University of Bradford eThesis

This thesis is hosted in Bradford Scholars – The University of Bradford Open Access
repository. Visit the repository for full metadata or to contact the repository team

© University of Bradford. This work is licenced for reuse under a Creative Commons
Licence.
MOBILE MULTIMEDIA SERVICE PROVISIONING

WITH COLLECTIVE TERMINALS

IN BROADBAND SATELLITE NETWORKS

M. HOLZBOCK

PhD

2011
MOBILE MULTIMEDIA SERVICE PROVISIONING

WITH COLLECTIVE TERMINALS

IN BROADBAND SATELLITE NETWORKS

An approach for systematic satellite communication system design for service

provisioning to collective mobile terminals including: mobile satellite channel

modelling, antenna pointing, hierarchical multi-service dimensioning and
aeronautical system dimensioning

Matthias HOLZBOCK

submitted for the degree

Doctor of Philosophy

School of Engineering, Design and Technology

University of Bradford

2011
Name:

Matthias Holzbock

Title:

Mobile Multimedia Service Provisioning with Collective Terminals in Broadband Satel-

lite Network

Keywords:

mobile, satellite, service, channel, collective, group, dimensioning, antenna, pointing

Abstract:

This work deals with provisioning of communication services via satellites for collec-
tively mobile user groups in a heterogeneous network with several radio access tech-
nologies. The extended use of personalised user equipment beyond the coverage of one
single terrestrial network by means of a satellite transport link, represents an increas-

ingly important trend in mobile satellite communication. This trend is confirmed by

the commercial introduction of broadband satellite communication to mobile terminals
mounted on vehicles, trains, ships or aircraft.

This work provides a consequent and structured approach for provisioning of services

to broadband satellite terminals for mobile user groups and addresses:

• a systematic satellite communication system design process for collective mobile

terminals;

• mobile satellite modelling at a wide range of frequencies, including current and

potential frequencies;

• an optimised Pointing Acquisition and Tracking (PAT) system design including

characterisation of moments for vehicle types of all mobile scenarios;

• a general hierarchical multi-service dimensioning methodology for collectively

mobile user groups, including voice, data, and multimedia services;

• an aeronautical system dimensioning scheme with (capacity and handover) re-

quirements analysis and evaluation of results for different satellite scenarios.

Contents

1 Introduction 1

1.1 Mobile Communications - a Satellite Opportunity . . . . . . . . . . . . 1

1.2 Mobile Satellite Terminals for Collective Use - Scope of the Thesis . . . 5

1.3 Objectives and Contributions of the Thesis . . . . . . . . . . . . . . . . 8

1.3.1 Mobile Satellite Channel . . . . . . . . . . . . . . . . . . . . . 9

1.3.2 Mobile Terminal Pointing, Acquisition and Tracking (PAT) . . . 10

1.3.3 System Dimensioning for Collective Mobile Terminals . . . . . 12

2 Satellite Communications Fundamentals for Mobile Services 14

2.1 Satellite - Terminal Geometry . . . . . . . . . . . . . . . . . . . . . . 15

2.1.1 Satellite Orbits . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2.1.2 Satellite to Mobile Basic Geometric Relations . . . . . . . . . . 17

2.1.3 Coverage Area and Constellation Parameters . . . . . . . . . . 21

2.1.4 Spot Beam Concept . . . . . . . . . . . . . . . . . . . . . . . 23

2.2 Satellite - Mobile Radio Link . . . . . . . . . . . . . . . . . . . . . . . 27

2.2.1 Antenna Radiation Pattern and Directivity . . . . . . . . . . . . 27

2.2.2 Boresight, Look Direction and Off Boresight Angle . . . . . . . 29

I
CONTENTS II

2.2.3 Gain and Efficiency . . . . . . . . . . . . . . . . . . . . . . . . 29

2.2.4 Beamwidth and Beam Contour . . . . . . . . . . . . . . . . . . 31

2.2.5 Radio Link System . . . . . . . . . . . . . . . . . . . . . . . . 32

2.2.6 Noise and Interference . . . . . . . . . . . . . . . . . . . . . . 33

2.3 Regulations for Mobile Earth Stations . . . . . . . . . . . . . . . . . . 35

2.3.1 Aircraft . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

2.3.2 Maritime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

2.3.3 Land mobile Train . . . . . . . . . . . . . . . . . . . . . . . . 39

2.4 Characteristics of the Mobile Satellite Channel . . . . . . . . . . . . . 41

2.4.1 Physical Phenomena of the Radio Channel . . . . . . . . . . . 41

2.4.2 Influences of Terminal Mobility on the Satellite Channel . . . . 42

2.4.3 Mobile Channel Characterisation . . . . . . . . . . . . . . . . 47

2.4.4 Lutz Channel Model . . . . . . . . . . . . . . . . . . . . . . . 50

2.5 Literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

2.5.1 Mobile Satellite Channel . . . . . . . . . . . . . . . . . . . . . 52

2.5.2 Satellite Terminal Antenna Pointing, Tracking and Acquisition . 54

2.5.3 System Dimensioning for Collective Mobile Satellite Terminals 55

3 Mobile Satellite Channel 57

3.1 S Band LMS Channel Characterisation by Measurement . . . . . . . . 59

3.1.1 S Band LMS Channel Measurement Setup . . . . . . . . . . . 59

3.1.2 S Band LMS Channel Measurement Results and Parameters . . 67

CONTENTS III

3.2 EHF Band LMS Channel Characterisation by Measurement . . . . . . . 69

3.2.1 EHF Band LMS Channel Measurement Setup . . . . . . . . . . 70

3.2.2 EHF Band LMS Channel Measurement Results and Parameters 72

3.3 Dominant Characteristics of LMS Channels at Higher Frequencies . . . 74

3.3.1 Outage Maps of Satellite Visibility . . . . . . . . . . . . . . . . 76

3.3.2 Results and Comparison with Statistically Derived Parameters . 80

3.3.3 Azimuth Diversity . . . . . . . . . . . . . . . . . . . . . . . . 82

3.4 Ka Band Aeronautical Satellite Channel Characterisation by Measure-

ments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

3.4.1 Ka Band Aero Satellite Channel Measurement Setup . . . . . . 83

3.4.2 Ka Band Aero Satellite Channel Measurement Results and Pa-

rameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

3.5 Dominant Characteristics of Aeronautical Satellite Channels at Higher

Frequencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

3.5.1 Aircraft Structure Shadowing . . . . . . . . . . . . . . . . . . 87

3.5.2 Cloud and Rain Attenuation . . . . . . . . . . . . . . . . . . . 90

3.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

4 Satellite Terminal Antenna Pointing, Tracking and Acquisition 97

4.1 PAT Basics and Definitions . . . . . . . . . . . . . . . . . . . . . . . . 98

4.1.1 PAT Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . 98

4.1.2 Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

4.1.3 PAT Design Methodology . . . . . . . . . . . . . . . . . . . . 100

CONTENTS IV

4.2 Beam Agility Requirements of the Terminal Antenna . . . . . . . . . . 102

4.2.1 Satellite Dynamics . . . . . . . . . . . . . . . . . . . . . . . . 102

4.2.2 Agility of Mobile Terminals . . . . . . . . . . . . . . . . . . . 103

4.3 Antenna Beam Steering . . . . . . . . . . . . . . . . . . . . . . . . . . 112

4.3.1 Mechanically Steered Antennas . . . . . . . . . . . . . . . . . 113

4.3.2 Electronically Steered Antennas . . . . . . . . . . . . . . . . . 115

4.3.3 Beam Steering Technology Tradeoff . . . . . . . . . . . . . . . 117

4.4 PAT Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120

4.4.1 Closed Loop PAT . . . . . . . . . . . . . . . . . . . . . . . . . 120

4.4.2 Open Loop PAT . . . . . . . . . . . . . . . . . . . . . . . . . . 124

4.4.3 PAT Algorithms Tradeoff . . . . . . . . . . . . . . . . . . . . . 125

4.5 PAT Errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126

4.5.1 Azimuth and Elevation Errors . . . . . . . . . . . . . . . . . . 126

4.5.2 Polarisation Error . . . . . . . . . . . . . . . . . . . . . . . . . 132

4.5.3 Latency Error . . . . . . . . . . . . . . . . . . . . . . . . . . . 136

4.6 PAT Solutions for Different Environments . . . . . . . . . . . . . . . . 137

4.6.1 Assessment of PAT Solutions for Different Environments . . . . 137

4.6.2 Open Loop Vehicular PAT Implementation . . . . . . . . . . . 139

4.6.3 Aeronautical INS Based Open Loop vs. Conical Scan Closed

Loop PAT Implementation . . . . . . . . . . . . . . . . . . . . 141

4.6.4 Vehicular Low Cost PAT Implementation . . . . . . . . . . . . 144

4.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152

CONTENTS V

5 System Dimensioning for Collective Mobile Satellite Terminals 155

5.1 Methodology of Hierarchical System Dimensioning . . . . . . . . . . . 156

5.1.1 Methodology - Service Mix, Models and Usage . . . . . . . . . 159

5.1.2 Methodology - User Group Size and Mobile Terminal Traffic . . 161

5.1.3 Methodology - Satellite Scenario and User Group Distribution . 162

5.1.4 Methodology - Capacity and Handover Analysis . . . . . . . . 164

5.2 Numerical Example - Aeronautical System Dimensioning . . . . . . . . 167

5.2.1 Aeronautical Example - Service Mix, Models and Usage . . . . 167

5.2.2 Aeronautical Example - User Group Size and Mobile Terminal

Traffic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172

5.2.3 Satellite Scenario and User Group Distribution . . . . . . . . . 174

5.2.4 Aeronautical Example - Data Traffic Analysis . . . . . . . . . . 179

5.2.5 Aeronautical Example - Handover Analysis . . . . . . . . . . . 182

5.3 Assessment of Satellite System Concepts for Aeronautical Communi-

cation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188

5.3.1 Coverage Optimisation . . . . . . . . . . . . . . . . . . . . . . 188

5.3.2 Handover vs. Capacity Optimisation . . . . . . . . . . . . . . . 190

5.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194

6 Summary 197

A Acronyms 206

B Notation and Symbols 214

CONTENTS VI

C Transformations 218

C.1 Earth - Mobile - IMU - Antenna Coordinate Transformation . . . . . . 218

C.1.1 Earth Fixed Coordinates E . . . . . . . . . . . . . . . . . . . . 219

C.1.2 Local Coordinates L . . . . . . . . . . . . . . . . . . . . . . . 219

C.1.3 Inertial Measurement Unit Coordinates I . . . . . . . . . . . . 220

C.1.4 Antenna Coordinates A . . . . . . . . . . . . . . . . . . . . . . 221

C.2 Polarisation Transformation . . . . . . . . . . . . . . . . . . . . . . . 222

D Traffic Models 225

D.1 Web Browsing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225

D.1.1 Packet Level Model . . . . . . . . . . . . . . . . . . . . . . . . 225

D.1.2 Page Level Model . . . . . . . . . . . . . . . . . . . . . . . . 227

D.2 Email . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229

D.2.1 Packet Level Model . . . . . . . . . . . . . . . . . . . . . . . . 229

D.2.2 Email Download/Upload Level Model . . . . . . . . . . . . . . 229

D.3 Voice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231

D.3.1 Packet Level Model . . . . . . . . . . . . . . . . . . . . . . . . 231

D.3.2 Call Level Model . . . . . . . . . . . . . . . . . . . . . . . . . 231

Bibliography 233
Chapter 1

Introduction

1.1 Mobile Communications - a Satellite Opportunity

Satellite Communications Satellites offer mobile services to users since the late
1970s. Starting from Inmarsat’s first maritime applications via geostationary earth
orbit (GEO) satellites, today the systems for mobile communication range from single
GEO satellites to sophisticated constellation networks with interconnected, regenerative

satellites (e.g., Iridium [Vat91], [Bru96]). While in the pioneering decades the commu-

nication companies were mostly governmentally held, nowadays telecom and satellite

companies are privatised providers, which have to operate in a highly competitive mar-

ket. In this competition, unfortunately only a few mobile satellite service (MSS) systems
turned out to be successful from a market perspective. The discussions about share of

resources, namely frequency spectrum allocation, have witnessed a shift towards the

strong arguments of the commercially more successful terrestrial mobile market.

Nevertheless, satellite communication features the inherent strength of wide area cov-

erage and as a consequence they are excellently suited or even indispensable for the

provision of service to globally scattered users. Bearing these strengths in mind, an

advantageous integration with terrestrial systems is the ultimate goal.

1
CHAPTER 1. Introduction 2

Terrestrial mobile communication systems of the 2nd generation were initially deployed

regionally with incompatible standards being concentrated on areas of high population

and service penetration. With this background, first attempts for worldwide personal

communication by satellites like Iridium or Globalstar [WV93], [Sch95] were made.

Using low earth orbit (LEO) satellite constellations, these systems (assembled some-
times by the naming “big LEOs”) have been proposed in the beginning of the 1990s and

became operational in early 2000. But although they were technologically convincing,
the economical breakthrough was not achieved and market expectations have not been

met, possibly prevented by the fast roll out and availability of the 2nd generation terres-

trial mobile communication systems. Nevertheless, after financial restructuring, Iridium

and Globalstar are still operational today with respectable subscriber numbers ( 400.000

subscribers Globalstar [Glo10]; 359.000 subscribers Iridium [Iri10]) and are launching
[Glo10] or planning [Iri10] their second generation satellites.

Initially targeting on the provision of a worldwide communication media for business

travellers, attempts introducing them for ’classical satellite niche markets’ like aeronau-
tical and maritime communications have been made in the last years.

Satellite systems like ASIA Cellular Satellite (ACeS) [TA96], [ACe11] or Thuraya

[Thu11] – both based on less expensive GEO satellites, without the need for a con-

stellation and intersatellite links (ISL) network and closely designed in compliance with
the global system for mobile communications (GSM) standard – were quite successful

from their very first commercial operation. ACeS has become part of Inmarsat in 2007

and Inmarsat extended its service portfolio with a handheld terminal named IsatPhone.

These systems are summarised with the term satellite personal communication network
(S-PCN) and were deployed for narrowband communication with restricted data capa-

bilities. They operate in the classical MSS frequency bands, namely L or S band and

are for the use with handheld terminals.

On the other hand, satellite services with portable or mobile platform mounted terminals

are evolving steadily towards higher data rates and also systems initially developed for
CHAPTER 1. Introduction 3

fixed service are considered for mobile usage.

Classical Inmarsat services came from a few kilobit per second (kbps) and support
nowadays a few hundred kbps with a sophisticated spacecraft incorporating narrow

spot-beams with frequency reuse and increased spectral efficiency by higher order mod-

ulation.

With the argument of spectrum limitations in the S- and L- bands, Ku/Ka band broad-
band systems are developed for mobile platform mounted use. This was maybe encour-

aged by successful prototype demonstrations of broadband satellite communications

within the frame of this thesis, to aircraft, cars, trains, and ships.

The research efforts into developing a broadband aeronautical communication

(AirCom) system over the last decade have now received a positive feedback from tech-
nology developers and airlines worldwide. This is evident from the introduction of ’In-
ternet above the clouds’ and the roll-out of ’GSM on board aircraft’. Inspired by enthusi-
astic estimations envisaging a multibillion US$ market, several significant steps towards

broadband aeronautical communications have been made, specifically the introduction

of Inmarsat’s Broadband Global Area Network (BGAN) and Swift Broadband (SBB)
services. With the introduction of Lufthansa’s FlyNet in 2004 using Connexion by
Boeing (CBB) [LR03] and its recent re-launch in 2010 [Luf10] using Panasonic’ eX-

connect as well as other networks e.g. ViaSat Yonder [Via11], LiveTV [Liv11], and

Row44 [Row11] which are all operating at Ku band.

With Inmarsat’s announcement of introducing GlobalXpress [Inm10] in 2014 offering

download rates up to 50 Mbps and upload speeds of 5Mbps by utilising Ka band this

frequency band is new to mobile satellite communication use.

A direct Air to Ground (AtG) system has been introduced by AirCell for the conti-

nental or contiguous US (CONUS) which is extended by an Inmarsat SBB link when

leaving the continental coverage. An AtG approach is taken for Europe and currently
under discussion by Conférence Européenne des Administrations des Postes et des

Télécommunications (CEPT) working group FM48 [CEP10].

CHAPTER 1. Introduction 4

With the launch of LightSquared’s SkyTerra 1 satellite in November 2010 [Lig11] the

first network has been introduced which is combing a terrestrial “long term evolution”
(LTE)and satellite infrastructure.

Terrestrial Communications GSM [Cox89], [MP92] – the most important represen-

tative of the 2nd generation (2G) mobile communication systems – was established in

1982 as an European standard for digital wireless communications, and when the first

system commenced commercial operation in 1991, nobody was able to anticipate its
overwhelming success. Advances in digital technology and integrated circuits allowed
miniaturisation of terminals to a comfortable size, thus mobile phones fit now into ev-

eryday’s life.

Nowadays the so called 3rd generation (3G) mobile communications systems became
omnipresent and 4rd generation (4G) is at the verge of being introduced. Key fea-

tures of these new systems are not only higher data rates, but also the establishment of
a worldwide mobile communication standard. Most 3G mobile communications sys-
tems are based on code division multiple access (CDMA) technology and the Universal

Mobile Telecommunications System (UMTS) standard as one of 3G mobile systems

representatives is operational in wide areas. Service operators are rolling out LTE a 3rd
Generation Partnership Project (3GPP) 3.9 generation standard in Europe [Vod10] and

USA [Ver10] for commercial use since end of 2010 enhancing the peak data rates to
300 Mbit/s downlink and 75 Mbit/s uplink [Nak09] by the use of orthogonal frequency

division multiplexing (OFDM). LTE-Advanced which is expected by the end of 2011

will comply with with 4G International Telecommunication Union - Telecommunica-

tion Standardisation Sector (ITU) 3GPP standard and will further enhance data rates to
mobile users up to download peak data rates of 1 Gbps.

Wireless local area network (W-LAN) hotspots are operating in a licence-free frequency
band and are based on the set of Institute of Electrical and Electronics Engineers (IEEE)

802.11 standards. This technology has set a widely used and accepted internet protocol
CHAPTER 1. Introduction 5

(IP) connectivity measure for people while being on travel and stopping for some time,

in hotels, lounges, stations, airports.

Mobile phones firstly introduced with pure voice and messaging capability, evolved

towards multi-application devices, supporting multi-standard wireless connectivity for

everyday communication, organisation, entertaining and much more.

1.2 Mobile Satellite Terminals for Collective Use - Scope

of the Thesis

Availability of terrestrial mobile communication services with personal devices and the
increased data rates of broadband mobile satellite system services called for usage of
the satellite terminals by multiple users which are collectively on the move.

Nowadays, the collective use of mobile broadband satellite terminals is widely recog-
nised to pose the most promising market potential in near future for mobile satellite

industry [NSR10].

The projected economic success of mobile satellite terminals which are collectively

used is strongly correlated with the requirement, that a multiple of services with hetero-
geneous access standard can be integrated for access by the end-users. On the other hand

provision of IP based transport bearers is the standard today adopted also by broadband

satellite terminals and thus baseline for this thesis.

This work is concerned with the collective use of broadband satellite terminals and the

issues imposed by the mobility of the user group. A baseline for a systematic system

design process will be given, including the mobile channel, antenna terminal pointing,
and finally the system dimensioning.

Figure 1.1 depicts an overview of a typical scenario with a collectively mobile user

group. This group (here passengers of an aircraft) is simultaneously using various ser-
CHAPTER 1. Introduction 6

Figure 1.1: Typical network scenario for service to collectively mobile users.

vices (voice, short messaging service (SMS), email, web, ...) with personalised de-
vices (cellular mobile phones, personal digital assistants, laptops, ...) of heterogeneous
communication standards (GSM, UMTS, W-LAN, ...) which have to be connected via

different wireless transport means (here two different satellite systems and one direct

terrestrial link) to the service provisioning networks on ground.

Throughout this work the wireless and wireline networks within the mobile user group

will be denoted as access network and the wireless network for connecting the collective
terminal will be entitled transport segment or backhaul. The term collective mobile

satellite terminal refers to the satellite (transport segment) terminal collectively shared

by a mobile user group.

The research work of this thesis was performed within the years 1996 to 2006. While
satellite industry learned hard lessons in the late 90s from the financial difficulties of

the "big LEOs", the work within the frame of this thesis concentrated from the very first
beginning on the collective use of broadband satellite terminals.

Besides the fact, that for this special group of users a global (or at least wide) coverage

is typically required, collectively mobile user groups enable the efficient use and sharing
CHAPTER 1. Introduction 7

of the broadband capabilities of satellite terminals.

Table 1.1 provides the timeline of the projects to which this PhD work contributed and
a reference to the associated chapter of the thesis.

Table 1.1: Project timeline and author’s contribution

project date author’s contribution w.r.t. the thesis chapter
SECOMS a 1995 - execution of EHF band LMS channel measure- 3.2
1996 ment campaign and analysis (narrowband)
- multi service modelling of aeronautical service 5.1
b
ABATE 1996 - execution of Ka band aeronautical channel mea- 3.4,
1998 surement campaign and analysis 3.5
- aeronautical Ka band terminal and PAT definition 4
- aeronautical Ka band terminal and PAT design 4.6
and development
- aeronautical attitude analysis 4.2
ICO c 1998 - execution of S band LMS channel measurement 3.1
1999 campaign and analysis
SUITEDd 2000 - vehicular Ka band terminal definition and PAT 4
2002 - vehicular Ka band PAT design and development 4.4
- vehicular attitude analysis 4.2
BayCES e 2000 - satellite visibility study 3.3
2001 - hierarchical multi service modelling 5.1
MOBILTY f 2001 - maritime terminal definition and PAT 4
2002 - maritime Ku band PAT design and development 4.4
- maritime attitude analysis 4.2
FIFTH g 2002 - train terminal and PAT definition 4
2004 - train terminal PAT design and development 4.4
- train attitude analysis 4.2
Wireless 2002 - aeronautical multi service modelling 5
Cabin h 2005
a
European Commission Framework Programme (EC FP) project Satellite EHF Communications for
Mobile Multi Media Services
b
EC FP project “ACTS Broadband Aeronautical Terminal Experiment” (ABATE) [HLCL99]
c
Industry contract "Supply of Propagation Data by Filed Campaign" [JHE+ 99]
d
EC FP project “Multi-Segment System for Broadband Ubiquitous Access to Internet Services and
Demonstrator” (SUITED) [HJG+ 00]
e
Bavarian Center of Excellence for Satellite Constellation System project 2220
f
EC FP project “Mobile Real Time TV via Satellite Systems” (MOBILITY) [LPG+ 03]
g
EC FP project “Fast Internet for Fast Train Hosts” (FIFTH) [CMSS03]
h
EC FP project Development and Demonstrator of Wireless Access for Multimedia Services in Air-
craft Cabins
CHAPTER 1. Introduction 8

Ku-Mobile a 2004 - vehicular Ku band receive only terminal and PAT 4

2007 definition
- vehicular Ku band receive only PAT design and 4.6
development
- vehicular attitude analysis 4.2
MOWGLY b 2005 - hierarchical multi service modelling of train, 5.1
2007 maritime, and aeronautical service
- vehicular, train, maritime, and aeronautical ter- 4
minal requirements definition and PAT
a
European Space Agency (ESA) project 17769 "Mobile Ku Band Receive-only Terminal
b
EC FP project MObile Wideband Global Link sYstem

1.3 Objectives and Contributions of the Thesis

The goal of the thesis is is to investigate the impacts of mobility onto broadband satellite

communications through a systematic and structured approach. Figure 1.2 depicts the
system design process for collectively mobile satellite terminals which is detailed within

the chapters of this thesis.

Spectrum and Terminal Wireless Channel

Satellite Constellation Satellite Radio Link
Antenna Regulations Characteristics

Chapter 2

Mobile Satellite Channel

Chapter 3

Mobile Platform
Mobile Terminal Design
Characteristics

Chapter 4

Mobile Group Capacity and Handover

Mobile Services
Characteristics Analysis

Chapter 5

Figure 1.2: System design process as followed in this work.

The original contributions of the research work are summarised in the subsequent sec-
tions.
CHAPTER 1. Introduction 9

1.3.1 Mobile Satellite Channel

Mobility of a satellite terminal affects the satellite to mobile propagation path. The
mobile satellite channel is suffering from time variant fading due to multipath effects

and shadowing resp. blockage. The latter effects become especially predominant for
higher frequencies and narrow beamwidth antennas, typically used for broadband satel-

lite communication. Thus, satellite visibility for any terminal – but mobile terminals
in particular – influences fundamentally the performance of a satellite communication.

Appropriate countermeasures can only be introduced, if the characterisation of the par-

ticular satellite channel is known. Several channel models and characterising parame-

ters have been presented for the land mobile satellite (LMS) channel (e.g. [LCD+ 91],
[GV95], [Jah99]). All models are based on statistical analysis of channel measurement
campaigns at the classical MSS frequencies in L and S band.

The thesis enhances the available mobile satellite models by LMS and aeronautical
satellite channel characteristics at higher frequencies.

Even if exhaustive channel measurement campaigns have been performed within the

thesis, it has been found that they can only provide limited data sets for few environ-
ments and satellite constellations. The resulting channel data is restricted to certain
environments, operational scenarios, frequencies, antenna radiation patterns and satel-

lite system dynamic parameters. The upcoming availability of three dimensional (3D)

data of cities within the last decade allows an accurate modelling of satellite visibility

maps and calculation of blockage probabilities at a wider area and with far less effort

compared to channel measurement campaigns.

Contribution to Mobile Satellite Channel

• The thesis introduces a LMS channel measurement campaign at 2 GHz (S band)

which has been performed for the satellite system ICO. Channel data have been

collected in down- and uplink of two diversity paths and are presented for selected
CHAPTER 1. Introduction 10

user environments and operational scenarios. These measurements have found a

recent interest revival by the emerging AtG plans discussed within CEPT working
group FM48.

• Within the thesis a LMS channel measurement campaign at 40 GHz (EHF band)
was performed and statistical analysis of narrowband channel parameters have

been evaluated and are presented.

• The thesis details the setup and results of an aeronautical channel measurement

campaign at 20 GHz (K band). This data is being of major interest for all broad-

band aeronautical MSS systems specifically Inmarsat GlobalXpress.

• The thesis presents a satellite visibility analysis by ray tracing of urban 3D mod-
els. This allows investigation of larger areas, with a variety of analysis and sim-

ulation possibilities. The analysis includes 75 km of streets for three different

satellite constellations, including potential constellations for a Sirius versus XM
Radio service expanded to Europe. The visibility results are also compared to the
blockage factor of the LMS channel measurements.

1.3.2 Mobile Terminal Pointing, Acquisition and Tracking (PAT)

Broadband satellite services need to employ antennas with a certain degree of directiv-

ity. These antennas have to be continuously pointed towards the satellite and have to
countermeasure variations in the satellite to mobile geometry.

Mobile satellite terminals at higher frequencies have been used for channel measure-

ments and demonstrations (e.g. [DJWW93], [HJ98b]) or within military implementa-

tions. These terminals have been designed without particular focus on optimised an-

tenna PAT algorithms and economic aspects of the terminal costs. Moreover, some an-

tennas meant for commercial use, neglected important technical requirements and eco-

nomic terminal and pointing, acquisition, and tracking (PAT) aspects; e.g. the phased
CHAPTER 1. Introduction 11

array antenna which was initially planned for use by the Connexion by Boeing system

was not able to provide sufficient angular beam agility range for intercontinental flight
routes.

System design and regulatory requirements call for certain antenna gain requirements
and define a maximum interference, but do not provide a structured approach to terminal

antenna and PAT design.

Indispensable input for a cost optimised terminal antenna and PAT system is the knowl-

edge about the mobile’s attitude movement. This data is mostly confidential data e.g.
by car manufactures because it is related to suspension comfort and chassis technology.

Contribution to Mobile Terminal PAT

• The thesis introduces a design methodology for cost optimised terminal antennas

and PAT systems utilised in different mobile environments.

• A unified description of azimuth, elevation and polarisation PAT errors is given

and resulting in a general approach to optimisation of pointing error vs. antenna

gain.

• The thesis provides characterisation of three axis angular movements for vehi-
cle types in range, rate, and acceleration for all mobile scenarios i.e. car, train,

ship and aircraft. The data has been collected from reference inertial measure-

ment systems during satellite service demonstrations and channel measurement

campaigns.

• The thesis presents a tradeoff analysis between mechanically and electronically

steered beams and closed and open loop PAT algorithms for mobile satellite ter-

minal antennas.

• Example implementations of optimised terminal antenna and antenna PAT sys-

tems are given for car, train, ship and aircraft.

CHAPTER 1. Introduction 12

1.3.3 System Dimensioning for Collective Mobile Terminals

Mobile terminals which are collectively used by a group of passengers and various end
user services pose a new traffic mix to satellite systems. Dimensioning of MSS was

traditionally based on single service traffic models with a geographical static user distri-
bution depending on market factors (e.g. gross domestic product, service acceptance).

Usage of multiple services within one satellite system was modelled in [WHS97].

This model was enhanced to cope with multi-service traffic for a mobile group, i.e.

traffic models and usage parameters for aeronautical passenger groups has been selected

and defined in order to provide dimensioning input for the groups terminals.

Models for group size and local and temporal distribution of the groups are required to
provide input for satellite system dimensioning or evaluate the performance of existing

satellite systems which need to cope with mobile broadband group terminals.

Today’s GEO systems used for mobile broadband satellite communication do not allow
seamless satellite handover, thus minimisation of handover occurrence is one of the
optimisation goals for service acceptance from user perspective.

Contribution to System Dimensioning for Collective Mobile Terminals

• In this work a hierarchical multi-service satellite communication system dimen-

sioning is detailed, first as methodology and

• second as the numerical analysis for the example of aeronautical communication,

reflecting i) the different traffic models, which are varying per service of the ser-

vice mix, ii) the different usage of the services, which is varying per group with
the travel characteristics, iii) the different group sizes, and iv) the local and tem-

poral distribution of groups within the service area.

• Results for capacity requirements on terminal level for typical aeronautical group

sizes are given.

CHAPTER 1. Introduction 13

• Aeronautical group and data traffic analysis on spot beam and satellite system

level is provided for a Inmarsat similar satellite system.

• Results for spotbeam and satellite handover occurrence of the aeronautical group
terminals is detailed, resulting is a handover signalling data overhead estimation.

• A analysis of satellite capacity requirement resulting from aeronautical passenger

communication versus satellite handover occurrence is given for different orbital

spacing of a three GEO satellite system.

Chapter 2

Satellite Communications
Fundamentals for Mobile Services

Spectrum and Terminal Wireless Channel

Satellite Constellation Satellite Radio Link
Antenna Regulations Characteristics

Chapter 2
Mobile Satellite
Channel
Chapter 3
Mobile Platform Mobile Terminal
Characteristics Design
Chapter 4
Mobile Group Capacity and
Mobile Services
Characteristics Handover Analysis
Chapter 5

Special geometrical conditions between the communication partners have to be consid-

ered in mobile satellite communication. This chapter lays a basis for a common under-

standing of mobility issues of satellite terminals and introduces the parameters used in
the subsequent chapters to enable a self contained document reading.

For a comprehensive introduction into satellite communication systems see [MB02],

[LWJ00], and [SH01].

14
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 15

2.1 Satellite - Terminal Geometry

2.1.1 Satellite Orbits

1 2
Satellite movement in relation to the earth is based on Kepler’s laws [Kep09],

[Kep19], which are generically applied for a two-body system. Satellite orbits were

introduced by Clarke 3 [Cla45], who constituted with his publication in 1945 the begin-

ning of satellite communication.

2.1.1.1 Orbit Types

Kepler’s first law on planetary motion adopted to satellite-earth relation generally de-

fines that satellites move around the earth in a plane with elliptical orbits and the earth in
the focal point. From elliptical orbits the special cases of circular and highly elliptical

orbit (HEO) became of importance for satellite communication systems. If the distance
between earth gravity centre and satellite is nearly constant the orbit is named a circular
orbit.

The angle under which the orbital plane intersects the equatorial plane of the earth is

called inclination. Orbits which are not coincidental with the equatorial plane conse-

quently are named inclined orbits, with the special case of polar orbits, if the inclination

is approximately 90 degree (cf. Figure 2.1).

2.1.1.2 Orbit Height and Orbit Period of Circular Orbits

The major characteristics of circular orbits are the orbit height hS and the orbit period

T . Assuming an ideal spherical shape for the earth, the satellite height over ground

is approximately constant and the relation between hS and T can be derived based on

Newton’s 4 law of gravity [New87] and Kepler’s third law by:

1
Latin: satelles ≡ bodyguard
2
Friedrich Johannes Kepler; * 27.12.1571 in Weil der Stadt; †15.11.1630 in Regensburg, Germany
3
Arthur Charles Clarke, * 16.12.1917 in Minehead, Somerset, UK
4
Sir Isaac Newton; * 04.01.1643 in Woolsthorpe-Colsterworth, Lincolnshire; †31.03.1727 in London
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 16

s
(hS + RE )3
T = 2π , (2.1)
γ0 ME

resp.

s µ ¶2
3 T
hS = γ0 ME − RE , (2.2)
2π

3
with the gravitational constant γ0 = 6, 672·10−11 kgms2 , earth mass ME = 5, 974·1024 kg,
and mean equatorial earth radius RE = 6378, 144 km.

LEO Orbits with heights of about 500 to 2.100 km above earth surface are named
LEO, with orbit periods of approximately 2 hours. This orbit height is between
the earth’s ionosphere and the inner Van Allen Belt 5 , a region of high ion con-
centration about one Earth radius above the Earths surface, which is reducing the

lifetime of satellites.

MEO A second region - the outer Van Allen belt - is reducing the useful orbit spacing
at regions 14.500 to 19.500 km above the Earths surface. Thus medium earth
orbit (MEO) (or synonymously: intermediate circular orbit (ICO)) satellites are

operating typically at approximately 10.000 km above Earth surface, with orbit

periods of 4 to 6 hours.

GSO An orbit height of 35.786 km is named geosynchronous earth orbit (GSO), with

a corresponding orbit period T equal to the duration of one Earth rotation TE . A

non-inclined GSO is called GEO. A GEO satellite remains fixed with respect to

Earth coordinates and is thus of particular importance for satellite communication

to fixed earth stations.

Figure 2.1 depicts these essential orbit types for satellite communication.
5
James Alfred van Allen; * 07.09.1914 in Mount Pleasant, Iowa; †09.08.2006 in Iowa City, US
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 17

polar LEO
GEO

inclined MEO

inclined LEO inclined HEO

Figure 2.1: Satellite orbit types.

2.1.2 Satellite to Mobile Basic Geometric Relations

2.1.2.1 Orbit Parameters

The orbit height is an important parameter for characterising satellite constellations.

Figure 2.2 depicts the geometric relation between satellite, terminal and geocenter and
allows the deduction of further geometrical parameters by the laws of sines and cosines
for triangles span by the geocenter, terminal and satellite, respectively satellite, terminal

and nadir [LWJ00].

The elevation angle ε denotes the angle at which a terminal sees the satellite above the

horizon. It can be derived by

à RE
! µ ¶
π cos ψ − hS +RE hS + RE
ε = − ϑ − ψ = arctan = arccos sin ϑ . (2.3)
2 sin ψ RE

The nadir angle ϑ denotes the angle at which a satellite sees the terminal with respect

to the sub-satellite point and can be derived by

µ ¶ µ ¶
RE RE sin ψ
ϑ = arcsin cos ε = arctan . (2.4)
RE + hS RE + hS − RE cos ψ
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 18

Satellite terminal d
h
g
hS

R RE Nadir, Sub-satellite point

Geocenter

Nadir, Sub-satellite point

RE
R hS
n
d
h

Satellite terminal

dmax
hmax
gmin
hS
RE

Rmax Nadir, Sub-satellite point

Coverage Area

Geocenter

Figure 2.2: Geometric satellite to terminal relations.

CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 19

The earth central angle ψ denotes the angle between satellite and terminal seen from

the geocenter and can be derived by

µ ¶ µ ¶
RE RE + hS
ψ = arccos cos ε − ε = arcsin sin ϑ − ϑ . (2.5)
RE + hS RE

If the satellite’s and the terminal’s position is given in earth spherical coordinates (ra-

dius, longitude, latitude (rS , λS , δS and rT , λT , δT )), ψ can be derived according to trans-
formations given in [MB02]

ψ = arccos(cos δS cos δT cos(λS − λT ) + sin δS sin δT ) . (2.6)

The elevation angle ε and the nadir angle ϑ are then derived in earth spherical coordi-
nates by

à !
rS (cos δS cos δT cos(λS − λT ) + sin δS sin δT ) − rT
ε = arcsin p ,
rS2 + rT2 − 2rS rT (cos δS cos δT cos(λS − λT ) + sin δS sin δT )
(2.7)

and

µ ¶
rt sin(arccos(sin δS sin δT + cos δS cos δT cos(λS − λT )))
ϑ = arctan . (2.8)
rS − rt (sin δS sin δT + cos δS cos δT cos(λS − λT ))

The relation between ψ, ϑ and ε is defined by

ψ + ϑ + ε = π/2 . (2.9)

The slant range d denotes the distance between satellite and terminal and can be derived

by
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 20

q
2
d= RE + (RE + hS )2 − 2RE (RE + hS ) cos ψ . (2.10)

or in earth spherical coordinates, d is calculated by

q
d= rS2 + rT2 − 2rS rT (cos δS cos δT cos(λS − λT ) + sin δS sin δT ) . (2.11)

The azimuth angle ϕ denotes the angle at which a terminal sees the satellite measured

– usually from North – around the horizon. Given the satellite’s and the terminal’s

position again in earth spherical coordinates, ϕ is calculated by

µ ¶
cos δS sin(λS − λT )
ϕ = arcsin , (2.12)
sin(arccos(sin δS sin δT + cos δS cos δT cos(λS − λT )))

for δS > δT and λS < λT .

More generally the following case differentiation provides the azimuth angle ϕ depend-
ing on the relation of the satellite’s and mobile’s longitude and latitude



 ϕ̂ if δS > δT and λS < λT





 · ¸
π + ϕ̂ if δS 5 δT and λS = λT cos δS sin|λS − λT |
ϕ= with ϕ̂ = arcsin ,

 sinψ

 2π − ϕ̂ if δS > δT and λS = λT



 π − ϕ̂ if δ 5 δ and λ < λ
S T S T
(2.13)

with 0 ≤ ϕ ≤ 2π and ϕ = π/2 corresponds to West.

2.1.2.2 Satellite Ground Tracks

As showed above, the GEO, is the only orbit type where the satellite position seen from

the earth is fixed. For all other orbit types the satellite moves with respect to the rotating
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 21

earth.

The time dependent satellite latitude δS and longitude λS of the satellite ground track

with respect to the earth can be calculated by

δS (t) = arcsin(sin i sin ω(t)) , (2.14)

cos ω(t) 2π
λS (t) = λ0 + arccos − (t − t0 ) . (2.15)
cos δS (t) TE

Hereby, λ0 denotes the latitude where the satellite passes the equatorial plane in north
direction (ascending node) resp. t0 denotes the time when the satellite passes the as-

cending node and ω(t) denotes the angle from the ascending node to the satellite (see
Figure 2.3).

Satellite

*S
i TS
8S
Geocenter 80
Reference meridian
Ascending node

Figure 2.3: Ground track of the satellite on the rotating earth.

2.1.3 Coverage Area and Constellation Parameters

2.1.3.1 Satellite Coverage Area

The orbital height determines also the maximal area visible from a satellite which forms

a spherical segment of the earth. The so called coverage area or footprint defines the
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 22

area of a satellite in which the service is available (see Figure 2.2). While the visible area

is purely related to the satellite earth geometry, the coverage area is including effects of
a possibly directed satellite antenna to focus the service area on a dedicated region.

2.1.3.2 Minimal Elevation Angle

Theoretically, with elevation angle ε = 0 the maximal coverage area is determined for a

given orbit height. Practically satellite systems limit the minimal elevation angle εmin to

a value greater zero, depending on the communication environment. Typically satellite

communication to fixed users is operational down to εmin = 5◦ . For mobile satellite

communication higher εmin values are desirable, not only because signal shadowing
due to obstacles is relaxed, but also steering range requirements for directed mobile

terminal antennas are less stringent.

Related to the minimal elevation angle is a maximum nadir angle ϑmax , a coverage angle

or maximum earth central angle ψmax and a maximum slant range dmax , according to

equations 2.5 to 2.10.

2.1.3.3 Angular Velocity of Satellites

The time-variant elevation angle ε(t) according to equation 2.7 resp. azimuth ϕ(t) ac-

cording to equation 2.13 under which a satellite is seen from a point on the earth can be

derived with time-variant satellite latitude δS (t) (c.f. equation 2.14) and longitude λS (t)
(c.f. equation 2.15).

Mobile satellite communication with directed antennas requires a steering of the an-

tenna towards the satellite, counteracting the mobile’s attitude and position changes.
The antenna movements are superimposed by the time-variant azimuth and elevation.

Figure 2.4 shows the elevation angle and the elevation angle speed for a MEO and LEO

overpass and an εmax = 70◦ MEO and εmax = 40◦ LEO passage.
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 23

80
MEO

elevation in deg
60
LEO
40

20

0
−60 −40 −20 0 20 40 60
time in min
elevation speed in deg/min

5 LEO

2.5 MEO
0
−2.5
−5
−60 −40 −20 0 20 40 60
time in min

Figure 2.4: Elevation angle (upper) and elevation angular speed (lower) over time for a LEO
(blue lines) and MEO (black lines) overpass (solid lines) and εmax = 70◦ MEO and εmax = 40◦
LEO passage (dashed lines).

2.1.4 Spot Beam Concept

Similar to terrestrial cellular networks, cellular patterns are used in satellite communica-
tion, too. Therefore the coverage area is divided into spot beams by means of multibeam
satellite antenna technology.

The lower bound of the spot beam size is determined by the antenna size of the satellite
and the operational frequency. Constrains are set by the possible launch size of antenna

dishes. The biggest antennas used currently for satellite communication reach a diame-

ter of ca. 12 metre in operation (e.g. ACeS Garuda Satellite [ACe11]). They are folded

during launch and unfold similar to an umbrella when set into operation. Next genera-

tion satellites will incorporate antennas of up to 22 metre [Mob06]. Figure 2.5 shows
the diameter of a central spot beam for different satellite antenna sizes and frequencies

for a GEO satellite according to equation 2.21.

The increased antenna complexity and mass for incorporation of spot beams pays off,

because it allows on the one hand the reuse of frequencies and thus increases system
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 24

spot beam diameter [km] 3

10

1.6 GHz
2 GHz
2
10

10 GHz
14 GHz
18 GHz
1
10
1 10 20 30
antenna diameter [m]

Figure 2.5: Diameter of central spot beam for different satellite antenna diameters and frequen-
cies for a GEO satellite.

capacity, and on the other hand the narrow antenna beam provides increased communi-
cation power due to its higher directivity.

R
2R 1 3
hmax 3R
Reuse Cluster K=3
h 1 3 2 1

3 2 1 3 2
Reuse Distance D= 3K R

1 3 2 1
D

gmin
2 3
1

Central Spotbeam

Coverage Area
Spotbeam Area

Figure 2.6: Spot beam pattern.

Figure 2.6 shows a spot beam pattern within a satellite coverage area defined by the

minimum elevation angle εmin and corresponding ϑmax . The angular spot beam width
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 25

ϑ is smaller than ϑmax and the spot beam pattern has to be designed in a way that the

coverage area is covered without gaps.

Typically regular hexagons are used for modelling purpose, which are in reality overlap-

ping circles of the spot beam contour as shown in Figure 2.6 on the left hand side. Very
commonly the spot beam centers are located on a regular triangular grid and aligned on

rings around the central spot beam.

The spatial distance allows the discrimination of co-channel terminals by the directed

satellite antenna and thus a reuse of the same frequency. Important parameters are the
reuse distance D describing the distance between the centre of two spot beams of same
frequency and the cluster size K denoting the number of neighbouring cells in that

different frequency bands are used. Typical cluster sizes are 1, 3, 4, 7, 9, 12, ... because
regular spot beam patterns are achieved [LWJ00].

Channel separation is typically achieved by frequency division multiple access (FDMA)

and time division multiple access (TDMA) systems. In these system the overall avail-
able bandwidth B can be reused in each cluster. The total bit rate capacity of a satellite

can be calculated by

nsb
Csatellite = ηmodcod B (2.16)
K

with B being the available frequency bandwidth, nsb the number of spot beams and

ηmodcod the modulation and coding (or spectrum) efficiency. The total bit rate capacity
depends on the applied modulation and coding technique. Other effects like the effi-

ciency of higher layer protocols and overhead due to signalling may also be applied.

For CDMA systems typically enough orthogonal codes are available for separation of
the terminals. But nevertheless, also in CDMA systems the influence of the narrow

beam on the link budget increases the maximum achievable bit rate for the terminal

link.
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 26

Very narrow spot beams, with spot beam diameter on earth surface less than the spatial

distance between single terminals, would allow to assign one single spot beam per ter-
minal and separation of co-channel terminals by their spatial distance. Combined with

the capability of actively steering these spot beams towards the terminal, one may speak

of a truly space division multiple access (SDMA) system. In this case a continuous
coverage of the spot beam within the coverage area of the satellite is not necessary any

more, but the antenna has to support same number of spot beams as active terminals.
Realisation of antennas allowing to form such beams is not possible in near future, how-

ever in chapter 4.3 a proposal for an optical terminal is given, which is approximating

such a system.

Figure 2.7 depicts such a regular pattern of nsb = 233 beams, similar to the one used
by Inmarsat’s fourth generation satellites. This pattern is generated by using a constant

spot beam beamwidth ϑ and circular arrangement of the cells, some of the outer cells

are omitted. The projection of this pattern on the spherical earth surface is shown. By
this projection the regular pattern is distorted due to the curvature of the earth and the
spot beam areas increase with increasing distance from the central spot beam until they
spill over the horizon. Equation 2.9 defines the relation of the earth centric angle ψ and
the angle of the spot beam centre with respect to the nadir.
Spotbeam Projection

50
Latitude in deg

−50

−150 −100 −50 0 50 100 150

Longitude in deg

Figure 2.7: Regular spot beam pattern of a GEO satellite with constant spot beam width pro-
jected onto the earth.
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 27

Spot beams with constant area size can be achieved by using constant earth centric

angles for the single spot beams. However then the spot beam size of the pattern is
defined by the outermost spot beam. The satellite antenna aperture size allows to achieve

higher antenna gains in the inner spot beam areas.

The reflector diameter of an Inmarsat 4 satellite is 9 m and the operational frequency is

approximately 1.6 GHz. With equation 2.21 a two sided 3 dB antenna spot beam width

of approximately 1.45◦ and a diameter of 750 km for the central beam follows.

In order to add some degree of flexibility a sophisticated satellite system can dynami-
cally assign resources to those spot beams where time and location dependent capacity

is required.

Besides the drawback of an increased antenna complexity and mass, the introduction

of spot beams leads for mobile users or for non-GEO satellite systems with non earth
fixed spot beam patterns to an increased spot beam handover (HO) rate. It is generally
desirable that HOs are performed without loss of an active communication link, thus a
HO is related with additional signalling traffic.

The formation of an earth-fixed spot beam pattern for non-GEO satellites as addressed in

[RM96] would require an active steering of the satellite antenna towards fixed spot beam

centers on ground and an active adaptation of the spot beam contour as the projection of

the spot beam pattern on the spherical earth surface is changing. Up to now, only non-

GEO satellite constellations with satellite-fixed spot beam patterns have been realised.

2.2 Satellite - Mobile Radio Link

2.2.1 Antenna Radiation Pattern and Directivity

An omnidirectional antenna is defined to radiate equally into all directions of space. It

is the idealised theoretical model for antenna characterisation. The power flux density
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 28

of such an isotropic 6 radiator at a distance d with a given radiated power PT is defined

by:

PT PT
Si (d) = = . (2.17)
spheric surface at distance d 4πd2

As the isotropic radiator is only a theoretical model, typical antennas emit their power

flux density not uniformly into all directions in space. Typically the radiated power of
the antenna is distributed in a solid angle less than the whole sphere and thus a power

flux density greater than Si is generated for this direction.

This antenna characteristics is called radiation pattern and is commonly normalised

to its maximum power flux density. Figure 2.8 shows different antenna radiation pat-
terns. The directivity of an antenna in a given direction is defined as ’the ratio of the

intensity of radiation, in that direction, to the radiation intensity averaged over all direc-
tions’ according to [ITU95a]. If an antenna has a distinct maximum directivity the term
directivity also is used for referring to this maximum antenna directivity.

Figure 2.8: Antenna radiation patterns of: a) an isotropic radiator (dashed red) b) dipole antenna
(dotted green) c) directed antenna (chain dotted blue).

The area around the local maxima of antenna directivity is called beam or lobe.

Directivity of an antenna can be achieved by focusing the radiation for instance with

parabolic reflectors, lenses, or an array of radiation elements.

6
Greek: isos ≡ equal; Greek: tropos ≡ direction
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 29

2.2.2 Boresight, Look Direction and Off Boresight Angle

The direction of a distinct maximum directivity is called antenna boresight. The direc-

tion perpendicular to the antenna aperture is called look direction. The antenna boresight

and look direction are identical, unless antennas are using a reflector with offset feed or

phased array antennas with shifted beamforming (cf. section 4.3.2).

Figure 2.9: Definition of the off boresight angle ϑ.

The off boresight angle ϑ defines the angle between boresight and a given signal prop-
agation vector (cf. Figure 2.9). For satellite antennas this angle is similar to the nadir
angle defined in equation 2.4. Later, it will be of importance for describing , e.g., inter-

ference or the mispointing of a mobile antenna.

2.2.3 Gain and Efficiency

The antenna gain G is defined by the ratio of the power flux density which the antenna
under consideration introduces at a defined point in space, with respect to the power

flux density which a reference radiator would introduce at the same point. The typical

reference antenna is the aforementioned isotropic radiator.

S
G= . (2.18)
Si

With above equations 2.17 and 2.18 the antenna gain is defined as a measure of power

emitted into a certain solid angle. If an antenna has one dominating radiation direction
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 30

the antenna is often refereed to as a directive or high gain antenna. The primary direc-

tion of radiation is called main lobe in contrast to the side lobes which describe other
local maxima. Radiation in other directions than the main lobe is unwanted, because of

interference to other systems. Side lobe suppression with respect to the main lobe is of

importance.

2h3dB side lobe

Gmax - 3dB h3dB

main lobe
Gmax

boresight

gmin

coverage area

3dB Contour

Figure 2.10: Directive Satellite Antenna.

According to [LWJ00] the maximum antenna gain Gmax is proportional to the effective

size of the aperture area Ae and reciprocally proportional to the wavelength λ, with

4π 4πf 2
Gmax = η Ae = η A e . (2.19)
λ2 c20

The antenna efficiency η is affected by loss, e.g. feed mismatch and thus 0 ≤ η ≤ 1. The

efficiency of aperture antennas is additionally influenced by, e.g. the illumination type,
spill-over, and surface finish of the antenna aperture. Typical values for the antenna

efficiency are between 0.55 and 0.65 [LWJ00]. The speed of light in vacuum c0 is

approximately 3 · 108 m/s and the frequency f = c0 /λ.

It has to be noted that the antenna gain is increasing with higher frequencies as well as

with increase of the aperture size.

CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 31

2.2.4 Beamwidth and Beam Contour

With increased focus of the radiated power in a solid angle the antenna gain increases
and obviously the antenna beamwidth decreases. The gain loss of an antenna with

circular-Gaussian shaped main lobe with respect to the boresight gain is given by
[LWJ00]

µ ¶2
ϑ
Gϑ = Gmax − 3 . (2.20)
ϑ3dB

The antenna one-sided half power (or 3 dB) beamwidth ϑ3dB defines the angle, relative
to the main beam axis, between the two directions at which the measured gain is 3 dB

below the maximum gain [ETS06b] (cf. Figure 2.10). Accordingly the two-sided half
power beamwidth 2 · ϑ3dB defines the angular span in which the antenna gain is grater

than half the maximum gain. Besides the 3 dB beamwidth also the 4.3 dB beamwidth
is often applied as system design parameter.

According to [LWJ00] ϑ3dB of an antenna with tapered power flux density aperture illu-
mination (e.g. typically for antennas with parabolic reflector of circular shaped aperture
area) can be approximated by

2ϑ3dB ≈ 70◦ · λ/D , (2.21)

with D being the diameter of the reflector.

Consequently the x dB beamwidth is a x dB contour on the earth surface. Within the

x dB contour area the power flux density emitted by the directed satellite antenna re-

mains greater than maximum power flux density -xdB. This contour is also called edge

of coverage (EOC).
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 32

2.2.5 Radio Link System

For antenna far fields the law of reciprocity applies and therefore above mentioned char-

acteristics apply to transmit as well as to receive antennas.

The power transmitted of a radio link system consisting of a directed transmit and a high

gain receive antenna is derived in the following.

The power flux density of a directed antenna at distance d with total transmit power PT

and a gain of GT in direction of the receiver is defined by as

PT GT
ST = . (2.22)
4πd2

Let the receive distance be d and the receiver has an antenna with an efficient antenna
aperture Ae (cf. equation 2.19). It can pick up the power

µ ¶2
PT GT GR λ2 PT GT λ
PR = Ae ST = Ae = = GR PT GT . (2.23)
4πd2 4π 4πd2 4πd

The last term of equation 2.23 represents the definition of the free space loss which
describes the loss due to the spatial distance d of receiver and transmitter

µ ¶2
4πd
L0 = . (2.24)
λ

The total loss due to propagation L = L0 + ∆L consists of the free space loss L0 and an

additional attenuation caused by gaseous, vaporous, and ionospheric effects ∆L, thus

the power transmit formula results as

PT GT GR
PR = . (2.25)
L

The product of transmitter power PT and maximum transmit antenna gain GT is called

effective isotropically radiated power (EIRP) and defines one of the most important
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 33

GR
PT GT PR

Receiver
Transmitter

Figure 2.11: Radio link budget.

characterisation values of a wireless transmitter. The EIRP is equivalent to the trans-

mit power of an isotropic radiator causing the same power flux density as the directed
antenna.

EIRP = PT · GT . (2.26)

For the more generic case that the satellite terminal is not located directly in the satellite
antenna’s boresight, the EIRP and GR are functions of terminal to satellite geometry

respectively the off boresight angle and the distance.

EIRP (ϑ) · G(ϑ)

PR = . (2.27)
L(d)

2.2.6 Noise and Interference

The received power is overlayed by thermal noise power N in the receiver and inter-

ference I caused by other users of the radio system IU and interference of other (radio)
systems IS . A commonly used measure of the quality of a radio transmission system is

the signal power (or carrier C) to noise N ratio signal to noise ratio (SNR) or carrier to
noise and interference I ratio signal to noise and interference ratio (SNIR). Assuming

an ideal bandpass filter at the receiver with centre frequency f and a noise equivalent
bandwidth B, the total noise power N in the receiver can be determined by
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 34

Figure 2.12: Radio link budget with interferer.

N = BN0 = BkT , (2.28)

where k = 1.38 · 10− 23 Ws/K is the Boltzmann7 constant, T the system temperature of
the receiver and N0 is the noise power spectral density. The system temperature of the
receiver is basically determined by the temperature picked up by the antenna aperture

and the noise of the first amplifier stage of the receiver.

Thus, the carrier C to noise power N ratio is

C EIRP · G EIRP · G/T

= = . (2.29)
N LN LBk

Likewise to the EIRP the gain to temperature ratio G/T is the typical characterisation
parameter of the radio receiver and called the figure of merit.

Interference caused by other systems IS can be described with given power flux density

spectrum PIS (f ) of the system and the above mentioned bandpass receiver filter as

fZ
+B/2

IS = PIS (f ) df . (2.30)
f −B/2

7
Ludwig Boltzmann, * 20.02.1844 in Wien, Austria; †05.09.1906 in Duino near Triest, Italy
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 35

The interference caused by one single other user Ii can be derived according to equation

2.27 and summed up over all interfering users NI , resulting in the cumulative interfer-
ence of other users IU .

NI
X NI
X EIRP (ϑi ) · G(ϑi )
IU = Ii = η i µi , (2.31)
i=1 i=1
L(di )

with µi being the activity factor of the interferer i and ηi the orthogonality factor of in-
terferer i. The activity factor reflects the burstiness of the interference traffic caused by

a terminal i whereas the orthogonality factor reflects the effective interference caused
by the terminal i. The predominant influence on the system performance is caused

by co-channel interferers, which are of particular interest, as for these ηi becomes 1.

In contrast to co-channel interference, adjacent or orthogonal channel interference be-

comes negligible.

The total carrier to noise N and interference I ratio is according to equations 2.29, 2.30,
and 2.31 determined by

C EIRP (ϑ) · G(ϑ)

= . (2.32)
N +I R
f +B/2
P
NI
EIRP (ϑi )·G(ϑi )
L(d) k B T + PIS (f ) df + L(di )
ηi µi
f −B/2 i=1

2.3 Regulations for Mobile Earth Stations

Regulations for radio emissions are typically given to protect different applications from
interference. Emissions by mobile terminals with directed antennas are subject to very

stringent regulations as antenna pointing errors in a mobile environment may occur more

often, depending on the agility of the mobile platform.

For mobile reception no special regulation apply, because here the antenna mispointing
leads only to degraded reception and does not cause interference to others. However
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 36

like any other receive terminal also collective mobile satellite terminals may be subject

to interference from other applications operating at the same frequency.

Regulations are given on international and national basis. On international basis the In-
ternational Telecommunication Union (ITU) defines spectrum allocations to the differ-

ent radio services (e.g. satellite, terrestrial, military, aeronautical) and rules international

interference.

National regulatory authorisations define the national spectrum policies and assign au-

thorisations to the spectrum users. In Europe the CEPT and the European Telecom-

munications Standards Institute (ETSI) provide harmonisation of spectrum policies and

technical standards between European countries.

In the international airspace and international waters beyond 125 km8 the ITU regula-

tions apply. In national airspace and national9 waters and international waters within
125 km the additional national regulations have to be considered.

Additionally to the applicable spectrum regulation and standards put into force by the
regulatory bodies, a coordination with the operator of the specific satellite and of the
adjacent satellites has to be performed. Moreover, the licensing conditions of the coun-

try apply, where the mobile platform is registered on which the satellite terminal in
installed.

Figure 2.13 details the frequency bands currently allocated for mobile and broadband

satellite communications, reflecting the higher frequency bands allocated for mobile

multimedia satellite communication.

Figure 2.13 shows the allocated frequency spectrum for mobile satellite communica-

tion. S and L band are classically used by MSS systems and the allocated spectrum is

appointed to this service for primary use. At the world radio conference (WRC) (the
tri-annual conference of ITU) in 2003 major steps have been taken to allow collective

mobile satellite terminals with aeronautical and maritime use at Ku band.

8
125 km between ESV and low-water mark as defined by the Coastal State
9
less than 12.5 nautical miles from coast
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 37

Figure 2.13: Frequency bands allocated to mobile (secondary allocations included) satellite
communication.

The secondary MSS allocation in the 14-14.5 GHz band was extended to include the
aeronautical MSS service. Before this date, the frequency band already included a sec-
ondary allocation to the MSS, but with the exclusion of the aeronautical MSS.

For earth station on vessels (ESVs) technical and regulatory concerns have been re-

solved and spectrum in the 5925 to 6425 MHz and 14 to 14.5 GHz bands was allocated.

Bidirectional satellite communications to trains was at WRC-03 not yet a topic for reg-
ulation, but with the increased demand of the train operators and successful technology

demonstration ETSI is putting a harmonised European standard (EN) for earth stations
on trains (EST) into force.

2.3.1 Aircraft

For aeronautical earth stations (AESs) the ITU-R recommendation M.1643 provides

technical and operational requirements. These requirements include for instance, that
the antenna is accurately pointed, that emission is under control of a network control

facility (NCF), and the protection of terrestrial and radio-astronomy sites by means of
 Table 2.1: Emission regulation for collective mobile terminals
international airspace & international wa- national airspace & international water < 125 km and national water & land
ters > 125 km
AES
• ITU-R Recommendation M.1643 • international conditions plus

• coordination with satellite operators • authorisation by overflying country (harmonised for Europe within ETSI EN
302 186)
• licensing from aircraft (A/C) registration
country

ESV
• ITU-R Resolution 902 • international conditions plus

• coordination with satellite operators • authorisation by coastal country (harmonised for Europe within ETSI EN 302
340)
• licensing from ship registration country

EST
• n/a • authorisation by country in which train is operating (harmonised for Europe
within draft ETSI EN 302 448)
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services
38
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 39

a power flux density limit per aircraft and a no emission restriction within 14.47-14.5

GHz. In European national airspace the harmonised ETSI EN 302 186 [ETS04] is in
force, covering the ITU-R recommendation.

2.3.2 Maritime

For ESVs the technical ITU conditions apply according to ITU-R Resolution 902

[ITU03c] which regulates the maximum emitted power, the tracking performance, and

states that a general control of the emission has to apply. In European national and in-

ternational waters within 125 km from an European coast the norm ETSI EN 302 340
[ETS05] is additionally regulating the emission of ESVs.

2.3.3 Land mobile Train

The harmonised ETSI standard for ESTs is derived on basis of the ESV ETSI EN.

For train no technical ITU regulation, but the draft ETSI standard ETSI EN 302 448
applies [ETS06a]. Only minor adaptations are made, e.g., no restrictions for the antenna
aperture size are given, as this parameter is meant for protection of adjacent satellites

interference and this is already regulated by the power flux density mask or a minimum

elevation angle relative to the horizon for transmission.

Table 2.1 provides an overview on telecommunications regulations and standards ap-

plicable for collective mobile satellite terminals at Ku band. The main technical re-

quirements which a collective mobile satellite terminal at Ku band has to cope with are
summarised in Table 2.2.
 Table 2.2: Main technical requirements for terminal emission at Ku band
Terminal type AES ESV EST
Antenna diameter - > 0,6 m -
Polarisation - Antenna polarisation polarisation Antenna polarisation polarisation
angle shall be adjustable within a angle shall be adjustable. a
range of at least 180◦
Tracking - Detection of antenna mispointing b same as ESV
Off-axis EIRP Maximum off-axis EIRP emission same as AES same as AES
density within an envelope c
Control and Monitoring The terminal is controlled and mon- similar to AES similar to AES
itored by a NCF
Cessation of emission The terminal has means to control similar to AES similar to AES
emissions including terminal iden-
tification and mechanisms to imme-
diately cease emissions
Emission to horizon On the ground, the AES transmits at Maximum ESV EIRP spectral den- The EST transmits at elevations ≥
elevations ≥ 7◦ relative to the local sity toward the horizon of 12.5 7◦ relative to the local horizon
horizon d dBW / 1 MHz and maximum ESV
EIRP towards the horizon of 16.3
dBW
a
For an adjustable polarisation range lower 90◦ the geographical region of operation with respect to the satellite position has to be declared.
b
Antenna mispointing shall be detected and emission switched off in case a pointing threshold of ∆ϑ is exceeded for more than T < 5 sec. ∆ϑ has to be designed in
compliance with EIRP spectral density limit
33 − 25 log(ϑ + ∆ϑ) − H [dBW/40kHz] for ϑmin ≤ ϑ + ∆ϑ ≤ 7, 0◦ ,
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services

c 12 − H [dBW/40kHz] for 7, 0◦ < ϑ + ∆ϑ ≤ 9, 2◦ ,

Maximum off-axis EIRP emission density
36 − 25 log(ϑ + ∆ϑ) − H [dBW/40kHz] for 9, 2◦ < ϑ + ∆ϑ ≤ 48, 0◦ ,
−6 − H [dBW/40kHz] for ϑ + ∆ϑ > 48, 0◦ ,
with ϑ is the off boresight angle, in degrees; ∆ϑ is the pointing error threshold, in degrees; H (in dB) is the maximum number of ESTs which always transmit at EIRPmax ;
and 2, 0◦ ≤ ϑmin ≤ 2, 8◦ depending on the satellite spacing.
40

d
except at locations where transmissions below 7◦ are permitted by the local administration
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 41

2.4 Characteristics of the Mobile Satellite Channel

2.4.1 Physical Phenomena of the Radio Channel

Communication between two parties uses a generic medium which is typically influ-

enced by physical phenomena resulting in attenuation or fading of the signal. In radio

communication the physical channel or satellite channel is exposed to multiple influ-

ences which are affecting the signal quality at the receiver [MB02]:

• Free space loss as described by equation 2.24 denotes the attenuation of the power

flux density with increasing distance from the receiver. In contrast to terrestrial
systems which operate at distances up to a few tens of kilometers, satellite systems
have to operate at very low link margins.

• Multipath propagation is caused by reflected signals from the surrounding envi-

ronment cumulating at the receiver with various weights, depending on the mor-
phological surface structure. Constructive and destructive superposition cause

fluctuations and phase shifting in the received signal. While terrestrial mobile

systems typically do not have direct visibility to the basestation and rely on multi-

path propagation conditions, satellite communication systems can gain only minor

advantages from multipath, because of the low link margins.

• Absorbtion is caused by ions, atmospheric gases (e.g. oxygen, water vapour) and
rain (rain fading), respectively, causing additional signal attenuation. The ITU

provides a set of recommendations dealing with atmospheric phenomena (e.g.

[ITU03b], [ITU99a], [ITU99b]. Attenuation due to rain is especially dominant in

the region of 20 and 30 GHz, while ionospheric effects are of minor importance

at frequencies above 10 GHz according to [ITU03a].

• Scintillation, caused by irregularities and local variations of electron density in

the atmosphere causes deflection along the signal propagation path. According to
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 42

[ITU95b] scintillation is significant for frequencies below 1 GHz particularly at

polar and equatorial regions and of less importance at frequencies used by MSS.

• Faraday-rotation of linear polarised electromagnetic waves may cause a polari-

sation mismatch at a receiver with corresponding linearly polarised antenna. At

lower frequencies circular polarisation may be used as a countermeasure, while

depolarisation is negligible at higher frequencies.

• Interference as described in equation 2.30 and 2.31 is caused by other radio sys-

tems or other users of the same communication system and is overlaying with the

wanted receive signal.

• Additive White Gaussian Noise is caused by thermal noise in the receiver as de-
scribed in equation 2.29.

2.4.2 Influences of Terminal Mobility on the Satellite Channel

Mobility of the communication partners introduces temporal and spatial fading statis-

tics to the radio signal determined by the vicinity of the mobile terminal. The relative
movement of the satellite and mobile terminal results also in a frequency shift to be

compensated. Both effects are detailed in the following.

2.4.2.1 Doppler Characterisation

The shift of the transmitted signal frequency with respect to the received frequency is

called Doppler 10 effect. The Doppler shift is defined by,

v
fD = · fc . (2.33)
c0
10
Christian Andreas Doppler; * 29.11.1803 in Salzburg; †17.03.1853 in Venezia
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 43

The velocity v is the instantaneous rate of change of the slant range as described in

equation 2.10 and 2.11. It is caused by the mobile’s velocity and the movement of the
satellite with respect to the earth and can be split into these two components.
frequency shift in Hz

time in sec

Figure 2.14: Doppler shift variations recorded during an aeronautical channel measurement
campaign at 18 GHz centre frequency

For a mobile to GEO satellite scenario the velocity v is determined by,

vz (rT − rS · cos∆λ · cos ∆δ) + vlat · rS · sin ∆λ − vlong · rs · cos ∆λ · sin ∆δ

v= p ,
rT2 − 2 · rT · rs · cos ∆δ · cos ∆λ + rs2
(2.34)

with:

rS Distance Earth centre to satellite

rT Distance Earth centre to mobile terminal

∆δ latitude of the mobile (positively counted if mobile is northern hemisphere)

∆λ longitude difference of mobile and satellite (positively counted if mobile is east of

the satellite)

vlat latitude velocity of mobile (positively counted if mobile is heading north)

vlong longitude velocity of mobile (positively counted if mobile is heading east)

vz vertical velocity of mobile (positively counted if mobile is ascending)

CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 44

The movement of a GEO satellite with respect to its nominal position due to orbit per-

turbations can be neglected.

Figure 2.14 shows the measured Doppler shift during an aeronautical channel mea-
surement [HJL99]. The worst case Doppler shift in a mobile to GEO scenario can be

expected in the aeronautical case, because of the high speed of the mobile platform. At

Ku band it is about 15 kHz.

The velocity v for a fixed terminal to non-GEO satellite scenario neglecting the earth’s

rotation is determined by,

sin ψ
v = 2πrs rT . (2.35)
dT

The worst case velocity due to the earth rotation occurs at equator in case the satellite is
visible at εmin = 0◦ and is determined by,

RE
v = 2π cos εmin . (2.36)
TE

From equations 2.36 and 2.33 follows a worst case Doppler shift introduced by the

earth’s rotation of fD = 1.55 · 10−6 fc , resulting in approximately 3 kHz at 2 GHz

carrier frequency.

Figure 2.15 shows the Doppler variations for a carrier frequency of 2 GHz during a LEO

and MEO passage. It can be seen that this effect is dominating the Doppler variations
caused by the mobile’s velocity.

2.4.2.2 Physical Propagation Paths

The environment in the vicinity of the mobile typically allows multiple physical propa-
gation paths. Thus, the radio signal consists of several superposition signals, which are

weighted with the antenna gain associated with each incident angle of the signal.
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 45

125
LEO 10 GHz
100

Doppler speed in kHz/min (dashed)

75 MEO 10 GHz
Doppler shift in kHz (solid)
50 LEO 2 GHz

25 MEO 2 GHz
0
−25
−50
−75
−100
−125
−60 −50 −40 −30 −20 −10 0 10 20 30 40 50 60
time in min

Figure 2.15: Doppler shift (solid lines) and Doppler shift speed (dashed lines) for a carrier
frequency of 2 and 10 GHz during a LEO (800 km) and MEO (10000 km) overpass.

Figure 2.16: Direct path (LoS, blocked, and shadowed), Reflected / scattered path, Diffracted
path.

Figure 2.16 shows the propagation paths for a LMS channel, which holds to a certain

extend also for aeronautical and maritime channel.

CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 46

• Direct path: A signal reaching the receiver with free line of sight (i.e. undis-

turbed) is called a line of sight (LoS) signal. Blockage and shadowing to the
direct path is caused by obstacles in the propagation path. Blockage and shadow-

ing is distinguished depending on the severity of the signal loss. Blockage in the

LMS is caused e.g. by man made structures as tunnels or buildings while vege-
tation is causing typically signal shadowing. Blockage and shadowing is of less

importance for aeronautical and maritime channel.

• Diffracted path: Diffraction is caused by the partial obstruction of the propagation

path and results in distortion of the signal. Of special importance for the satellite
to mobile propagation is the near field diffraction in the vicinity of the mobile.
11
The so called first Fresnel zone is defining a volume in the antenna radiation

pattern which should be kept free from obstructions. In order to avoid signal
attenuation obstacles outside the Fresnel zone are considered of marginal impact
to the signal strength.

• Reflected / scattered path: Reflected or scattered radio signals may occur due to
the mobile to satellite geometrical relation. Depending on the ratio between wave-
length and surface structure of the mirroring object the radio signal is reflected or

scattered. At low elevation angles reflection at sea or ground is becoming rele-

vant for the aeronautical and maritime satellite channel. Surface state of the sea

respectively morphological data of the ground surface is heavily influencing the

signal quality.

Signals on reflected/scattered paths are called echoes since time delay is caused by the

longer propagation path compared to the direct path. All of these propagation paths are
affected by the physical phenomena described in chapter 2.4.1 and the Doppler shift

above.

Depending on the frequency and the antenna characteristics these effects are more or
11
Augustin Jean Fresnel; * 10.05.1788 in Brogue; †14.07.1827 in Ville-d’Avray
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 47

less distinct. Generally, the use of high gain antennas make shadowing effect more

dominant than reflection, as the small antenna beamwidth discriminates reflected and
scattered signal paths.

In terrestrial systems scatterers can be found in the vicinity of both basestation and
mobile terminal whereas in the satellite to mobile geometrical relation, relevant objects

which are influencing the radio propagation are obviously only in the vicinity of the

mobile terminal.

The mobile to satellite geometrical relation differs considerably from a terrestrial sce-
nario by the ratio of the distance between mobile and satellite respectively basestation.

Even for LEO satellites the distance of the mobile terminal to the satellite is bigger by

order of magnitudes than the distance from the mobile to the scatterers. This results in a
very narrow angular spread of the received signals, in contrast to the terrestrial scenario

e.g characterised in [PMT01].

The direct path components in down- and uplink are typically reciprocal correlated.
Effects like diffraction, reflection and scattering in the near field of the mobile is causing

uncorrelated fading with different influences to the down- and uplink signal [VTL95],
[PEBB96], [JHE+ 99].

2.4.3 Mobile Channel Characterisation

For mobile communication frequencies at L, S, X, Ku, Ka and EHF (V) band (cf. Fig-

ure 2.13) are of interest and thus channel characterisation for mobile terminals is of

importance at these frequencies.

Mobile channel characterisation is especially focused on multipath effects. Other phe-

nomena of the radio channel as described in section 2.4.1 can be overlayed to the mul-
tipath characterisation and ITU-R provides a set of recommendations based on fixed,

long term measurements in different regions of the world.

CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 48

Bello introduced a fundamental characterisation of time and frequency variant radio

channels with a description of the physical channel by means of power delay profiles
and Doppler spectra to explain and reproduce the observed fading statistics. According

to [Bel63], [Par92] the characterisation of multipath effects is done by a linear time-

variant channel impulse response h(t, τ ), with t being the observation time and τ the
delay or dispersion of time caused by the echoes. This channel impulse response h(t, τ )

is derived by the output of the channel at time t, with a Dirac impulse as input to the
channel at time (t − τ ).

The received signal r(t) resulting from a transmit signal s(t) is given by the convolution

between the transmitted signal and the channel impulse response:

Z ∞
r(t) = s(t) ∗ h(t, τ ) = h(t, τ )s(t − τ )dτ . (2.37)
−∞

The characterisation shows the two effects i) how fast the channel impulse response
varies with t and ii) what delays τ occur on the channel.

Fourier transformation of the channel impulse response into frequency domain by the

observation time t and the delay τ leads to a total of four equivalent characterisations

of the channel. These are named time variant impulse response h(t, τ ), time variant
transfer function T (t, f ) , delay-Doppler spread function S(fD , τ ), and Doppler-delay

spread function H(fD , f ) [Bel63].

T (t, f ) = Fτ {h(t, τ )}. (2.38)

S(fD , τ ) = Ft {h(t, τ )}. (2.39)

H(fD , f ) = Ft {T (t, f )} = Fτ {S(fD , τ )}. (2.40)

CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 49

Since the time behaviour of the channel impulse response is not known accurately

Bello introduced the wide-sense stationary uncorrelated scattering (WSSUS) assump-

tion. This assumption enables the characterisation of the channel by its statistical corre-

lation functions which can be obtained by measurements. The channel is in wide sense

stationary in case the autocorrelation of two impulse responses is independent of the

absolute observation times t1 and t2 and only determined by the difference t1 − t2 . Un-

correlated scattering assumes that the different scattered signals coming from different
location (or delays) are uncorrelated in amplitude and phase.

Measures for the variability of the channel with resect to t respectively f are the Doppler

spread BfD (and associated coherence time Tc ) derived by the Doppler power spectral
density and the delay spread ∆τ (and the associated coherence bandwidth Bc ) derived
by the delay power spectral density.

The power spectral density defines how much spectral broadening the channel
causes. The delay power spectral density is the Fourier transform of the frequency-

autocorrelation function and defines the average power of echoes vs. delay τ . Doppler
power spectral density is the Fourier transform of the time-autocorrelation function and
shows the spreading of a sinusoidal carrier in the frequency domain [Jak75].

The symbol length TS and the signal bandwidth BS in relation to the coherence time
TC and coherence bandwidth BC gives a classification of the radio channel into time

non selective channel in case TC /TS À 1 and frequency non selective channel in case
BC /BS À 1.

A frequency non selective channel is also known as narrowband channel. Assuming

that for non code division multiplex systems the ratio of symbol length TS and signal
bandwidth BS is approximately BS ≈ 1/TS the narrowband assumption holds for BS ≤

1M Hz in the L band, and even up to BS ≤ 30M Hz in the EHF band according to

[LWJ00].
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 50

2.4.4 Lutz Channel Model

In N-state Markov12 models the mobile satellite channel behaviour is determined by the
channel state. The channel state, e.g. LoS or non-LoS is described by the dominating

propagation path. These different channels states can be assigned to a Markov process
with a certain probability for state transition.

p11 p12

1 2
p22
p21

Figure 2.17: Two-state Markov process, e.g., used in the Lutz model.

Figure 2.17 shows as example a two state Markov process as used for the Lutz model

described hereunder. The two states can be assigned to a LoS and shadowed channel
condition. The transition probabilities between the two states are inversely proportional

to the mean durations for which the channel remains in the states. The Markov process
assumes that only the current state of the process is relevant to determine the next state
(i.e. memoryless process). Furthermore, a homogeneous Markov process is used for the
modelling of radio channel characteristics assuming that the transition probabilities are

independent of time.

The Markov process is represented by a set of probabilities

P (sn+1 = j|sn = i) , (2.41)

with n being the number of states and 0 < i, j < n.

A N-State model Markov process can therefore be described by its transition probabili-

ties written as a matrix of size N.

12
Andrej Andreevič Markov; * 14.06.1856 in Rjasan; †20.07.1922 in Petrograd
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 51

 
p p12 · · · p1N
 11 
 
 p21 p12 · · · p2N 
 
(pij ) = P =  . .. ..  . (2.42)
 . ... 
 . . . 
 
pN 1 pN 2 · · · pN N

A model which can be used to generate time series of narrowband fading is the Lutz

model [LCD+ 91]. The Lutz model is based on a two state Markov process as described

above. The LMS channel is characterised by a good ’unshadowed’ and bad ’shadowed’

state. The good state is characterised by a Ricean [Ric48] distribution of the signal

power envelope and the bad state with a Rayleigh13 distributed fading.

• Good (LoS) state:

√
In case the direct path is unobstructed, the received signal envelope S follows a
Ricean distribution function and consists of an overlay of the strong direct compo-
nent and the multipath components. The probability density P of the momentary
receive power S can be characterised by a Ricean distribution

√
PRice (S) = ce−c(S+1) I0 (2c S) , (2.43)

with I0 (x) being the modified Bessel 14 function of order zero and the Rice factor

c describing the ratio between direct and multipath components [LWJ00].

• Bad (non-LoS) state:

In case of shadowing the envelope of the sum of reflectional scattered multipath

components is modelled with a Rayleigh distribution with a short term average

power S0 , the probability density of the power follows an exponential distribution

1 − SS
PRayleigh (S|S0 ) = e 0 . (2.44)
S0
13
John William Strutt (Lord Rayleigh); * 12.11.1842 in Langford Grove; †30.06.1919 Terlins Place
14
Friedrich Wilhelm Bessel; * 22.07.1784 in Minden; †17.03.1846 in Königsberg
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 52

The average power in non-LoS S0 may vary for each non-LoS state. Typically,

this variance is modelled by a Lognormal distribution

µ ¶
10 (10 log S0 − µdB )2
PLN (S0 ) = √ exp − 2
, (2.45)
S0 2πrσdB ln 10 2σdB

with σdB the received power variance of the averaged shadowed component and
µdB is the mean power level decrease in dB.

Finally the resulting probability function of the received power can be derived by a

combination of the above mentioned distributions. The ratio of mean duration for which

the channel status is bad to the entire observation time is called shadowing factor A.

Z ∞
PLutz (S) = (1 − A) PRice (S) + A PRayleigh (S|S0 )PLN (S0 )dS0 . (2.46)
0

In [Lut96] the Lutz model is enhanced to include besides the shadowing characterisa-

tion also the angular correlation of two satellite channels. Satellite diversity as described
earlier is a very powerful countermeasure for land mobile satellite channels. The corre-

lation of the channel states of two satellite channels h1 (t, τ ) and h2 (t, τ ) in dependence

on the azimuth separation ∆ϕ and elevation separation ∆ε can be modelled as a Markov

model. Thus, the two shadowing states of the both channels are resulting in four states
of the new model.

2.5 Literature

2.5.1 Mobile Satellite Channel

The time variant mobile satellite channel influences satellite to mobile transmission and
thus mobile satellite services. Planning of countermeasures requires knowledge of the

satellite channel characteristics.

CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 53

Terrestrial wireless system design and cellular planning tools are often based on deter-

ministic channel propagation models which determine channel parameters from topo-
graphical and morphological databases. These models used to have a minor role for

satellite channel modelling so far, because the accuracy of the databases was not suffi-

cient to determine reliable satellite channel characteristics. A combined deterministic

and statistic model is presented in [DJKB98].

Empirical LMS channel models have been derived by Goldhirsh and Vogel. The model

defines the signal attenuation in dependence of the percentage of the driven path length,

the elevation angle and frequency. The ERS model [GV88] was enhanced for a wider
frequency range, elevation range and probability range in the combined empirical road-
side shadowing (CERS) [BER92], the modified empirical roadside shadowing (MERS)
[SBB93], and the extended empirical roadside shadowing (EERS) [GV95] models,

which is the most cited empirical model. An overview of different empirical channel

models is given [KC96].

Determination of statistical channel parameters of the satellite channel variance is one

possibility for characterisation. Channel models can be derived from the statistical pa-
rameters. Statistical distribution functions for describing the time variant fading are

Rice [Ric48], Rayleigh [Jak75], Lognormal [Loo85], and Suzuki [Suz77] distributions.

The Lutz Model [LCD+ 91] combines Rice and Rayleigh / Lognormal by a Markov pro-
cess for characterisation of the LMS channel. Neul [Neu89] presents an aeronautical

model based on LMS data. An overview of different statistical channel models is given
[DEJ+ 98].

A recent survey on LMS channel models is given in [HHG10].

Besides the modelling a wide range of satellite channel measurements have been carried

out by e.g. the European Space Agency (ESA), the German Aerospace Center (DLR),
the Jet Propulsion Laboratories (JPL), the University of Surrey, the University of Texas

and associated channel models and parameters have been published. The following table
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 54

gives an overview of those which are of relevance for the measurements performed and

parameters presented within the thesis.

Table 2.3: Overview of Satellite Channel Measurement Campaigns Status

Frequency Band Type Reference
UHF L LMS [GV88]
L LMS [Lut88]
L S Ku LMS [BER92]
Ka LMS [Dav93]
LS LMS [PEBB96]

At higher frequencies, the availability of data in 1996 was scarce and this is still the case.

LMS data at EHF band and aeronautical channel data at higher frequencies was com-
pletely unknown before the measurement campaigns presented within this thesis. This

thesis contributes with statistical Lutz model parameters of a LMS channel measure-
ment campaign at 2 GHz (S band), a LMS channel measurement campaign at 40 GHz
(EHF band) and an aeronautical channel measurement campaign at 20 GHz (K band).
The thesis presents a satellite visibility analysis by ray tracing of precise, wide area ur-

ban 3D models, which became accessible in 2002 and overcoming the restrictions of
topographical databases until this date.

2.5.2 Satellite Terminal Antenna Pointing, Tracking and Acquisi-

tion

Emerging MSS in US brought along the topic of using higher gain antennas and the

need of pointing and tacking systems. [BN86] proposes in an early publication about

steered mobile satellite antennas a combined open loop closed loop tracking mechanism
for a 10 dBi higher gain antenna. Intensive work has been done at Jet Propulsion Labo-

ratories (JPL) in the late 1980s for higher gain antennas [Woo88] and later for very high

gain antennas within the Advanced Communications Technology Satellite (ACTS) pro-
gramme [Den93]. The Defense Satellite Communications System (DSCS) also called

for steered mobile terminal antennas [RCD94].

CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 55

Different antenna concepts have been proposed such as mechanically steered [KYE88],

electronically steered [HHE89], and hybrid [SLH+ 00] with various techniques for ex-
traction of a receive signal of a closed loop pointing like monopulse [BB88] or dither-

ing [DG96]. Monopusle techniques require additional receiver effort e.g. a four array

receiver to extract a steering signal, dithering poses a constant movement to the an-
tenna mechanics. The thesis presents a conical scan design for a mechanically steered

Cassegrain antenna which imposes no additional movement to the antennas elevation

and azimuth axes at a very high scanning speed which is resulting in high tracking per-

formance.

Most publications summarised technology demonstrations of antennas and antenna

steering platforms, but missing to define minimum angular accuracy requirements appli-
cable to the mobile platform and type of antenna. The thesis provides a design method-

ology for cost optimised terminal antennas and PAT systems utilised in different mobile

environments and a general approach for optimisation of pointing error vs. antenna
gain.

2.5.3 System Dimensioning for Collective Mobile Satellite Termi-

nals

System dimensioning for mobile satellite networks started with models for subscriber

distribution, service penetration, and acceptance models extended by the inhomogene-

ity of global traffic flows and daily usage variation. Until the mid of 1990’s satellite
communication systems have been focusing on telephony services. Thus, system di-

mensioning has been based on Erlang traffic models using parameters derived from
telephone traffic measurement in the global PSTN [BJL+ 93] [Vio95]. The introduction

of asynchronous transfer mode (ATM) networks offering multiple services in an inte-

grated network called for extending the single service models to multi service models

[WHS97].
CHAPTER 2. Satellite Communications Fundamentals for Mobile Services 56

System dimensioning for aeronautical satellite systems used for communications collec-

tively used by a group of passengers has not been in the focus of research until ACTS
technology demonstration in US [AP95] and ABATE in the EU [HJ98a] showed feasi-

bility of such a service. Traffic models on which a system dimensioning could be based

on have not been available.

The work presented in this thesis extends the multi service model presented in [WHS97]

by collective mobile groups and layers the model to extract numerical values on group,

spot beam, satellite and system level. Service mix, usage parameters, group size, tem-

poral and spacial user and group distribution, bit rates, traffic volumes, and handovers
are presented for the aeronautical case and reused with various tuning on parameters in
many later projects (cf. table 1.1).
Chapter 3

Mobile Satellite Channel

Satellite Spectrum and Term. Wireless Channel

Satellite Radio Link
Constellation Antenna Regulation Characteristics
Chapter 2

Mobile Satellite Channel

Chapter 3
Mobile Platform Mobile Terminal
Characteristics Design
Chapter 4
Mobile Group Capacity and
Mobile Services
Characteristics Handover Analysis
Chapter 5

The design of a satellite system for mobile applications requires detailed knowledge of
the satellite to mobile terminal propagation path. This chapter will present the mobile

satellite channel characterisation as a basis for the identification of influences on the

various aspects of mobile satellite system and especially terminal design as well as

appropriate countermeasures against radio signal fading and blockage.

Long signal propagation paths and the associated free space signal loss force satellite
systems to operate with very limited power margins. In contrast to terrestrial mobile

systems, where link margins and multipath propagation allow to overcome signal at-
tenuation by obstacles like buildings or trees, satellite communication systems require

typically a direct, undisturbed signal propagation path from the mobile terminal to the

satellite. This effect becomes especially predominant for higher frequencies and narrow
beamwidth antennas.

57
CHAPTER 3. Mobile Satellite Channel 58

Thus, satellite visibility for a terminal influences fundamentally the performance of a

satellite communication system. The visibility is the main characteristics of the mobile
satellite channel, which suffers from time variant fading, because of the terminal’s and

satellite’s movement. Systems with non-GEO satellites show apart from the influence

of the user’s mobility, a time varying delay and free space signal loss due to satellite’s
movements relative to the receiver.

The channel characteristics is basically dependent on the frequency band, the topolog-

ical and morphological environment, the directivity of the used antenna, the satellite
constellation parameters (mainly the elevation under which the satellite is seen from the

terminal, because low elevation angles make the signal more susceptible to obstructions)
and the terminal’s mobility.

Severe satellite channel blockage, for instance in urban environment, leads to conse-
quences for the handover process and poses requirements for the signal acquisition time

of a steered high gain antenna with narrow beamwidth. Fading events with fast signal

loss reduces the time left for handover. Short, but frequent signal losses may increase
the synchronisation overhead, moreover, the movement of the mobile during a signal
fade poses severe problems, if steered small beamwidth antennas with high gain are

used.

For the design of collective mobile terminals the statistical characterisation of the satel-
lite to mobile propagation channel becomes an essential input. Thus, in the following

characteristics of the satellite to mobile satellite channel will be introduced. Chan-

nel characterisation parameters for mobile platforms in land and aeronautical scenarios

have been derived in this work by measurements as well as by deterministic modelling.

Theoretically the satellite to mobile channel can be entirely characterised by the trans-

mission system functions (cf. chapter 2.4.3). However, the computation of all echoes,

the Doppler delays, and the phase shifts is practically unfeasible since the necessary time
behaviour input for computation is not available in detail. Pure deterministic computa-

tion of the channel impulse response h(t, τ ) would require a complete electromagnetic
CHAPTER 3. Mobile Satellite Channel 59

description of the environment in the vicinity along the mobile path with the involved

geometry of manmade and natural objects (buildings, mountains, vehicles or people)

including their surface structures and dielectricity.

Statistical channel models are used to describe the properties of the channel. Character-
isation of the channel impulse responses are derived from these models. To determine

the model parameters, measurements have been carried out and the model parameters

have been derived.

3.1 S Band LMS Channel Characterisation by Mea-

surement

A measurement programme at S band was performed to describe the ICO subscriber

link [JHE+ 99]. ICO was a planned system for voice and data communication and based

on a MEO constellation with 10 satellite in 2 planes at 10390 km orbit height (cf. Figure
2.1) [Pos96].

Channel data has been collected in various environments to characterise the performance
of the mobile user link in up- and downlink direction. The experiments used also a

second diversity path to compare the channel quality on different links. In this way,

amplitude and phase correlation between return and forward links can be investigated
as well as diversity and handover strategies.

3.1.1 S Band LMS Channel Measurement Setup

The measurement campaign used an Zeppelin airship to emulate slow angular changes
of the satellite orbit characteristics (cf. figure 2.4). A second transmitter/receiver was

placed on top of high buildings to simulate diversity reception. A user terminal (UT)
was built to transmit and receive a continuous wave (CW) signal in different frequency
CHAPTER 3. Mobile Satellite Channel 60

Zeppelin NT 01

2200 MHz
Tower 1986 MHz

2195 MHz

1986 MHz ICO UT

Figure 3.1: Channel measurement setup with UT and diversity base stations at Zeppelin and
tower.

bands simultaneously. The frequency bands used were: 1986 MHz for the subscriber
up-link, 2200 MHz and 2195 MHz for the subscriber down-link.

Figure 3.2: LMS S Band measurement equipment.

Figure 3.2 shows pictures of the measurement van and UT on tripod in open area (left);

the UT detail with original antenna and in mockup housing (middle), and the Zeppelin
(right).

Figure 3.3 shows the equipment setup for user terminal. In figure 3.4 the equipment

setup of the Zeppelin is shown. The tower equipment was equal to the Zeppelin in
downlink path and similar in uplink path differing only in the necessary filter adaptations

for the uplink frequency.

The received signal was recorded in inphase and quadrature component on a digital dig-

ital audio tape (DAT) recorders. Additionally, the positions and the attitude of the user
and Zeppelin were recorded on laptop computers. A video footage of the user oper-

ation was recorded during all tests, too. Differential global positioning system (GPS)
CHAPTER 3. Mobile Satellite Channel 61

f_RX1 = 2200MHz
f_RX2 = 2195MHz
A<0.9dB
@ f=f_RX1 or Handheld
f=f_RX2
A>90dB f_c=2185MHz
@ f=f_TX A_12<1dB
A_13>25dB
AFD3 @ f=f_RX or
G=37dB f=f_TX
U=15V
I=0.1A
P_in<-25dBm LNA
Low loss
cable
A<6dB connection

f_c=1985MHz
F_c=2200MHz A<2dB @ f=1985MHz

ZKL 2R5
G=28dB ZVG 8G
U=12V
AMP G=30dB
I=0.2A HPA P_out=30dBm
P_in<-15dBm U=12V
P_out=16dBm I=2A
ZKL 2R5 AMP
G=28dB U=15V, I<0.15A
U=12V U=5V, I=0.4A
I=0.2A (each Synt.)
P_in<-15dBm
3dB LO f=f_TX=1986MHz
ZFM-15
f=2125MHz A=6dB A<6dB A=6dB f=2130MHz
R10 A_RF=24dB
LO ATT A_LO=20dB ATT LO 3dB R10.
P_LO=10dBm
f_c=70MHz,
BW 150kHz
A=14dB A=6dB ATT ATT A=3dB
Zeppelin signal

AM112
Tower signal

G>15dB X-tal Osci.

P_out=6dBm
AMP U=15V AMP LO U=15V
I<0.05A
I=0.4A
f_c=70MHz
f=10MHz
+/-5kHz
A<6dB
@ f=f_c
AM112 External connection
G>15dB
U=15V
AMP I<0.05A AMP Internal connection
WORK
Synthesizer

ZFMIQ-70D LO ZFMIQ-70D
A=8dB P_LO=10dBm
P_LO=10dBm

f=70.005MHz
BLP-2.5
f_u=2.5MHz
A<2dB
@ f=5kHz
Ch3
IRIG GPS
DAT
Ch2 Ch1 Clock

Figure 3.3: LMS S Band user terminal up/downlink equipment.

receivers were used for enhanced accuracy of the position data. Synchronised reference

clocks in all terminal were used to time-stamp the data on DAT, laptop and video.

The user terminals had a dynamic range of the overall measurement set-up of better

than 30 dB at worst case conditions. During nominal conditions the dynamic range of
CHAPTER 3. Mobile Satellite Channel 62

f_RX = 1986MHz

f_c=1995MHz 2 f_c=2200MHz
A=1dB
@ f=f_RX 1
3
A>93dB A=2dB @ f=2.2GHz
@ f=f_TX
A_12<0.6dB
AFD3 A_13>25dB ZVG 8G
G=37dB @ f=f_RX or G=30dB
U=15V LNA f=f_TX HPA P_out=30dBm
I=0.1A U=12V
P_in<-25dBm I=2A
ZKL 2R5 f=f_TX=2200MHz
G=28dB P_out=16dBm
U=12V AMP LO U=15V, I<0.15A
I=0.2A U=5V, I=0.4A
P_in<-15dBm (each Synt.)
A=6dB
ZFM-15
A=4dB
A_RF=24dB ATT LO
A_LO=30dB
P_LO=10dBm
f=1916MHz
AM112 ATT A=6dB ATT
G>15dB
U=15V AMP
I<0.05A
X-tal Osci.
P_out=6dBm
f_c=70MHz
U=15V LO
+/-5kHz
I=0.4A
A<5dB
f=10MHz
@ f=f_c

AM112
G>15dB
U=15V
AMP
I<0.05A WORK
Synthesizer
ZAD-1
A=6dB LO External connection
P_LO=10dBm
f=70.005MHz
Internal connection
BLP-2.5
f_u=2.5MHz
A<2dB
@ f=5kHz

Ch2
IRIG GPS
DAT
Ch1 Clock

Figure 3.4: LMS S Band Zeppelin up/downlink equipment.

the equipment guaranteed a signal reception with a C/N of 48.9 dB measured in 2 kHz

bandwidth.

Various trials have been performed in June 1999 in the Bodensee area near Lindau and

Friedrichshafen, Germany. The trials reflected several operational scenarios and envi-
ronments. The environments comprised open, rural (various degrees of shadowing),

suburban, urban, indoor hotel, indoor residential, indoor office, lakeside and ferry en-
vironments. Operational scenarios have been typically handheld scenes, but also in-car
CHAPTER 3. Mobile Satellite Channel 63

scenarios, and scenarios where the handset was placed in pockets, briefcases, on tables,

bedside tables and so forth.

Figure 3.5: Measurement browser with video of surrounding and data, and the map of the
surrounding.

In total, the amount of data that has been recorded corresponds to an useful measurement

time of 15 hours. Data post-processing and error correction was applied to correct the
received data with respect to varying free space propagation losses and antenna gain
variations due to different flight attitude of the Zeppelin. The recorded channel data

together with the video footage of the respective trial is displayed in Figure 3.5 for a
driving environment.

The user terminal is held at the window with an elevation to the Zeppelin of over 70◦ .
The main screen displays the video and the data plot as well as the map. In the map,

the positions of the tower (red), and the paths of the UT (magenta) and the Zeppelin
(yellow) are marked.

In Figure 3.6 the power series and the azimuth and elevation plot are depicted in better

resolution. The upper two figures of the signal plot show the receive power of the user
CHAPTER 3. Mobile Satellite Channel 64

Figure 3.6: Measurement browser with detailed power series (upper) and azimuth and elevation
plot (lower).

terminal from the Zeppelin and the tower, the lower two figures the receive power at the

Zeppelin and the tower.

CHAPTER 3. Mobile Satellite Channel 65

A detailed set of environments and operational scenarios have been defined. The en-

vironments and test scenarios for the channel measurements are characterised by five
classes:

• Mobile Terminal User Environments; according to Table 3.1.

In total measurement series at ten different mobile terminal environments have

been performed.

Table 3.1: Mobile Terminal Environments

EN #, Code Description Comment
EN1, O Open Clear sky, no obstructions
EN2, R Rural, light shadowing One-sided tree shadowing
EN3, M Rural, medium shadowing Group of trees and bushes
EN4, T Rural, heavy shadowing Heavy tree shadowing
EN5, S Suburban Residential areas
EN6, U Urban 4 - 5 storeys, heavy traffic
EN7, H Indoor, hotel Small room, typical furniture
EN8, W Indoor, office building Open-plan office
EN9, I Indoor, residential house Living room with furniture
EN10, F Ferry At lakeside and on ferry
EN11, A Autobahn Similar to open, but driving

• Operational Scenarios; according to Table 3.2.

Different operational scenarios for the user terminal have be defined to ensure

measurement series for fixed terminal positions and with user interaction.

• Satellite (Zeppelin) Flight Scenarios; according to Table 3.3.

Basically, two modes of operation were adopted for the Zeppelin flight paths: i)

hovering above fixed positions or flying circles around the user terminal position
which is resulting in fixed elevations, and ii) flight trajectories to simulate the el-

evation and azimuth characteristics of an ICO satellite from a given user position.
Elevation ranges from 5◦ to 85◦ with flight altitudes between 1100 m or 1600 m

above ground.

• UT Antenna Position; according to Table 3.4.

CHAPTER 3. Mobile Satellite Channel 66

Table 3.2: Operational Scenarios

OS #, Code UT Position Comment
OS1, T Tripod-mounted Fixed position
OS2, S Handheld, user is standing With user interaction
OS3, M Handheld, user is walking With user interaction, walking figure ’8’
OS4, P Handheld in pocket Fixed position
OS5, B Handheld in briefcase Fixed position, several briefcase types
(leather, nylon, aluminium case etc.)
OS6, D Vehicle at speed, dashboard Fixed UT position in car
OS7, V Vehicle at speed, handheld With user interaction
OS8, W Indoor, held near window With user interaction
OS9, Y Indoor, on table Fixed position
OS10, Z Indoor, on bedside table Fixed position in hotel

Table 3.3: Satellite Flight Scenarios

FS #, Code Flight trajectory Comment
FS1, F Hovering or circles Fixed elevation angles in 5◦ steps
FS2, S Satellite flight trajectories Simulate satellite overpass

Three different antenna positions were used at the user terminal to emulate typical
user behaviour.

Table 3.4: Antenna Position

AP #, Code Antenna Position Comment
AP1, S Antenna extracted, straight
AP2, E Antenna extracted, 45◦ Normal position above user’s head
AP3, F Antenna folded for call signalling environments

• Diversity Scenarios

The diversity measurement to the tower have been performed for most environ-
ments. In some scenarios diversity measurement were not possible, for instance

if the mobile moved to locations with blocked conditions to the tower or moved
even outside the tower transmitter coverage.

Table 3.5: Diversity scenario

DS #, Code Diversity Comment
DS1, N No diversity
DS2, V Visibility to second satellite
CHAPTER 3. Mobile Satellite Channel 67

3.1.2 S Band LMS Channel Measurement Results and Parameters

Figure 3.7: Surrounding and power plot at rural environment with sequential tree shadowing to
tower and to Zeppelin.

A measurement in a suburban user environment is plotted in Figure 3.7 (EN5,S; OS3,M;

FS1,F 45◦ ; AP2,E; DS2,V;). The user is walking towards a leafy tree with the terminal
and the antenna unfolded. Firstly the tower is shadowed by this tree. At about second
35 of the power series the path to the tower get free sight, whereas the user is walking

and remaining under the tree up to second 75 of the plot. The tree results in a signal loss
of about 20 dB.

Figure 3.8: Surrounding and power plot at lakeside environment with multipath fading.

Figure 3.8 shows a measurement at the lakeside (EN10,F; OS3,M; FS1,F 5◦ ; AP1,S;

DS1,N;). The user is slowly walking from the lakeside backwards and returns to the
lakeside. The Zeppelin is hovering at low elevation above the lake. The tower is not

visible during this measurement set. Especially in the receive signal at the user terminal
CHAPTER 3. Mobile Satellite Channel 68

a strong multipath fading can be observed, originating from a reflected second path at

the lake.

Figure 3.9: Surrounding and power plot at open environment with heavy rain and umbrella
shadowing to Zeppelin.

Figure 3.9 shows a measurement in an rural light shadowed user environment (EN2,R;

OS2,S; FS1,F 45◦ ; AP2,E; DS2,V;). The user is walking with the terminal and the
antenna unfolded. The tower is shadowed by small trees leading to a signal decrease

of about 7 dB. The Zeppelin is visible until second 53 and then shadowed by a wet
umbrella. This results in a signal loss of about 4 dB.

Figure 3.10: Surrounding and power plot at hotel while blinds are opened.

Figure 3.10 shows the power plot measured in typical a hotel room while the metallic

blinds are opened (EN7,H; OS2,S; FS1,F 10◦ ; AP2,E; DS2,V;). The user terminal is
held and quasi fixed near the window. The tower is at very low elevation visible from

the hotel window. The Zeppelin is at a higher elevation.

CHAPTER 3. Mobile Satellite Channel 69

Figure 3.11: Surrounding and power plot at office building while windows are opened and
closed.

The last example shown in Figure 3.11 represents a typical office building (EN8,W;
OS1,T; FS1,F 65◦ ; AP1,S; DS2,V;). The windows are coated on the outside with a
silvered reflecting surface. The user terminal is tripod mounted. Tower terminal is

located at the multi-storey building and directly visible thought the open window. The
Zeppelin is at very high elevation. The path is diffracted at the top edge of the window
and leading to a signal loss of 20 dB. The window is closed from second 22 to 35. The
coated windows lead to a signal loss of about 30 dB.

Channel model parameters for the Lutz model have been derived for selected scenarios.

Results are given in Table 3.6.

3.2 EHF Band LMS Channel Characterisation by Mea-

surement

For LMS channel characterisation at 40 GHz, measurements have been performed for
future vehicular multimedia satellite services at EHF band [JH98]. Narrowband and

wideband measurements have been performed by simulation of the satellite link with a

suitable aircraft carrier (CESSNA 207), since satellites were not available with sufficient
link margins and transponder bandwidth. This approach allowed also to test the channel

properties for a wide range of environments in full azimuth and elevation ranges.
CHAPTER 3. Mobile Satellite Channel 70

Table 3.6: S band channel parameters

scenario elevation shadowing Rice fac- µdB / dB σdB / dB
factor A tor c / dB
rural UT from Zep. 15 0.99 23 -12.1 24.8
rural UT to Zep. 15 0.99 22.1 -10.6 20.5
rural UT from Zep. 25 0.36 11.1 -7.2 17.9
rural UT to Zep. 25 0.36 11.4 -7.1 18.9
rural UT from Zep. a 35 0.7 10.1 -9.9 5.9
rural UT to Zep. 35 0.7 12.9 -9.5 11.3
b
suburban UT from Zep. 45 0.60 13 -10.3 18.2
suburban UT from Zep. 45 0.60 11.8 -9.6 11.3
c
suburban UT from Zep. 55 0 17 - -
suburban UT to Zep. 55 0 19 - -
urban UT from Zep. 20 0.52 6.5 -6.4 11.1
urban UT to Zep. 20 0.52 10.4 -4.6 24.2
urban UT from Zep. 70 0.40 15.8 -5.4 0.8
urban UT to Zep. 70 0.40 17.9 -5.9 4.5
a
forced tree shadowing
b
forced head shadowing
c
UT antenna tripod mounted

3.2.1 EHF Band LMS Channel Measurement Setup

The measurement set-up basically consists of a transmitter in the aircraft and a receiver

part in a vehicle. The transmitter radiates a test signal, either wideband or narrowband,

and the receiver is probing the channel for the transmitted test signal. The carrier fre-
quency is 40.175 GHz. The receiver consists of two-stage demodulators transferring

the 40 GHz EHF-signal to a 1.8 GHz intermediate frequency band. Commercially off

the shelf channel sounders working in this band have been used to measure the received
signal. The dynamic range of the measurement system is better than 30 dB.

Figure 3.12 shows a block diagram of the receiver. Two independent receivers allowed

the simultaneous reception with the different antennas or different receiver types. Thus,
it was possible to measure narrowband and wideband signals, or narrowband signals

with an omni-directional antenna and a steered high gain antenna.

Figure 3.13 shows the different antennas used for the channel measurements. A high

gain antenna with 25 dBi two sided antenna gain was used and mounted together with
CHAPTER 3. Mobile Satellite Channel 71

Figure 3.12: LMS EHF band measurement equipment.

a video camera on a mechanical steering platform (cf. right hand side of Figure 3.13).
An omni-directional dipole was used as second antenna and as reference for picking up
the multipath effects (cf. left hand side of Figure 3.13).

video camera feed antenna

gain 25 dBi telemetryantennas steering platform

40 GHz receiver
steering platform demodulator compass fish eyevideo GPS antenna with omni-antenna

Figure 3.13: Roof top mounted antennas on van

CHAPTER 3. Mobile Satellite Channel 72

3.2.2 EHF Band LMS Channel Measurement Results and Parame-

ters

Narrowband power plots versus time in Figure 3.14 compare the performance of a high
gain antenna (25◦ beamwidth) with an omni-directional one in a provincial road envi-

ronment at an elevation angle of 25◦ . Little multipath fading can be observed besides
short shadowing events. It is obvious that the omni-directional antenna picks up more

multipath due to the lower discrimination of echoes, but for the same reason it is less

sensible to short blockage. Note that for the purpose of better presentation, the series

of the high gain antenna was offset by -20 dB. The Rice-factor in non-shadowed condi-

tions for the steered high gain antenna is 21.5 dB, whereas the omni-directional antenna
yields to an Rice-factor of 17 dB.

Figure 3.14: Narrowband channel series in EHF band; rural road; 25◦ elevation; upper curve:
hemispherical antenna, lower curve: steered antenna (offset by -20dB) [JH98].

The cumulative distribution function of the channels shows, in situations with direct
sight to the satellite, a moderate link margin (less than 4 dB) is sufficient for a service

availability of more than 95%. However, in shadowed environment the required link

margin would be as high as 30 dB to get a good availability. Such high margins are
CHAPTER 3. Mobile Satellite Channel 73

Table 3.7: EHF band channel parameters, rural road, standing car, steered antenna (note that
the increase of the shadowing with elevation is caused by the location of the receiver under trees
with foliage) [JH98]
elevation shadowing factor A Rice factor c / dB µdB / dB σdB / dB
25◦ 0.21 21.5 -15.4 2.2
◦
35 0.40 22.3 -17.1 4.4
◦
45 0.81 20.6 -25.3 7.2

Table 3.8: EHF band channel parameters, urban, standing car, steered antenna [JH98]
elevation shadowing factor A Rice factor c / dB µdB / dB σdB / dB
15◦ 0.71 13.2 -21.5 2.1
25◦ 0.17 16.8 -26.1 3.3
◦
35 0.24 15.7 -27.2 1.6
45◦ 0.08 14.3 -17.5 1.2

unrealistic for satellites from a cost point-of-view and show the distinct on/off channel
behaviour at higher frequencies.

24
22
20
Rice factor in dB

18
16
14
12
10
8
6
4 rural road
2 urban
urban
0
0 10 20 30 40 50
elevation in deg

Figure 3.15: Rice factor of the EHF band LMS channel (steered antenna) [JH98]

Figure 3.16: Rayleigh/Lognormal parameters of the EHF band LMS channel (steered antenna)
[JH98]
CHAPTER 3. Mobile Satellite Channel 74

The LMS channel parameters according to the Lutz model have been determined for

these EHF band channel measurements by least-square fits to the measured data. Figure
3.15 and 3.16 show a graphical representation of the Rice-factor and the parameters of

the lognormal distribution, respectively . Table 3.7 and 3.8 give the numerical values.

3.3 Dominant Characteristics of LMS Channels at

Higher Frequencies

One main result from the channel measurements at higher frequencies and narrow an-
tenna beamwidth is clearly that the shadowing factor A and the LoS factor 1-A become

the dominant parameter of the channel model. Within this section an analytical visi-
bility study is given and the results are compared to the shadowing factor derived from

channel measurement campaigns.

Even exhaustive channel measurement campaigns can only provide limited data sets for
few environments and satellite constellations to be characterised. The resulting chan-
nel models are restricted to certain environments, operational scenarios, frequencies,

antenna radiation patterns and satellite system dynamic parameters. In comparison to

channel measurements, shadowing and signal outage probabilities derived from ray trac-
ing calculations using urban 3D models, allow the investigation of larger areas, with a

variety of analysis and simulation possibilities.

Deterministic channel models derive propagation parameters from topographical and

morphological environment data. This technique is today well known for terrestrial

propagation modelling and essential part of mobile network planning tools. Since the
computation of all propagation paths would require a complete description of the geom-

etry along the mobile path, several approaches have been undertaken to overcome the

limitations of statistical and empirical channel modelling for land mobile satellite chan-
nels. For instance the project “LMS Propagation Model For Non-Urban Areas” (LMS-

PRO) an ESA funded study [DJKB98] was based on ray tracing techniques combined
CHAPTER 3. Mobile Satellite Channel 75

with statistical modelling. The relevant rays of the satellite to mobile communica-

tion path are determined based on topographical and morphological data. An intensive
search for digital elevation models, morphological data and wide areas of architectural

three dimensional models came to the conclusion, that – at this time – no database was

available to determine precisely the interactions with the earth surface in the proxim-
ity of the mobile. Therefore, a so-called ’Landscape Generation Model’ was included

to create such near range obstacles along the mobile trajectory using again appropriate
statistical processes for buildings and roadside trees.

Figure 3.17: Example of satellite visibility in urban areas (red: roof tops, green: streets, gray:
blockage, yellow: vegetation and road limitations included but not rendered).

Important work for generation of 3D building, roof and vegetation shapes has been
performed the in the late 90s (e.g. [HB99]). Accurate three dimensional data models for

urban and rural areas including structural engineering information as well as landscape

architecture and vegetation in public areas became accessible, for instance for the city
of Stuttgart in 2002. Such data were getting more and more available also for other

regions. Used by public land surveillance offices for the modelling and visualisation
CHAPTER 3. Mobile Satellite Channel 76

of cities, these databases are highly reliable in accuracy and resolution and include a

variety of additional topological information in public areas, ranging form site borders
to single kerbstones limiting streets, paths, parking areas and pavements.

Figure 3.17 depicts a detail of such a database for the city of Stuttgart. Based on a
ray tracing engine the picture is visualising in a top view the satellite visibility area

of one GEO satellite at 21◦ East and shows the rendered three dimensional models of

buildings. The bright (green) areas indicate visibility of the satellite for terminal on

ground. The dark areas show the shadowed areas of the satellite signal and medium

(red) ones represent the roofs of the buildings. The impact of vegetation can be rendered
from information about vegetation, mainly trees, in public areas (yellow). Additionally
to the data of vegetation also roadside boundaries (yellow) are overlayed as example in
Figure 3.17.

3.3.1 Outage Maps of Satellite Visibility

Such database is used for the generation of outage maps for satellite visibility over areas

of several square kilometres for various satellite constellations, leading to a precise eval-

uation of system performance especially at higher frequencies. Several parameter sets

can be easily changed in these simulations, e.g. receiver antenna height, vehicle loca-
tion, distribution and velocity on different road types. The type of shadowing (building

or vegetation) and possible diffraction areas can be identified as well as the analysis

areas can be restricted to roads or may include pavements and public parks. From the

shadowing obstacles the statistics of fade depth can be observed. For every system
performance evaluation the receiver distribution in the service area becomes important,

when the service availability has to be given in relation to the addressable market. E.g.

car receivers are the core market for DAB, thus reducing the outage maps to street lay-

outs is of key interest. The 3D models give excellent information about kerbstones,
roadside limits and thus real street widths. Furthermore, the overlay of the 3D model
CHAPTER 3. Mobile Satellite Channel 77

with additional street information allows to distinguish the number of lanes of any street

and to add a distribution of cars with respect to the street width.

The most dominant shadowing effect is obviously blockage by buildings. This shad-
owing decreases at higher elevation angles in general, but shadowing from vegetation

still remains under trees. This applies to the propagation of the direct signal. Multipath
reflection is negligible for unshadowed signals (lower than 25 dB with respect to the

direct path signal). Furthermore, in shadowed situations multipath especially at higher

frequencies is known to be insufficient for service provisioning since link margins are
typically far too low in satellite systems to compensate fades caused by shadowing. This

allows focusing on the pure LoS conditions for sake of simplicity. Furthermore a com-
bination of the measured fade depth for different obstacles and the analytically derived
shadowing cause can be made.

These outage maps are basis for a detailed system performance analysis incorporating
an assessment of the effectiveness of several shadowing countermeasures and even cost

estimation of those. Bases for the generation of the signal outage maps are three tiles of
1x1 km each including more than 75 km of streets from a 3D city model and digital city

map of different regions of Stuttgart.

• Stuttgart Mitte: inner city, with a mix of multi-storey offices, business and resi-

dential buildings and different types of city roads, streets are close to the build-
ings, public recreation areas and parks are without car access.

• Bad Cannstatt: typical suburban area with a mix of low storey single-and multi-

family houses, villas, row-houses mostly with garden and public areas like a soc-

cer and sports field.

• Neugereut: typical social residential area with high-storey or tower buildings and

some low storey single-and multi-family houses.

Three reference constellations, two GEO and one HEO, have been chosen as basis for
CHAPTER 3. Mobile Satellite Channel 78

the analysis of a system comparison. Table 3.9 summarises the elevation angle ε and

azimuth angle ϕ of the satellite(s) as seen from a terminal in Stuttgart.

Table 3.9: Satellite azimuth and elevation in Stuttgart for the chosen scenario
Constellation Name 1GEO 2GEO 3HEO
Elevation 31.7◦ Sat1 30.9◦ Sat2 30.9◦ > 70◦
Azimuth 165.8◦ Sat1 160.7◦ Sat2 199.3◦ 180◦

• 1 GEO: one GEO satellite located at 21◦ E; WorldSpace’s AfriStar is located at

this position and its footprint potentially covers Europe

• 2 GEO: two GEO satellites located at 5◦ W and 25◦ E; similar to the XM Radio
system (see section 2.1.1), with 30◦ orbital spacing difference of the satellites
which is yielding minimum correlation of the channels.

• 3 HEO: Three satellites in a classical Tundra like 24-hour orbit with about 55◦

inclination, ensure a minimum elevation of 70◦ throughout the whole service area;
similar to the constellation design of Sirius Satellite Radio (see section 2.1.1).

Figure 3.18 shows the signal outage maps due to manmade structures for the three satel-
lite constellations, from left to right in the case of 1GEO, 2GEO and 3HEO and top to

down for the reference environments of Stuttgart Mitte, Stuttgart Mitte in detail, Bad
Cannstatt and Neugereut. Again the bright (green) areas indicate visibility of the satel-

lite on ground. The dark areas show the satellite-shadowed areas. In the middle column

a differentiation between one shadowed satellite and no satellite visible can be made.

Building roofs are medium (red).

Table 3.10 summarises the corresponding statistical outage probabilities. The visibility

on ground and the percentage of built-up to ground areas are given, too. It can clearly be

seen that in the inner city the density of built-up areas is higher than in the two suburban
areas of Bad Cannstatt and Neugereut. Also the building types in all scenarios are quite

different.
CHAPTER 3. Mobile Satellite Channel 79

Figure 3.18: Shadowing maps for the three satellite constellations. All areas are of 1x1square
km. Rows from top to down: 1) Stuttgart Mitte, 2) Stuttgart Mitte Detail, 3) Bad Cannstatt, 4)
Neugereut. Columns from left to right: 1GEO, 2GEO, 3HEO.

In 3.19 a dark (blue) mask yields the reduction of receiver location probability to streets.

Table 3.11 shows the corresponding statistical outage probabilities. The visibility on
ground and the percentage of streets with respect to the whole area is given. Here a

doubled street density can be noticed for the inner city with respect to the suburban
CHAPTER 3. Mobile Satellite Channel 80

Table 3.10: LoS and shadowing factor at ground level for all non build up areas
Visibility on ground LoS to one (both) Shadowed in % Buildings in % of
(without building areas) satellite in % overall area
Stuttgart Mitte
GEO1 45 55 34
GEO2 65 (35) 35 34
HEO3 86 14 34
Cannstatt
GEO1 86 14 17
GEO2 90 (16) 10 17
HEO3 98 2 17
Neugereut
GEO1 80 20 16
GEO2 88 (18) 12 16
HEO3 96 4 16

areas.

Table 3.11: LoS and shadowing factor at ground level reduced to street area
Visibility reduced to LoS to one (both) No satellite visi- Reduction in %
streets at street height satellite in % ble in % of overall area
Stuttgart Mitte
GEO1 44.35 55.65 14.35
GEO2 68.11 (35.38) 31.89 14.35
HEO3 92,89 7.11 14.35
Cannstatt
GEO1 84.80 15.20 8.54
GEO2 93.26 (74.17) 6.74 8.54
HEO3 99.88 0.12 8.54
Neugereut
GEO1 81.17 18.83 6.04
GEO2 94.09 (71.80) 5.91 6.04
HEO3 99.54 0.46 6.04

3.3.2 Results and Comparison with Statistically Derived Parame-

ters

For further comparison, an installation height of the car antenna at 1.5 metre is con-

sidered in Table 3.12. With this the areas of LoS condition are by the order of a few
CHAPTER 3. Mobile Satellite Channel 81

Figure 3.19: Shadowing maps for the three satellite constellations. All areas are of 1x1square
km, masked with the street layout. Top to down: Stuttgart Mitte, Bad Cannstatt, Neugereut. Left
to right: 1GEO, 2GEO, 3HEO.

percent higher compared to visibility on ground and it is quite obvious that an increased

elevation lowers this influence.

A comparison of the analytical results with statistically derived parameters validates

channel measurements performed at Ku and EHF band. A typical shadowing factor A

of the Lutz model for vehicular receivers in an urban environment at 30◦ elevation (one
satellite) ranges from 0.4 to 0.6 (e.g. [ESK+ 02]) compared to the result of A = 0.53 in

Table 3.12.

It can be noticed that in the suburban areas the HEO constellation reached over 99

percent visibility in the streets. A further analysis in which way the visibility of the three
CHAPTER 3. Mobile Satellite Channel 82

Table 3.12: LoS and shadowing factor for receiver heights at 1.5 reduced to street area
Visibility reduced to streets, LoS to one (both) No satellite visi- ground to 1,5
receiver height 1.5 m satellite in % ble in % metre LoS ratio
Stuttgart Mitte
GEO1 46.83 53.17 1.0559
GEO2 71.84 (39.08) 28.16 1.0548
HEO3 96.01 3.99 1.0336
Cannstatt
GEO1 89.94 10.06 1.0606
GEO2 96.75 (81.81) 3.25 1.0373
HEO3 99.93 0.07 1.0009
Neugereut
GEO1 86.65 13.35 1.0675
GEO2 95.36 (77.49) 4.64 1.0135
HEO3 99.63 0.37 1.0009

different satellite constellations influences the system costs can be found in [HBWJ03].

3.3.3 Azimuth Diversity

Figure 3.20: Satellite LoS percentage at ground without building areas (blue line) and reduced
to streets only (pink line) in azimuth dependency for the tile Stuttgart Mitte at elevation 31.7◦ .

The LoS conditions in dependence of satellite azimuth are shown in Figure 3.20. The
model allows rotating the orientation of the city under the chosen satellite constellation.

The city of Stuttgart (tile Stuttgart Mitte) as seen in Figure 3.18a is rotated in 10◦ az-
CHAPTER 3. Mobile Satellite Channel 83

imuth steps under scenario 1GEO conditions. The elevation angle ε is fixed at 31.7◦

and the azimuth angle ϕ is rotated clockwise starting at azimuth 0◦ pointing to the top
in Figure 3.18a.

The variation of the shadowing factor A (in Figure 3.20 the LoS factor 1 − A is shown)
is between 0.38 and 0.54 for the whole area. The effect is even more visible if the
area of evaluation is again reduced to the streets. The shadowing minima can easily be

verified in Figure 3.18a by the obvious main direction of the roads. The LoS minima

show a dominant periodic occurrence of 90◦ , the LoS maxima a very dominant periodic

occurrence of 180◦ .

Thus, Figure 3.20 allows the conclusion of the azimuth correlation of LoS respectively

shadowing. As the main directions of the roads are orthogonal a satellite azimuth diver-
sity of about 45◦ is leading to an effective diversity gain.

3.4 Ka Band Aeronautical Satellite Channel Character-

isation by Measurements

Similar to the LMS channel characterisation, aeronautical channel characterisation fo-

cused for a long time on lower frequency bands. ITU-R recommendation P.682-1
[ITU92] deals in particular with the propagation effects encountered for earth-space

aeronautical mobile telecommunication systems. However, the methods contained in

that document are mainly applicable to frequencies below and near 1.5 GHz. The fol-

lowing sections will focus on the aeronautical channel at higher frequency bands.

3.4.1 Ka Band Aero Satellite Channel Measurement Setup

Classically VHF band is used for air-to-ground and S band for air-to-satellite links, thus,
multipath propagation for higher frequencies is not described in ITU-R recommendation

P.682 [ITU92].
CHAPTER 3. Mobile Satellite Channel 84

Aeronautical channel measurements at Ka band has been performed as a first step to

validate the technical feasibility of broadband aeronautical communication at higher

frequencies as shown in the project ABATE [HLL99]. The measurement set-up basi-

cally consisted of an airborne receiver for measuring a pilot tone at 18.685 GHz of the

Italsat satellite (cf. Figure 3.21).

Figure 3.21: Channel measurement setup for aeronautical Ka band measurements.

Low Noise Amplifier Downconverter Downconverter

20 / 2, 15 G H z 2 , 1 5 G H z / 7 0 MH z

0 dBm

70 MHz

20 GHz 2,15 GHz 0 dBm

ITALSAT
18,685 GHz 16,535 GHz 2,08 GHz
LIN-Vert.Pol
192
PLO 1 PLO 2 20
1 1

86,11979 GHz 104,0 MHz

SYNTHES. 2

10 MHz

SYNTHES. 1 Normal-
frequenz

Figure 3.22: Aeronautical Ka Band receiver.

Multimedia communication at higher frequencies typically uses high gain antennas. For

the campaign a Cassegrain antenna with a two-sided half power beamwidth 2ϑ3dB = 4.2◦
and a side lobe suppression of min. 22 dB was used as receiving antenna. An ex-

treme light weight structure yielded excellent pointing performance. The antenna was
CHAPTER 3. Mobile Satellite Channel 85

mounted on a steering platform which allowed tracking of the antenna with an azimuth

speed of 70◦ /s and elevation speed of 40◦ /s. The antenna pointing performance was
better 0.3◦ throughout all measurements. The antenna and steering platform is depicted

top right in Figure 4.12.

A CW signal was recorded in inphase and quadrature component on DAT recorders.

Time synchronisation of all data recorded with differential global positioning system

(D-GPS) and attitude and heading reference system (AHRS) data of the aircraft allowed

a precise post-processing of the measurement results. A digital elevation model of the

earth surface and a structural model of the aircraft allowed to determine special situa-
tions such as wing or tail shadowing. Figure 3.23 visualises a post-processed flight path
with the environmental surrounding.

Figure 3.23: Typical flight path of a measurement scenario with environment based on digital
elevation and morphological data.

Depending on geometrical relation of transmitter and receiver, multipath can occur by

reflections at the aircraft structure itself or at the environment. The aircraft motion and

the surface of the environment (sea state, land surface type) is influencing the occurrence
of multipath propagation.
CHAPTER 3. Mobile Satellite Channel 86

Trials with different flight scenarios have been performed, all in an area of southern

Bavaria, Germany. The flight conditions comprise: Start and landing, ascent and de-
scent, normal cruise, in flight manoeuvres, touch and go, and typical holding patterns.

It was possible to perform manoeuvre to force geometrical relation for multipath of

wings and ground reflection, as well and shadowing by the aircraft structure. For exam-
ple, flight manoeuvres with up to 50◦ banked circles or ascent and decent with nose up

(15◦ pitch) and nose down (-15◦ pitch) were performed. However, flight situations over
large water areas were not possible.

3.4.2 Ka Band Aero Satellite Channel Measurement Results and

Parameters

Figure 3.24 shows two sample results for a LoS situation and a wing shadowing situation
at clear sky weather condition. A Rice factor c > 30 dB can be derived. Shadowing of

the metallic aircraft structure leads to a blockage of the received power, noticeable by
an attenuation up to range limit of the measurement setup.

Figure 3.24: Receive power for LoS (left) and wing shadowing (right).

No multipath effects have been observed, even though extreme flight manoeuvres have

been forced. As expected the very narrow beamwidth of the used high gain antenna

suppressed any overlaying propagation paths.

CHAPTER 3. Mobile Satellite Channel 87

Typical antenna beamwidth for mobile multimedia communication with aircraft at high

frequencies are in the range of 5◦ to 10◦ . Thus, even at continuous low elevation angles
the probability of receiving a second path is negligible.

The trials also comprised different weather conditions such as rainy, cloudy, sunny,

flights below, through and above clouds. The measured receive power showed an atten-
uation of less than 1.5 dB for flights under rain clouds.

3.5 Dominant Characteristics of Aeronautical Satellite

Channels at Higher Frequencies

3.5.1 Aircraft Structure Shadowing

As pointed out in the previous section the aeronautical channel at higher frequencies is
a fairly steady channel, which suffers almost only from shadowing of the aircraft struc-

ture. Antenna location on the aircraft and the antenna beam agility range is therefore of
major importance offering great optimisation potential.

The antenna beam agility range is typically depending on the antenna type, i.e. beam

steering method. Installation options are dominated by structural design issues, as in-

tegration of high gain antennas impacts the mechanical structure of the aircraft due to
their aperture - and thus overall - size.

In contrast to electronically steered antennas, mechanically steered antennas do not lose

antenna gain when the signal incident angle differs from antenna boresight (cf. 4.3.2).

Installed at an optimised location these antennas types can provide constant gain perfor-
mance over a signal incident angular range of over 180 degrees.

Table 3.13 summarises the elevation of the tail structure and horizontal tail plane seen
for an antenna installed at the aircraft front, middle, and behind the wings. Especially

the smaller aircraft types use high wing airplane models. Besides the position at the
CHAPTER 3. Mobile Satellite Channel 88

top of the tail structure, a position near the cockpit would obviously be best suited to

minimise shadowing probability by the tail.

Table 3.13: Shadowing angles of tail structure in degree.

Airplane max. TP max. TS max. TP max. TS max. TP max. TS
a b
elev. elev. elev. elev. elev. elev.
Airbus front middle behind wings
A340-200 <0 10 <0 15 <0 30
A340-300 <0 9 <0 14 <0 26
A320-200 <0 11 <0 18 <0 29
A321-100 <0 12 <0 15 <0 25
A310 <0 13 <0 20 <0 35
Boeing front middle behind wings
747-400 <0 9 <0 15 <0 30
737-300 <0 12 <0 20 <0 30
737-400 <0 12 <0 19 <0 30
737-500 <0 13 <0 20 <0 30
Dornier front middle behind wings
328 14 15 20 22 30 33
Fairchild front middle behind wings
Metro 23 1 10 2 14 3 19
Cessna front middle behind wings
CitationJet 12 14 18 19 28 30
Citation S/II 4 15 7 21 8 34
Citation VI 15 15 20 20 36 36
Citation V 4 15 7 21 8 34
a
tail-plane
b
tail-structure

An aircraft can change its attitude in a wide range. Roll angles up to 25 degrees and

pitch angles of -5 to 15 degrees occur during normal flight manoeuvres. This means

that for example the virtual horizon of the aircraft is lowered (or increased) by about 25
degrees while flying a normal autopilot manoeuvre.

Today’s commercial systems for aircraft communication are based on GEO satellites.
The main critical issues in a GEO system are the coverage deficiencies at higher latitudes

and the extreme antenna steering requirements at lowest elevation angles (i.e. again

highest latitudes), and this in the light of the commercially important near-polar flight

routes in the northern hemisphere.

CHAPTER 3. Mobile Satellite Channel 89

The minimum elevation angle εmin is defined in equation 2.7 as a parameter of earth-

space geometry and related to terminal location on earth only. In the aeronautical com-
munications scenario, however, it is important to consider relative elevation angles with

respect to the virtual horizon of the attitude-changing aircraft, which is the plane defined

by the roll and pitch axes of the aircraft body, cf. Figure 3.25.

±70˚

N
min
. ele
vati
25 on
˚r ant
enn
ol a =
la 20˚
ng
le
vir
tua
lh
ori
zo
n
ho
ri z
o

total min. elevation = 45˚

n

Figure 3.25: Illustration of relevant angles concerning aircraft-satellite.

For further considerations, let us assume a phased-array antenna mounted on top of

the aircraft’s fuselage in the front region, with a rotational-symmetric (around the yaw

axis) radiation pattern and a 3dB-steering range of, say, ±70◦ . This yields a minimum
relative elevation of +20◦ . In order to derive the effective (nominal) minimum elevation

angle (from horizon) for satellite coverage under permanent line-of-sight conditions, in

general a “worst-case overlay” of both body structure elevation angles and attitude has
to be considered for all flight phases where service shall be available.

With the assumed antenna position, however, any body structure blockage is outside
the steering range of the antenna. Moreover, during the cruising flight phase for in-

stance, pitch values are usually negligible with respect to standard roll angles of 25◦

during manoeuvre. Thus the worst-case situation arises when the aircraft is at the edge
of (nominal) coverage, its flight vector is orthogonal to the line between aircraft and

satellite, and a manoeuvre requires rolling to the opposite side from satellite direction.
CHAPTER 3. Mobile Satellite Channel 90

Then the minimum relative elevation angle of 20◦ and the maximum roll angle of 25◦

add positively up to a 45◦ requirement for the nominal minimum elevation angle of the
system in order to guarantee permanent visibility. A GEO system solution faces severe

coverage problems at higher latitudes, especially with demanding availability figures

under extreme conditions; with a 45◦ minimum elevation requirement, it can cover lo-
cations up to approximately 40◦ latitude, only.

3.5.2 Cloud and Rain Attenuation

As especially higher frequencies are of interest for multimedia communication to col-

lective terminals, the tropospheric and environmental effects will be detailed in the next

section with special focus on the aeronautical channel at frequencies above 10 GHz.
Ionospheric effects are of minor impact at frequencies above 10 GHz according to
[ITU03a].

Particularly for tropospheric effects, the flight height has significant impact on the aero-
nautical satellite channel. Thus, a classification into flight phases is necessary for

channel characterisation. Table 3.14 and Figure 3.26 show seven phases of a typi-

cal flight profile, based on the standard flight phases defined by Aeronautical Radio
Incorporated (ARINC) 429 specification [ARI04] and measured data of typical flight
profiles [DFS06].

Table 3.14: Aircraft Flight Phases

TE #, Code Description Comment
AF1, H Taxi-out On ground
AF2, E Take-off First 5 to 10 min of flight
AF3, N Climb With manoeuvre, might be still under clouds
AF4, U Cruise Normal flight with standard manoeuvre at cruise height
AF5, T Descent Holding patterns, might be under clouds
AF6, T Landing 5 to 10 min to touch down
AF7, T Taxi-in On ground

Depending on the type of application, there might be no need for a satellite connection
CHAPTER 3. Mobile Satellite Channel 91

Figure 3.26: Seven phases of a typical flight profile.

during taxi-out and taxi-in, either because of safety regulations, regulatory issues, or an
available terrestrial connection alternative.

Especially for analytical considerations of passenger communication, the phases take-

off and climb respectively descent and landing can be summarised to one single phase,
which will be named "climb" and "descent". During climb phase, as well as during

descent phase, the aircraft possibly passes through cloud layers. To assess the time

interval during which data reception/transmission is possibly influenced by clouds above

the aircraft, the relevant parameters are the duration of take-off/climb phase Ta , duration

of descent/landing phase Td , and, of course, the duration of the cruise phase Tc , together
with the cruise altitude hc .

During climb and descent, the vertical flight profile depends mainly on safety regula-
tions and noise protection in the vicinity of airports, but also on the particular aircraft

model. Further, the vertical flight profile followed by an aircraft depends on whether

a short, medium, or long haul flight is considered, e.g. with respect to cruise altitude.

Justified by comparison of flight profiles of long-, medium-, and local/short-haul flight

profiles, the following can be stated:
CHAPTER 3. Mobile Satellite Channel 92

• the time for climb and descent amounts to approx. Ta = Td = 0.5 hour each in

all cases,

• the cruise altitude typically amounts to 10000 m to 12000 m for medium- and
long-haul flights (the cruise altitude inside Europe is about 8000 m to 12000 m),

• the cruise altitude for local/short haul flights less than 1 hour may be below 10000

m, as the final cruise are not reached.

With the promotion of Ku and Ka band usage for aeronautical satellite communication,

the newer revisions of the ITU-R recommendation P.682 [ITU92] also refer to recom-
mendation P.618 [ITU03a] with respect to cloud and rain attenuation as described in
section 2.4.1. The methods presented therein are valid for a broader frequency range
and also hold for the aeronautical case.

In particular for systems above approximately 18 GHz and especially for low elevation

angles, several sources of tropospheric attenuation must be considered, which are (cf.
[ITU03a]:

• attenuation due to rain,

• attenuation due to clouds and fog,

• gaseous attenuation due to water vapour and oxygen, and

• attenuation due to tropospheric scintillation.

It can be assumed that attenuation due to clouds is comparably small, except for clouds

of high water content. Rain attenuation can have a significant impact on the link quality,

in particular at higher frequencies such as in Ka band. Furthermore, although clouds

can occur up to high altitudes, the probability of clouds quickly drops with increasing
altitude. This holds in particular for clouds with considerable water content, as data

available from satellite measurements suggests [ITU99b].

CHAPTER 3. Mobile Satellite Channel 93

Attenuation due to atmospheric and tropospheric gases becomes already important at

frequencies above 10 GHz [ITU03a], but these effects will also diminish for aircraft
at cruise flight altitudes. ITU-R recommendation P.618 is using a model of mean rain

height which varies with longitude and latitude, as discussed in ITU-R recommendation

P.839-3 [ITU99a].

Figure 3.27: Global mean rain height hR above mean sea level (based on data available on
[ITU05]).

Figure 3.27 shows the global mean rain height based on data available on [ITU05].

Especially, with respect to the main flight routes in the northern hemisphere, it can
easily be verified that the mean rain heights are less than the cruise heights.

The rain attenuation is zero for receivers at an altitude higher than the respective rain

height hR , therefore the times during which an aircraft is typically at an altitude higher

than the rain height is assessed. As shown in Figure 3.27 the mean rain height is world-

wide well below the cruise altitude of long- and medium-haul flights, and also for typical
short-haul flight this can be expected (the maximum of approximately 6 km is reached

in Tibetan Highlands). Considering the climb and descent phase, it can be expected that
the total time TR between take-off and touch-down during which the satellite link could

be affected by rain attenuation, amounts to approx. (values of Ta = Td = 30 min, and hc

= 12000m according to previous section)

CHAPTER 3. Mobile Satellite Channel 94

Ta + Td 3600sec
TR = hR = hR . (3.1)
hc 12000m

Inserting hR = 4000 m as the mean cloud height being the maximum in most parts of
the world (e.g. Europe, cf. Figure 3.27) and hR = 6000 m as the worst-case yields TR =

20 min and TR = 30 min, respectively.

Figure 3.28: Rain rate distribution exceeded for p=0.001% for world-wide coverage (based on
[ITU03b].

Within this period the actual rain probabilities and rain rate probabilities of the specific

region has to be applied. A spatial characterisation of rain fall rates expended for a

probability p in % of an average year can be found in ITU-R recommendation [ITU03b].

The rainfall rate on world-wide coverage for p=0.001% is shown in Figure 3.28.

As shown above the attenuation effects of rain and clouds on the aeronautical channel

at frequencies above 10 GHz are of negligible impact, especially during cruise phase.
CHAPTER 3. Mobile Satellite Channel 95

3.6 Conclusions

Channel measurements at S-Band have been performed at a very wide range of scenar-

ios. Power series including a diversity path have been input and further processed for
ICO system design. Scenarios of major interest show up to 20 dB attenuation by rural
tree shadowing with very high probability for diversity gain and approximately 30 dB

attenuation for indoor scenarios with no probability for diversity gain. The ICO system

design laid special focus on successful call signalling for indoor UT locations. Once
signalling reached the UT the user may move to a window location during the call.

Channel measurements at higher frequency have been performed for the 40 GHz LMS

channel and 20 GHz aeronautical channel. Parameters of the Lutz model have been
derived and showed that the shadowing factor or satellite visibility becomes the pre-
dominant channel parameter. Very little to no multipath propagation can be observed.
The result of the distinct on/off behaviour of the channel provides input to the further

mobile satellite system design process.

With the knowledge that the shadowing factor is the predominant channel parameter the
huge effort for channel measurement campaigns e.g. approximately 4 man years plus
flight time and equipment cost for the S band ICO measurements becomes somewhat

disputable.

The thesis discusses a LMS visibility study which covers three different satellite con-

stellations in parallel at a much bigger area i.e. 75 km of roads which is more than
any cannel measurement campaign performed within this thesis. A comparison of the

results of the visibility study prove the shadowing factor which has been derived by
channel measurements.

The visibility study compares the LoS factor of three satellite constellations (one GEO

satellite, two GEO satellites, three HEO satellites) and the LoS factor at ground exclud-
ing non build up areas, reduced to streets, and reduced to streets at a receiver height of

1.5 metre. The results show that the three HEO constellation is able to cover 96.01% of
CHAPTER 3. Mobile Satellite Channel 96

the urban area and 99.93% of the suburban area with visibility. In comparison the two

GEO system visibility is only 71.84% for urban area and 96.75% for suburban area.

Shadowing for the aeronautical channel is depending mainly on the antenna mounting

location at the aircraft. Mounting at the front of the aircraft or at the top of the tail
structure minimises aircraft structure shadowing, but in most cases aircraft structural

design dictate the mounting location. Especially for flight paths at high latitudes an

aircraft standard roll angle during flight manoeuvre restricts additionally the antenna to

satellite visibility.
Chapter 4

Satellite Terminal Antenna Pointing,

Tracking and Acquisition

Satellite Spectrum and Term. Wireless Channel

Satellite Radio Link
Constellation Antenna Regulation Characteristics
Chapter 2

Mobile Satellite
Channel
Chapter 3

Mobile Platform
Mobile Terminal Design
Characteristics

Chapter 4
Mobile Group Capacity and
Mobile Services
Characteristics Handover Analysis
Chapter 5

Enabling broadband communication via satellite at higher frequencies requires high

gain antennas at the satellite and on the ground terminals. If mobile reception shall be

supported, the movement of the mobile terminal must be compensated by continuous

steering of the antenna towards the satellite to achieve the maximum antenna gain.

High gain terminal antennas, which could be implemented in an unobtrusive way on a

mobile platform – in order to comply with wind drag, power consumption, and aesthet-

ical constraints – have been within the research interest since a long time. Satellite op-

erators and industries are still searching for economical antenna solutions, which would
open the market for a broad field of applications, but an optimal technical solution at

reasonable price seems not being achieved yet.

97
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 98

The use of higher frequencies lowers the required antenna size, but still antennas need

to be steered towards the satellite. Thus, also antenna PAT algorithms and the sensor
hardware (HW) are part of an integrated economic approach to mobile high gain termi-

nals.

An optimised antenna design and the associated PAT algorithm would provide the min-

imum required performance, particularly necessary for a given mobile satellite scenario

at lowest implementation and hardware (HW) effort. Especially, precise sensors for

PAT might pose a challenge for an economical solution. Therefore, a tradeoff between

required antenna gain and sensor accuracy has to be found.

4.1 PAT Basics and Definitions

4.1.1 PAT Problem

The general PAT problem is the maximisation of the antenna gain, by sufficiently accu-

rate matching of the antenna boresight vector with the vector from the mobile antenna to
the satellite. Figure 4.1 depicts the general PAT problem. A mismatch of the two vectors

translates into a loss of signal power and an increase of other system interference.

Satellite direction

Dh

Antenna look angle

loss
actual antenna gain max. antenna gain

Figure 4.1: Satellite mobile geometry.

CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 99

In this thesis, only mobile to satellite pointing vector for the PAT problem, which is

dynamically influenced, is considered. The relative movement of satellite and terminal

with respect to earth is affecting the PAT. For GEO satellite communication the terminal

mobility – and especially the variance of its attitude – is the dominating influence. For

non-GEO constellations or inclined GSOs, terminal antenna beam agility requirements

resulting from both satellite movements and terminal mobility have to be considered.

4.1.2 Definitions

In order to establish a communication link with high gain antennas, the PAT process
steers the antennas of the communication partners towards each other. It is obviously

necessary that the transmitted signal of the sender reaches the detector of the receive
terminal with sufficient power and vice versa. PAT is a challenge for narrow beamwidths
due to the high angular resolution, and laser communications has dealt with this problem
for a long time. The process of pointing, acquisition and tracking is defined in the

following [GK76]:

• Pointing is the operation of aiming a transmitter in the proper direction of the

receiver, whereas

• (spatial) acquisition is the receiver operation of determining the direction of ar-

rival of an impinging beam, and finally

• tracking is the subsequent operation of maintaining the pointing and acquisition

throughout the communication time period.

PAT is a process with several stages. First, the antenna of the transmitter has to be

pointed towards the receiver. The mobile is assumed to be within the coverage area
of a satellite. The satellite antenna is pointed to the terminal and can typically not be

steered. Setting up a communication link is in this case only an issue for the mobile
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 100

terminal. The footprints of the satellite are considered fixed or moving, depending on

the constellation, but not steered towards one single user terminal. This would be a step
towards SDMA and is not dealt hereafter.

Then the mobile terminal has to acquire the satellite and optimise the receive signal.

This might be supported by some pre-knowledge (e.g. last elevation angle) of the lo-
cation of the satellite. If the angular uncertainty area of the satellite position is greater

than the terminal receive antenna beamwidth, the terminal needs to search within this

uncertainty area.

After acquisition, the tracking process follows, maintaining the antenna direction to-

wards the satellite throughout the communication period.

For bidirectional communication the mobile receive and transmit antennas are usually
located at close spatial distance compared to the antenna beamwidth. Consequently, the
same tracking angle for the mobile receiver and transmitter is resulting.

4.1.3 PAT Design Methodology

The design methodology for an optimal PAT system is depicted in Figure 4.2. The
system design involves:

• the antenna parameters and regulatory standards;

The required data rates and related link budgets impose the directivity of the ter-
minal, ranging from low quasi omni-directional to high gain directive antennas.

Based on the antenna beamwidth and possibly a polarisation matching, the re-
quired PAT system accuracy and resolution of beam movement can be defined.

• the movement of the satellite and the mobile;

Knowledge of the translatory and angular movement of satellite and mobile ter-
minal is leading to determination of angular range, rates and acceleration require-

ments of the mobile terminal antenna and the velocity of the PAT algorithm.
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 101

Input: Sensors
Input: System
Input: Antenna - available attitude / Input: Channel
- operational scenario
- antenna parameters position information - shadowing factor
and area
- regulatory standards - available signal - fade duration
- satellite constellation
feedback

Angular range, rate

PAT accuracy and acceleration Additional sensors Reacquisition time
Velocity of PAT
algorithm

Cost and performanceoptimisedPAT system

Figure 4.2: PAT Methodology.

• the availability of mobile position and attitude parameters;

If sensors for position and/or attitude information and/or a receive signal strength
feedback are accessible, this information can be exploited by the PAT system and
reduces the effort for implementing additional HW. However, also a PAT system

design which is independent from the mobile platform might be a requirement.

• the channel characteristics.

In case the chosen PAT system is relying on a signal strength feedback or informa-

tion gathered from the decoded signal, the channel characteristics like occurrence

of shadowing and fading influences the acquisition and re-acquisition times of the

mobile terminal antenna.

The PAT design process starts with the evaluation of the required PAT performance for
a given communication system and mobile environment. The performance criteria of a

PAT system are:

• the angular accuracy in azimuth, elevation and possibly polarisation, and

• the max. angular range, rate, and acceleration that need to be supported by the

PAT system.
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 102

Based on this knowledge an optimised PAT system can be derived.

Low costs are particularly important for products targeting the mass market. Conse-

quently, a cost optimised PAT design is identifying the system which is providing the
required performance, particularly necessary for a given mobile satellite scenario at

lowest implementation and HW effort.

Different approaches for determination of minimal PAT performance requirements and

cost-efficient PAT systems are detailed in the remainder of this chapter.

4.2 Beam Agility Requirements of the Terminal An-

tenna

In this section the dynamics of the pointing vector will be determined which translates
directly into the agility requirements of the terminal antenna beam. The pointing vec-
tor is impacted by movements of the satellite and the mobile which will be separately
addressed.

4.2.1 Satellite Dynamics

4.2.1.1 Dynamic Location of non-GEO Satellites

Communication with satellites at non-GEO orbits require terminal antenna agility in

order to follow satellite movement. Angular range, rate and acceleration requirements
for the antenna can be calculated from the geometric relations between the satellite orbit

and the terminal as introduced in section 2.1.2.

At the terminal antenna, worst case elevation and azimuth values and their time deriva-

tives have to be taken into account, which occur at overhead pass of a satellite.

Figure 2.4 shows the maximum angular speeds for a LEO and MEO overpass. The next

sections demonstrate that the rates are far below the angular speed encountered for the
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 103

attitude changes of a mobile platform. However, this does not hold for the singularity

of an overpass where the discontinuity in the zenith (cf. Figure 2.4) leads to an infinite
azimuth speed for typical mechanical antennas.

4.2.1.2 Satellite Attitude Changes and Orbit Perturbations

Changes of the satellite attitude due to effects like vibrations cannot be counteracted by

a PAT process at the mobile antenna. Typically, they can be neglected with respect to the
satellite antenna beamwidth, as most communication satellites are three axes stabilised.

Satellite orbit perturbations lead to a movement with respect to the nominal satellite
position. These movements are typically small and slow. Their impact on angular range,
rate and acceleration requirements for the mobile’s antenna can be typically neglected
as well, if the satellite is under operational window keeping control [MB02].

End of life satellites can be attractive for mobile applications as angular changes due to
orbit perturbations could be compensated by the mobile terminal antenna.

The satellite ASTRA 1A is taken as an example. It has been launched 1988, is now al-

ready in an inclined orbit. Its nominal orbital position is 5.2◦ E longitude, corresponding

to the longitude of its ascending node. The orbit inclination is evolving with time by
about 1◦ per year and will possibly reach a maximum of 5◦ at end of life.

4.2.2 Agility of Mobile Terminals

Movement characteristics of the mobile terminal are defined by:

• translatory movements (position changes, i.e. longitude and latitude) inside the
satellite’s service area.

• attitude changes (rotations around the three body axes, i.e. yaw/heading, pitch,

and roll),
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 104

Small vehicles are typically highly agile and pose more stringent requirements on the

PAT system than larger vehicles. This trivial fact is a potential serious problem for
mass market products. The effort and cost to guarantee a certain pointing accuracy

strongly depend on the particular mobile scenario (i.e. the faster the movement the

more expensive is the PAT solution). Considering different scenarios (e.g. car, yacht,
train, aircraft) the cost of the PAT system alone can vary in a wide range and in fact,

effort and cost of implementing the PAT system can prevail the pure antenna cost and,
thus, be the limiting factor.

4.2.2.1 Translatory Movements

Translatory movement is usually defined in the earth centered, earth fixed (ECEF) ref-

erence frame using longitude, latitude, and height.

Equations 2.12 and 2.7 define the azimuth and elevation angle of the mobile terminal

antenna, respectively, depending on the mobile’s and satellite’s position in longitude

and latitude.

It is obvious that the angular range for azimuth is 360◦ . The elevation range of the

terminal antenna which is communication with a GEO satellite1 , is mainly influenced

by the operational area of the mobile within the satellite system coverage. Aircraft and

ships are mobile in large areas and thus require steering range of 0◦ to 90◦ or respectively
0◦ to 180◦ in the elevation axis. Land mobile vehicles in contrast are typically bound

to a certain operational area, thus, the elevation angular range of the terminal antenna

might be restricted to a defied market e.g. CONUS or Europe.

Figure 4.3 shows the elevation angles encountered for a geostationary satellite posi-
tioned at 5.2◦ E. An elevation range of 25◦ to 45◦ is resulting for mid-latitude Europe.
1
For non-GEO systems the required elevational angular range of the terminal antenna is dominated
by the satellite dynamics described above.
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 105

Figure 4.3: Elevation angle (in ◦ ) depending on the longitude and latitude of a mobile terminal
for a GEO satellite at 5.2◦ E longitude.

4.2.2.2 Attitude Impact

The mobile platform attitude impacts the antenna pointing system agility requirements

in terms of rate and acceleration. This effect dominates over the translatory movements

of a mobile.

Attitude changes are characterised by the three angles heading/yaw, pitch, and roll.

These angles are defined within a local reference frame, given by the local horizon and

the East or North vector (east north up (ENU)2 , or north east down (NED) frame). The
ENU frame is most commonly used, while the NED frame is usually applied in the

aeronautical scenario.

For both frames the heading denotes the rotation angle around the vertical axis with

respect to the true-North vector (or equivalently the East vector). Yaw denotes the rota-
tion around the vertical axis with respect to the mobile’s local coordinate system, pitch
2
ENU, i.e. x-axis points towards East (≡ 0◦ ), y-axis points towards North, and z-axis towards zenith
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 106

denotes the rotation around the lateral axis, and roll around the longitudinal axis (cf.

Figure 4.4).

Figure 4.4: Mobile Attitude Definition.

The roll, pitch and yaw angles can be related to azimuth and elevation angles that define
the direction to which the antenna must point. The antenna elevation range is deter-
mined, on the one hand, by position changes on a large scale and, on the other hand,

the vehicle attitude changes. To evaluate the maximum elevation range due to large
scale position changes, the lower and upper limit is determined by the service area of
the satellite or the operational area of the mobile, respectively. The actual range must

be increased by the attitude changes of the mobile.

The elevation range for non-GEO constellations may prevail these restrictions, while
rates and accelerations are dominated by the attitude changes of the vehicle.

4.2.2.3 Vehicular Platforms

Attitude changes for land based vehicles originate from the changing properties of the

route which a vehicle is taking (e.g. road or track slope, road or track bank angle) and
from movements caused by the inertia of the vehicle mass (e.g. roll during a u-turn,

pitch while accelerating a car). In order to gather statistics of yaw, roll, and pitch values
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 107

both effects can either be analysed or measured during a test campaign. Analysis would

require knowledge about car inertia dynamics, which is typically not disclosed by car
manufactures.

Thus, several measurement campaigns including worst case scenarios (continuous cir-
cle, speed bumps, ...) with different vehicle types (car and van) and have been per-

formed [HLO00]. Table 4.1 shows the worst-case angular range, rate, and acceleration

for a driving car.

Table 4.1: Worst-case angle, angular rate and acceleration for yaw, pitch, and roll for cars
Parameter Angle in ◦ Rate in ◦ /s Acceleration in ◦ /s2
Heading/Yaw 360 40 50
Pitch ±5 20 100
Roll ±5 10 100

During normal cruise the values are typically well below the worst-case values and the
measurements have shown that the typical angular range for pitch and roll is smaller,

approximately 3◦ for roll, and 2◦ for pitch. However, these values apply for a car on
an almost horizontal plane. Considering hilly terrain in particular the pitch angle may

exceed 5◦ , depending on the road slope which can amount to 10% (≡ 5.7◦ ) and more
in the extreme. Figure 4.5 shows the complementary cumulative distribution function

(CCDF) of the elevation angle for two different test drives in a rural hilly environment
(route 1 and 2 cf. Figure 4.6). The elevation angular difference to nominal is derived

according to the co-coordinate transformation described in Annex C.

4.2.2.4 Train

The train movements can be divided into short term variations (e.g. due to vibrations)

and long term variations, caused by the movement of the train. Long term changes are

caused by i) position changes of the train, ii) heading, roll and pitch changes along a

curved, sloped (orthogonal to direction of the train movement) or inclined (in direction
of the train movement) track and iii) yaw, roll and pitch changes due to the chassis
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 108

Empirical CCDF
0
10
Route 1
Route 2

P(elev. error exceeds abscissa)

−1
10

−2
10

−3
10

−4
10

−5
10
0 2 4 6 8
elevation errror w.r.t. nominal in deg

Figure 4.5: Complementary CDF of elevation angle error with respect to nominal (route1 and 2
as depicted in Figure 4.6).

Figure 4.6: Route examples for empirical attitude measurements of cars.

movement of the train. The layout of a train track is limited to a minimum radius of

curves and a maximum angle in track slope and inclination/declination (uphill/downhill)

for to passenger comfort. Table 4.2 presents measurement values for a typical high speed
train.
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 109

Table 4.2: Worst-case angle, angular rate and acceleration for yaw, pitch, and roll for trains
Parameter Angle in ◦ Rate in ◦ /s Acceleration in ◦ /s2
Heading/Yaw 360 0.18 3.4
Pitch ±0.06 0.3 4.6
Roll ±6.9 4.9 22.9

The measurement were performed on a non inclined track, thus the values of pitch,
pitch rate and pitch acceleration reflect attitude changes due to chassis movement of the

train. Assuming furthermore that a train chassis is a relatively stiff vehicle in travelling

direction and that no long term pitch nutations are occurring, the measured values are

can safely be considered as vibrations. The values are relatively small with respect to
long term pitch changes.

The low heading acceleration value of 3.4◦ /s2 can be interpreted such that the maximum
heading rate is resulting from track layout restrictions due to passenger comfort.

The occurrence of angular roll changes in a train scenario can be caused by different

effects. Slope angles of the track are built in curved tracks to compensate centrifugal
forces. Modern trains are able to control actively the angle between the carriage and the
wheels (chassis angle). In this way they compensate centrifugal forces for the conve-

nience of the passengers. This effect increases or decreases the overall roll angle. The

measured roll values are the sum of slope angle and chassis angle variations. Angular
rate and acceleration are worst for the roll angle with respect to pitch and yaw in the

train scenario, but still less than in other mobile scenarios.

4.2.2.5 Maritime

In the maritime scenario the antenna agility requirements strongly depend on the par-

ticular ship type. Each ship design (e.g. size, height of centre of gravity above water

level) is subject to a particular movement behaviour. For larger ship types, the influ-
ence of roll is dominant, whereas smaller ships show pronounced pitch movements,

too. Furthermore, cargo ships may show a completely different movement behaviour
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 110

(also depending on the load) than passenger ferries of similar size, because the latter

are usually actively compensating the roll and pitch angles by hydraulic stabilisers or
water tanks. It is obvious that smaller ships pose higher requirements on the antenna

and PAT system than larger ships. Worst case weather conditions may pose additional

requirements on the antenna agility.

For an emergency communication system, the worst weather conditions represent the

benchmark for the terminal, requiring a high system availability, while for a SAT-TV

application a lower availability may be sufficient, allowing a much cheaper solution.

Depending on the location of the antenna aboard the ship and the type of ship, high

inertial forces may act on the antenna structure in bad weather conditions due the high

lateral accelerations. The resulting mechanical stress can result in a pointing error.
Also vibrations caused by the ship’s motor or by waves may pose high demands on the

electronics and the mechanics of the antenna and PAT system.

Attitude results are derived from measurements for a 10 m sailing yacht and 150 m
passenger ferry. The two boat types define the possible range for different agility re-

quirements for a mobile satellite platform in the maritime environment.

Based on 195 minutes of data recorded with different weather conditions [BH00],

[Hoc00], the requirements for the antenna and PAT system for small ships are extracted.

The accuracy of the employed attitude and heading reference system (AHRS) is ±0.1◦ /s

for roll, pitch and yaw rates. The data is categorised into various classes according to
the maximum and minimum wind speed that occurred during the measurements to gain

some knowledge about the relation between weather conditions and occurring angular

rates and accelerations (cf. Table 4.3).

The resulting attitude angle, rates and accelerations for a 10 m sailing yacht are shown

in Table 4.4. The roll angle appears high for the sailing yacht compared to other types

of small ships. For motor yachts commercial products support ranges of ±25◦ for roll
and ±15◦ for pitch.
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 111

Table 4.3: Definition of different maritime environment categories based on occurring wind
speeds.
Category Mean wind speed
1 6 m/s
2 7.7 m/s
3 10.1 m/s
4 11.6 m/s

Table 4.4: Worst case angle, angular rates, and accelerations for 10m sailing yacht
Parameter Category Angle in ◦ Rate in ◦ /s Accel. in ◦ /s2
Roll 1 ±32.5 7.8 20
2 ±44.9 12.4 43.4
3 ±44.7 18.6 47.8
4 ±55.5 22.1 38.9
Pitch 1 ±4 8.9 62.7
2 ±4.9 11.4 64.7
3 ±8.2 18.5 90
4 ±10 20 91.4
Yaw 1 360 6.5 30.1
2 360 7.9 45
3 360 21.2 104.2
4 360 18.7 102.8

Measurements for a 150 m ferry during a measurement campaign for the project MO-
BILTY in the region of the Canary island have been perfomed. The characteristics of the
AHRS was: maximum dynamic angular error for roll, pitch and yaw is 0.1◦ , max. rate is

500◦ /s, sampling and output rate is 200Hz. The ship has a total length of 151m, width

26m, max. speed 18knots. The capacity of the ferry is 556 passengers and 170 vehicles.

The AHRS was installed on the upper most deck, approx. 35m above sea level. In total,

approx. 65h of recorded data are was evaluated with sampling rates between 4Hz and

20Hz.

The ferry was equipped with a hydraulic stabilisation system to dampen the roll move-

ments caused by the sea. Fairly calm weather conditions, and hence ship movements,
are encountered on approx. 360 days throughout the year on this particular route and

have been experienced also during the measurements. On the other hand it is clear that
other routes (e.g. transatlantic, in Pacific Ocean etc.) mean different changing weather
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 112

conditions throughout the year, causing different movement behaviour of the ship, with

or without stabilisation. Finally, the results are shown in Table 4.5. Measured pitch and
roll angle data showed a typical period of approximately 10s.

Table 4.5: Worst case angle, angular rates, and accelerations for 150m ferry
Parameter Range in ◦ Rate in ◦ /s Accel. in ◦ /s2
Roll ±3 1.5 2
Pitch ±3 1.5 3
Yaw 360 2 10

4.2.2.6 Aeronautical

As discussed in section 3.4 for the aeronautical case the angular agility ranges are dom-

inated by the wide scale mobility of the aircraft. An additional standard roll angle of

±25◦ and pitch angle ranging from -5◦ to 15◦ has to be considered and is influencing the
virtual horizon. Antennas with beam agility rates of about 70◦ /s in azimuth and 40◦ /s
in elevation provided sufficient performance during the aeronautical channel measure-
ments described in section 3.4.1. However, smaller or military aircraft may require

higher agility of the terminal antenna.

4.3 Antenna Beam Steering

The agility of the antenna must be sufficient to compensate vehicle movements for con-

tinuous availability of the satellite link. Outage times due to signal blocking cannot be

compensated by the pointing system, but the PAT system must minimise or even avoid
re-acquisition times after periods of signal blockage.

Agility of the antenna main lobe is achieved by mounting a reflector or fixed array

antenna on multi-axle pivots or employing an electronically steered active array.

CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 113

4.3.1 Mechanically Steered Antennas

Mobile terminal antennas are classically realised by reflector antennas or direct radiation

array antennas mounted on a gimbal which is providing a two axis rotation for azimuth

and elevation steering. A third degree of freedom (DoF) has to be employed in case
polarisation mismatch has to be countermeasured. Schematic implementations for a

parabolic and a Cassegrain antenna with three mechanically steered axes are shown in

Figure 4.7.

Elevation

Polarisation
Elevation
Polarisation

Azimuth Azimuth

Figure 4.7: Implementation examples of 3-axis mechanically steered antennas. Elliptically

shaped parabolic reflector (left) and Cassegrain reflector (right).

The height of an antenna is generally a critical issue of mobile terminal design. A non-
protuberant integration and the overall allowed height of the terminal out door unit are

limiting the antenna dimensions. Wind drag may impose restrictions, too.

Mechanically steered antennas therefore often employ elliptically shaped reflectors with

low height, but larger width. This results in a small beamwidth in the azimuthal plane
and wide beamwidth in elevational plane. For a mobile terminal which is operational

at higher latitudes in a GEO scenario the beam pattern in this case is advantageously

shaped to cope with the regulatory off axis EIRP as described in section 2.3 in azimuth
direction. This implies a higher azimuthal accuracy. But, yaw agility - which translate

most directly into the azimuthal requirements - is the most dynamic for all type of

mobile platforms as shown in the last section.

Each axial DoF increases the implementation effort of a mechanical steering platform.
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 114

Figure 4.8 shows different design options, for example a gimbal with two axis cardan

joints, a three axis spherical joint, and three composite rotational joints.

Figure 4.8: Gimbals with two axis cardan joints (left), a three axis spherical joint (middle), and
a three axis composite joint (right).

Mechanical steering platforms are limited in angular speed and torque of the motors
actuating the associated axis. Typically the mechanics for the first rotation axis has to
cope with weight of the mechanics of the second axis.

Advantageous gimbal layout would place the centre of gravity of the mechanics and
the antenna into the rotation axes of the steering platform. In this way static forces for
holding the mechanical structure are minimised. Moreover, PAT control systems for
mechanical steered antennas have to cope with the inertial response of the system to

dynamic changes. Countermeasures for overswing harmonics have to be implemented

e.g. by the employment of proportional plus integral plus derivative (PID) controllers

for actuation of the motors.

Mechanical beam steering might not only be implemented for terminal antennas. Also

satellite antennas are operational which allow to move the reflector and the footprint

with it. If single spot beams of a multibeam antenna can be adjusted independently, this
is typically based on electronically steered array antennas described below.

In [GHLW03] an example is given of an optical multibeam satellite antenna for aero-

nautical application. Optical free space communication allows very narrow spot beams,

because of the short optical wavelength. These spot beams may not necessarily cover a
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 115

contiguous area, but can be steered towards a single aircraft. Figure 4.9 show the design

option of a 64 beam optical transmit antenna based on spherical joints.

0,8 m

0,8 m

Figure 4.9: Mechanically steered satellite optical Tx-antenna array; 64 (8x8) Tx-telescopes with
ca. 90◦ pointing-range in each axis; each telescope having 3 cm diameter resulting in a far-field
on-axis gain of 98 dB at 1550 nm wavelength.

4.3.2 Electronically Steered Antennas

Array antennas open the possibility to control the main beam direction and beam pattern
shape electronically. This is done by a variable phase and, in general, amplitude of each
element. If the phase and amplitude can be controlled electronically this is called an
electronically steered phased array antenna. The sum of all amplifiers and phase shifters

is called beamforming network.

Figure 4.10 shows the principle of a phased array antenna. The planar wavefront of the

signal is tilted from the orthogonal angle of the antenna plane, if the single array signals
are shifted in phase and amplitude. In the depicted example the amplitude of the upper

element has to be amplified according to the additional path length x and the phase has
2π
to be shifted by λ
x. The number of active elements i.e. amplifiers and phase shifters is

determined by the number of array elements. The antenna directivity is determined by

the number of elements and spatial distribution. A λ/2 spacing of the array elements is
typically applied.

The gain of an antenna is proportional to the antenna size according to equation 2.19. If
the angle of incidence is not orthogonal to the antenna plane the absolute antenna gain is

reduced. Thus, array antennas have typically a limited angular range of less than ±60◦ .
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 116

Figure 4.10: Array antenna principle.

The beamforming network can be realised by programmable electronic phase shifters

and amplifiers at the single array elements or completely digitally by utilising digital
signal processors. Antennas with electronic phase shifters require a more complex ad-
justment and calibration of the beam forming network. A trade-off between calibration

effort and production accuracy of array element spacing and programmable electron-
ics results typically in this requirement of calibration of each single produced antenna.
Beamforming by digital phase shifting by represents a very flexible implementation, but

is today, especially at higher frequencies and signal bandwidths, limited by the required

processing power.

Not only the control of the main beam direction, but also the placement of nulls in
the direction of interferers is possible with array antennas. In this case one speaks of

an adaptive array. Adaptive beamforming arrays allow either smaller antennas or higher

data rates by reduction of multipath and interference suppression if applied to the receive
path and can suppress neighbouring satellite interference for transmit antennas.

For multibeam antennas, the signals of each array element are fed into a multiple of
beamforming networks as depicted in Figure 4.11. Receive and transmit paths are gen-

erally separated by a diplexer for each array element, unless the antenna is not com-
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 117

pletely detached in a transmit and receive array elements. The number of beamforming

networks determines the number of beams which can be formed by an antenna. Each
beamforming network includes the same number of active elements. Thus, an antenna

array with M beams and N active elements incorporates at least M · N active elements.

Figure 4.11: Beam forming network principle.

Very flexible – so called smart antennas – are sampling and converting the signals of
each array elements. Digital processing power is then the only limiting factor for the
adaptivity of the antenna and the number of beams.

4.3.3 Beam Steering Technology Tradeoff

Potential of optimisation for antenna designs in size, accuracy, axial range, degrees
of freedom, and especially costs is given when using mechanical or electronical beam

steering. Thus, a trade-off to satisfy the requirements being of foremost importance for

a particular scenario has to be found.

The maximal angular range, rates and accelerations of a steered antenna pose limitations

for the different antenna design options. Angular rates and acceleration are of special

interest, when a mechanical steering axis is involved in the antenna design, because
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 118

torque and speed are finite. Electronically steered axes are limited by the computational

power of the beamforming network or they may depend in pointing accuracy on the
resolution and switching velocity, if discrete phase shifters are used.

Several design approaches can be found in implementations incorporating the advan-

tages and drawbacks of the chosen technology.

Electronically steered antennas will definitely be an option for the future, but especially

at higher frequencies they are still costly because of the number of active elements and

their power consumption. Moreover, the restriction in angular agility range prevent a
mass deployment for platforms.

Connexion by Boeing for example introduced such antennas [AVA+ 00] for their aero-
nautical test flights with Lufthansa (cf. Figure 4.12) and British Airways, but costs and

angular agility range forced to change to mechanically steered antennas for the com-
mercial implementation [MMI+ 03].

Conformal terminal antennas which are shaped around the body of an aircraft could
be an option for solving the restricted agility range. However, this technology is still
subject to prototypal research today.

Nevertheless, for mass market applications (e.g. in cars), the availability of cheap adap-

tive array antenna technology with beam steering capability seems vital for competitive

commercial products.

Pure mechanically steered antennas have been used for a long time in maritime envi-

ronments over a broad frequency range and are still representing the most commonly

installed antenna types for mobile reception in general.

Their drawbacks lie in antenna height, and limited angular rates and accelerations.
Moreover, mechanical pivots are sensitive to vibration which occur in some mobile

platforms. Thus, endurance of the mechanical structure is challenging, especially for

train installations.
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 119

The capability of forming multiple beams with independent steering is of special interest

if a seamless HO between satellites has to be supported. This is the case for wide-area
mobile platforms and certainly for non GEO constellations. Mechanical solutions are

very restricted in this sense and typically two independent antennas have to be installed

for HO support.

Table 4.6 summarises tradeoffs of mechanical and electrical beam steered antenna

agility axes.

Table 4.6: Terminal antenna designs tradeoffs

mechanical beam steering axis electronical beam steering axis
mechanical joint must support angular scan angle requirements to be fulfilled
range
full gain in all directions antenna gain decreases with increased
steering angle
multiple beams possible, but not individ- multiple beams, each individually
ually steered steered are supporting seamless satellite
HO
angular speed determined by the axial sufficient computation speed and algo-
motors rithm convergence needed
sufficient motor torque and controllers restriction in angular resolution if dis-
for system inertia needed crete phase shifters are applied
susceptible to vibration and agitation complex calibration
today typically less expensive today typically more expensive

To date there is no one-fits-all mobile satellite terminal antenna solution that would
satisfy all requirements equally well (i.e. being small, highly accurate, and cheap),

instead there is always the need to find an compromise. The best compromise would
yield a design that shows sufficient antenna agility to support the required data rates,

with an acceptable platform size and acceptable cost.

A combination of both methods may be advantageous as well. A mechanical turn table

for azimuth steering with electronically elevation agility reduces the number of active
elements significantly and allows low implementation heights of the antenna. Such

hybrid antennas are commercially available for vehicular (e.g. [KVH06]) and train (e.g.
[Ray06]) applications. However, these antenna types are mostly for receive only.
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 120

Figure 4.12 is depicting several prototypes and commercially available antenna types

which have partly been used for measurements or application demonstration of this
work.

4.4 PAT Algorithms

PAT algorithms determine the mobile to satellite pointing vector pT S and control the

antenna main lobe by actuating motors, by setting weights of the beam forming network
or by digital computation. Depending on the antenna design and utilised sensor input
the PAT algorithm includes compensation for mechanical inertia of the antenna (e.g. by

a PID controller) and prediction of the mobile platform movement (e.g. by Kálmán3

filtering).

There are in principle two possible approaches to implement a PAT system:

• Closed-Loop PAT, where feedback from the received signal is exploited, and

• open-Loop PAT, where a sensor system is deployed to measure the platform po-

sition and attitude, allowing direct computation of the optimal antenna pointing
direction.

Both options will be detailed in the following section, and their implementation advan-

tages for different antenna designs will be discussed.

4.4.1 Closed Loop PAT

Closed loop PAT, relies on information about received signal quality (e.g. signal

strength, bit error rate, ...). In a first phase of the PAT process the satellite must be
acquired in a certain (e.g. hemispherical) area. For this, the received signal strength
3
Rudolf Emil Kálmán; * 19. May 1930 in Budapest
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 121

Figure 4.12: Antenna examples (from upper left to lower right):

- two mechanical axis Cassegrain antenna, Rx, K/Ka band, aeronautical use, [HJL99]
- mechanical azimuth and electronical elevation steering antenna, Rx/Tx, K/Ka band, vehicular
use, [APG+ 01]
- two mechanical axis fixed array antenna, Rx/Tx, K/Ka band, aeronautical use, [HLCL99]
- mechanical azimuth and electronical elevation steering antenna, Rx, Ku band, vehicular use,
[KVH06]
- two axis electronically steered phased array antennas, one RX one Tx, Ku band, aeronautical
use, [Luf06]
- two axis electronically, one axis mechanically steered antenna, RX, Ku band, maritime use,
[LH03]
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 122

must exceed a certain limit. Then a tracking algorithm like conical-scan or quadrant

detection technique tracking can maintain the antenna in the direction of the satellite.

Closed loop PAT is inappropriate for systems developed for mobile application with

spreading technologies. In the forward link of spread spectrum systems, a low SNR
is resulting before de-spreading of the receive signal. A closed loop PAT algorithm is

depending on a feedback of the receive signal after de-spreading. This may result in

a slow performance of a closed loop PAT implementation or even preventing this PAT

option, totally.

If linear polarisation is used, correction of polarisation mismatch is complex by pure

signal power measurements, thus, closed loop PAT is mainly applicable for azimuth and

elevation steering.

With conical scan tracking, the antenna detector maximum is moved with two orthogo-
nal functions (sinus and cosines), thus the beam rotates around the antenna boresight. If
the antenna boresight does not coincide with the LoS angle to the satellite, the rotation
of the variant antenna gain modulates the amplitude of the received signal (cf. Figure

4.13). From this amplitude modulation, a control signal for antenna tracking can be
derived.

Figure 4.13: Modulation path of the LoS direction on Gaussian antenna detector profile.

A control signal for antenna tracking can also be derived by regular, stepwise displace-

ment of the antenna boresight with respect to the LoS. Again, displacement results in a
degradation of the receive amplitude. Convergence to LoS can be achieved by evaluat-

ing the difference of received signal amplitudes as the gradient of the detector.
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 123

For both techniques a trade-off between offset angle (or respectively step width) and

receive amplitude loss has to be found. One tenth of the 3 dB angle of the antenna is
typically used [Ric86] as offset angle. Moreover, both techniques are based on sequen-

tial measurements of the degraded receive amplitude at different off boresight locations.

The performance of the PAT algorithm is depending on the sampling rate of these mea-
surements.

Also the quadrant detection technique (derived from the radar mono-pulse technique)

needs an adaptation of the antenna, if mechanical reflector antennas are used, but does

not require continuous movement of the antenna for deriving a PAT control signal. The
quadrant detection technique utilises the signals received from four separate detectors,
which deviate from the main antenna direction. The data signal results from adding up
the signals of all detectors, while the antenna control signal is calculated from differ-

ences between the signal amplitudes.

The pointing performance of closed loop systems is heavily influenced by signal shad-

owing, because there is the possibility that the mobile changes its direction during shad-
owing and hence the pointing angle of the antenna changes. The re-acquisition of the
satellite after shadowing is thus more difficult. The re-acquisition rate – mainly influ-

enced by the shadowing rate – is a very important parameter for closed loop PAT and

influences the re-synchronisation process of the transmission link.

The mean time for re-acquisition may be reduced by the use of additional data e.g. the
last known elevation angle of the satellite position relative to the terminal or sensors.

Without additional sensors the antenna pointing position of the last good channel state

is the only indication for the initial search position. From this point the system can start
re-acquiring the satellite.

Tracking by conical scan (or any kind of dithering) is imposing a continuous beam
movement for deriving a PAT control signal, which may lead for mechanical steered

axes to additional requirements or reduced lifetime of the mechanical parts.

CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 124

4.4.2 Open Loop PAT

Determination of mobile’s attitude and position and the satellite’s position is another

option for PAT and is called open loop PAT. This method is especially interesting for
highly agile vehicles in shadowing channels. In this case, open loop PAT is more robust

than closed loop PAT as it does not rely on received signal strength.

The accuracy of open loop PAT depends on sufficiently accurate knowledge of mo-

bile position, mobile attitude and satellite position. There is no feedback of the signal
strength, so open loop systems can only steer (and not control) the mobile antenna on

knowledge of location and attitude. An open loop PAT system requires no hardware

modification for the antenna itself, but, sensors for the mobile location and attitude are
necessary. The accuracy of these information determines the resulting pointing error

for low frequent movements of the mobile. An assessment of applicable sensors with

respect to accuracy and price is given within section 4.6.4.2. The compensation of high-
frequent attitude changes is limited by the sampling rate of the attitude sensors and the
velocity of the antenna.

The antenna control being based on an open-loop PAT algorithm requires knowledge of:

• antenna platform location being typically provided in ECEF coordinates, i.e. lon-

gitude, latitude, height,

• antenna platform attitude w.r.t a reference frame, e.g. provided in Euler angles
yaw, pitch, and roll, and

• satellite location being typically provided in ECEF, coordinates, i.e. longitude,

latitude, height.

With this information, the pointing of the antenna towards the satellite is a geometric

calculation where the line-of-sight vector pE

T S is transformed from the ECEF coordinate

frame to the antenna platform coordinates. This coordinate transformation is given in

annex C.
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 125

4.4.3 PAT Algorithms Tradeoff

Table 4.7 summarises the aforementioned drawbacks and advantages of the two different

PAT algorithm solutions.

Table 4.7: PAT algorithms tradeoffs

open loop PAT algorithm closed loop PAT algorithm
performance not influenced by signal performance influenced by signal shad-
shadowing owing
additional sensors for mobile attitude autarkic from attitude (and position) sen-
(and position) necessary sors, but introduces continuous move-
ment of beam, antenna detector modifi-
cations, or additional beamforming net-
work
heading determination difficult by sen- same accuracy for all DoF
sors (compass, dual antenna GPS, kine-
matic GPS)
sensors cost increase drastically when sampling rate and computational power
high accuracy is required at highly agile increases with required accuracy and mo-
mobiles bile agility
suitable for polarisation determination polarisation mismatch complex to be
counteracted by signal strength measure-
ment
independent of receive signal accurate control signal is difficult to ac-
cess, if spread spectrum signals are used

Combinations of closed and open loop with different focus are often advantageous.
Various combinations with different weight on the two pure algorithms can be used and

their selection depends on the particular application scenario of the terminal and the

terminal antenna technology.

For example in an aeronautical application, the availability of precise attitude and posi-
tion data in the aircraft, provides all necessary information for an open loop implemen-

tation. In addition a closed loop focused approach is a viable solution, too, since the

aeronautical channel is not suffering from fading and shadowing influences.

The land mobile channel with its severe shadowing would demand for an open loop sys-
tem, but precise attitude and position data are expensive and not accessible in vehicles.

A combination of cheap but less accurate sensors for the provisioning of heading, roll
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 126

and pitch angles of the mobile can be implemented in the antenna controller and pro-

vide sufficient information of the actual attitude. The closed loop tracking process can
start when the received signal strength exceeds a certain threshold. Thus, the angular

accuracy of such sensors need to be in the range of the corresponding antenna angle

only.

For land mobile and aeronautical environments example implementations of combined

PAT algorithms are given later on in this chapter.

4.5 PAT Errors

Mismatch of the antenna boresight with respect to the mobile to satellite pointing vec-

tor pT S translates into loss of signal power and increased interference. This pointing
error may occur due to:

• inadequate angular precision of mechanical steering platform or electrical phase

shifter resolution

• inaccurate sensors

• overall latency, e.g. due to sampling rate, computation power, system inertia

Typically, a certain pointing error and an associated signal loss and/or interference can

be tolerated and is included in the link budget calculation of a mobile satellite system.
In the following the different errors which accumulate to the pointing vector will be

given.

4.5.1 Azimuth and Elevation Errors

4.5.1.1 Angular Azimuth and Elevation Error

The LOS vector in the local coordinate frame is defined according to

CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 127

   
 pLTSx   cos ϕ cos ε 
   
pLTS = L
 pT Sy
 =  sin ϕ cos ε
 
 ,
 (4.1)
   
pLTSz sin ε

with ϕ and ε being azimuth and elevation, respectively, in the local coordinate frame,
defined in equations 2.13 and 2.7.

Accordingly, the LoS vector in the antenna coordinate frame is defined according to

   
 pA
T Sx   cos ϕA cos εA 
   
pA
TS = A
 pT Sy
 =  sin ϕ cos ε
  A A
 ,
 (4.2)
   
pA
T Sz sin εA

with ϕA and εA being azimuth and elevation, respectively, in the antenna coordinate
frame.

The following equations determine the antenna azimuth ϕA and elevation εA against the
attitude parameters of the mobile platform

à !
pLTSx (−sy cr + cy sp sr ) + pLTSy (cy cr + sy sp sr ) + pLTSz cp sr
ϕA = arctan 2 , (4.3)
pLTSx cy cp + pLTSy sy cp − pLTSz sp

and

³ ´
εA = arcsin pLTSx (sy sr + cy sp cr ) + pLTSy (−cy sr + sy sp cr ) + pLTSz cp cr , (4.4)

where cy = cos(α) , cr = cos(β) , cp = cos(γ) and sy = sin(α) , sr = sin(β) ,

sp = sin(γ) denote the cosine and sine, respectively, of the yaw angle α, roll angle β ,

and pitch angle γ.

CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 128

The equations can be used to assess the pointing error that would occur if errors of the

mobile attitude or position determination are present. The mispointed antenna azimuth
and elevation angles can be calculated by adding the error in attitude determination to

the yaw, roll and pitch angle. cy , cr , cp and sy , sr , sp are substituted by c(y+∆y) ,

c(r+∆r) , c(p+∆p) and s(y+∆y) , s(r+∆r) , s(p+∆p) , respectively.

The mispointed antenna boresight vector is defined by

 
 cos(ϕA + ∆ϕA ) cos(εA + ∆εA ) 
 
pA
T ∆S =
 sin(ϕA + ∆ϕA ) cos(εA + ∆ϕA )
 ,
 (4.5)
 
sin(εA + ∆εA )

with ∆ϕA and ∆εA being the angular error of the antenna in azimuth and elevation,
respectively.

Mobile platforms are often operational at small roll and pitch angles, due to the fact that
large movements are impacting the comfort of passengers. Therefore the equations can
be simplified for small roll and pitch angles.

For roll and pitch angles being zero, equation 4.3 simplifies to

à !
−pLTSx sy + pLTSy cy
ϕA (β = 0, γ = 0) = arctan 2 , (4.6)
pLTSx cy + pLTSy sy

and obviously with roll and pitch angles being zero, the antenna elevation is a direct
component of the LOS vector and equal to the elevation of the local frame (cf. equation

4.4).

¡ ¢
εA (β = 0, γ = 0) = arcsin pLTSz = ε . (4.7)

In section 4.2.2 values for range, rate, and acceleration of each rotational movement roll,

pitch, and yaw of different mobile platforms are given separately. Worst case angular
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 129

ranges, rates and accelerations for local or antenna azimuth and elevation can be derived

by superposing these to an overall movement.

Equation 4.8 can be used to estimate the worst case elevation error depending on the
maximum roll and pitch angle errors ∆βmax and ∆γmax .

p
∆εmax ≈ min(90◦ , ε0 ± 2
∆βmax 2
+ ∆γmax ), (4.8)

with ε0 being the elevation with respect to horizontal and a variation around this eleva-

tion resulting from mobile platform attitude changes.

A typical sensor system for the determination of the mobile attitude is an inertial nav-
igation system (INS) consisting e.g. of gyroscopes, accelerometers and compass. The
sensor accuracy is in general the same for roll and pitch and differs for the heading as

determination of the north vector involves another sensor type. For radial symmetric
antennas the maximum overall pointing error can be approximated by

p
ep ≈ ∆ϕ2max + ∆ε2max . (4.9)

4.5.1.2 Signal Losses

Signal loss and interference caused by azimuth and/or elevation mismatch is directly

related to the shape of the receive detector.

Equations 2.27 and 2.32 give the receive power and the carrier to noise and interference

ratio dependent on the antenna off boresight angle. Equation 2.20 defines the power loss

of the antenna with circular-Gaussian shaped main lobe with respect to the boresight

gain. With these equations, an acceptable maximum off boresight angle ϑmax can be
derived by

r
1
ϑmax = ϑ3dB (Gmax − G(ϑmax )) . (4.10)
3
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 130

Assuming that a pointing loss of 0.5 dB is acceptable for the received power it yields

1
ϑmax = √ ϑ3dB ≈ 0.4 ϑ3dB . (4.11)
6

4.5.1.3 Optimisation of Pointing Error vs. Antenna Gain

Equation 4.10 defines the maximum acceptable off boresight angle in dependence on

the antenna gain and beamwidth, respectively. Figure 4.14 plots the gain of an antenna

against its off boresight angle and its beamwidth, reflecting, that the gain loss near the
boresight angle is very small due to the shape of antenna gain contour. By interpretation

of the acceptable off boresight angle as an allowed pointing error, this representation of
equation 4.10 allows the derivation of the maximum achieved gain in case of a given
pointing accuracy of the antenna and PAT system. Thus, due to the shape of antenna

gain contour the allowed pointing error can be traded off versus the antenna gain for
receive antennas.

An antenna with a 3 dB beamwidth of 8.4◦ achieves an antenna gain of 16 dBi at ap-

proximately 10◦ off boresight angle (which states the 4.3 dB beamwidth). This is the
maximum gain which a terminal can achieve with 10◦ pointing error (cf. red lines in
Figure 4.14.

The Figure 4.14 allows determination of antenna beamwidth versus allowed pointing
error for a given required antenna gain. For example, when a receive gain of 20 dBi

has to be achieved for a terminal, this can be reached by using an antenna with 8.5◦

beamwidth and 1◦ maximum pointing error as well as by an antenna with 7◦ beamwidth

and 5.4◦ maximum pointing error or an antenna with 5.2◦ beamwidth and 6.4◦ maxi-

mum pointing error. Vice versa the presentation of antenna beamwidth versus pointing
error in Figure 4.14 also reflects, that with a maximum pointing error 6.4◦ a maximum

terminal antenna gain of 20 dBi can be achieved.

It can be derived from Figure 4.14, that a cost optimised pointing accuracy of re-

ceive antennas is achieved when the allowed PAT accuracy equals the 4.3 dB antenna
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 131

beamwidth. The pointing accuracy impacts the costs of a PAT system, while the antenna

size and gain determine the antenna costs. Both factors depend on the utilised PAT al-
gorithm and antenna technology. Lower PAT accuracy can be traded off with increased

antenna size and gain, up to a maximum mispointing of 4.3 dB.

Counteracting the mispointing of 4.3 dB by an increased antenna gain (or respectively

size cf. 2.19) equals to an increase of antenna diameter of factor 1.6. The associated

variation of the receive power is typically compensated by power control of the mobile

receiver.

0.5 dB 1 dB 2 dB 3 dB
15 15

14 15
16

13
13 16

14
4.3 dB
12 17
16
15
17
11
antenna beamwidhth ϑ 3dB in deg

18
increasing size of antenna

10 18 17

9 19

12
13
14
20
16

8 10
15

20
18

7 11
19

21

6 23 22 12
13
17

24 14
5 25
1615
26
4
20

27
18
23

28
3
21

19
24

30 22
25

2
26
27
32
34

1
36

increasing cost of PAT system

38 0

4442
4

0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
antenna off boresight angle ϑ in deg

Figure 4.14: Antenna beamwidth ϑ3dB in degree versus off boresight angle ϑ (translating into
the allowed pointing error) for a given required antenna gain (solid blue lines in dBi).
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 132

4.5.2 Polarisation Error

4.5.2.1 Angular Polarisation Error

A loss of the received power due to a polarisation mismatch in mobile scenarios is

only occurring when linear polarisation is used. Classically circular polarisation is used

for mobile satellite applications, but with the upcoming trend of utilising Ku and Ka

band satellite services for broadband mobile terminals also linear polarisation has to be
considered.

In the following, the equations for deriving the power loss due to polarisation mismatch
for linear polarised satellite signals will be given.

The previously discussed maximum roll and pitch angles of a mobile vehicle allow an
estimation of the polarisation loss due to attitude changes for linearly polarised satellite
signals. Cross correlation can be estimated, too, if vertical and horizontal signals are

used in the system simultaneously.

Besides the effect of attitude changes of the mobile, polarisation loss may also occur

due to position changes if the platform is mobile in a large operational area.

The two linear polarisations are called vertical and horizontal polarisation. They are

defined by two orthogonal vectors, both perpendicular to the propagation vector.

Figure 4.15 shows the definition of the horizontal and vertical polarisation vectors with
respect to the local horizontal plane of mobile terminal antenna. The horizontal polar-

isation angle of the mobile terminal ςP,H is defined as angle between the line of inter-

section (of the local horizontal plane and plane spanned by the polarisation vectors) and
the horizontal polarisation vector.

If the relative longitude of the satellite orbital position and the satellite footprint centre

as well as the latitude of the satellite footprint centre is not zero, a polarisation offset ςP 0
is typically introduced at the satellite antenna for pre-compensation, e.g. by tilt of the

antenna feed. In this case, one speaks of a shift between the satellite’s equatorial/polar
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 133

Figure 4.15: Definition of polarisation angles ςP,H and ςP,V .

and horizontal/vertical polarisation planes (cf. Figure 4.16). In general, any inclination
i of the satellite would additionally rotate the equatorial/polar and horizontal/vertical
polarisation.

Figure 4.16: Definition of polarisation tilt angle.

CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 134

In [CCI] an equation for the equatorial polarisation angle ςP,E is provided for the case

of a GEO satellite scenario:

 
µ ¶s µ ¶2
∆λ sinψ
ςP,E = arctan sin 1+ , (4.12)
tan δT rn − cos ψ

where:

∆λ : relative longitude of the satellite and the terminal,

ψ : earth central angle between Sub-satellite point and the receiving ground terminal,
δT : latitude of receiving ground station,

rn : the radius of the geostationary orbit normalised by the radius of the earth.
Consequently, equation 4.12 has to be extended for the general case to

 v 
u à !2
 sin(λS − λT ) u
t1 + sinκ 
ςP,H = ςP 0 + i − arctan  rS −rT . (4.13)
tan(δT − δS ) RE
− cos κ

with κ = arccos[cos(λS − λT ) cos(ϕT − ϕS )]

Figure 4.17 shows the lines of isocline polarisation tilt angles. Wide range translatory
movement of a mobile terminal perpendicular to this isoclines requires active polarisa-

tion steering to compensate for signal losses.

4.5.2.2 Signal Losses

The polarisation mismatch angle is the sum of effects due to position changes (to a
nominal fixed polarisation preset) and attitude changes of the vehicle. In Figure 4.18

the influence of a polarisation mismatch ∆ς on the antenna receive signals is illustrated,

where

   
 − sin ∆ς   cos ∆ς 
sP (t) = sP (t)   and sE (t) = sE (t)  
cos ∆ς sin ∆ς
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 135

Figure 4.17: Isocline line of polarisation tilt angle for a satellite position at 5.2◦ E longitude and
0◦ E latitude and satellite tilt angle offset = 0.

are signal vectors describing both the co-polarised as well as the cross-polarised signal

components.

Figure 4.18: Impact of mismatch between signal and antenna polarisation (polarisation mis-
match).

For ∆ς = 0 the receive signals rE (t) and rP (t) carry the wanted signals sE (t) and sP (t)

without interference from the cross-polarised signal component, such that (noise is not
relevant for these considerations):



 rE (t) = sE (t)
∆ς = 0 ⇒ . (4.14)

 rP (t) = sP (t)
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 136

For ∆ς 6= 0 interference is introduced from the cross-polarised signals and it yields:



 rE (t) = sE (t) cos(∆ς) − sP (t) sin(∆ς)
∆ς 6= 0 ⇒ . (4.15)

 rP (t) = sE (t) sin(∆ς) + sP (t) cos(∆ς)

Cross polarisation loss Lxpol due to polarisation mismatch is determined by:

µ ¶
(cos ∆ς)2
Lxpol = 10 log in[dB] . (4.16)
(sin ∆ς)2

This is resulting in e.g. a maximum acceptable polarisation angle mismatch of 1.8◦ if

30 dB cross polarisation discrimination has to be achieved.

Neglecting cross-polarisation, the loss to the receive signal due to polarisation mismatch
Lpol can be estimated according to

Lpol = 20 log(cos ∆ς) in[dB] . (4.17)

For vehicle movements with pitch and roll typically less than 5◦ , losses are below
0.05 dB and hence considered negligible.

4.5.3 Latency Error

System latency is the dead-time in the antenna pointing system that results from the

inevitable delays introduced in the signal processing chain. Latency is resulting from

• filtering and averaging of sensors,

• sampling rate,

• computation power, and

• and mechanical movement of motors (if applicable).

CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 137

For an open loop with mechanical axes this latency for instance is summarising from

i) mobile platform movement, ii) sensor averaging, iii) antenna pointing system detects
movement and determines motor control commands, iv) motor counteracts mobile plat-

form movement sensor.

For a closed loop system with electronically steered axes, latency typically summarises

from averaging window for receive power and computation of new beamforming net-
work weights.

The maximal angular accelerations are occurring in vehicular and yacht environments

with a value of 100◦ /s2 . The tolerable latency yielding a pointing error below ep can be

calculated by:

r
ep
tlatency = 2 . (4.18)
100◦ /s2

This would lead to approximately 40 ms system latency for an acceptable angular error
of 0.1◦ .

4.6 PAT Solutions for Different Environments

4.6.1 Assessment of PAT Solutions for Different Environments

This section discusses PAT algorithms for different mobile environments and applica-

tions. The following constraints impact the PAT selection:

• Antenna Parameters and Regulations:

Regulatory standards are applicable (and are influencing the requirements of PAT

accuracy) if a mobile transmit link is considered, independently of the mobile

environment. For receive only applications a certain PAT accuracy has to be

achieved as well to ensure service quality. Depending on the data rates of the
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 138

application and the associated antenna gain, the required PAT accuracy can be

determined. In general, closed loop algorithms allow higher accuracy, but both,
open and closed loop algorithms, require increased effort with higher antenna

gains.

If the application requires compensation of polarisation mismatch, closed loop

algorithm relying on signal power feedback reach their limit and determination of

the platform attitude angles becomes necessary.

• Operational Scenario and Constellation:

Worst case angular rates and accelerations are occurring in land mobile car and
maritime environments. Aircraft and ships operate on a large scale, thus no re-

striction in the operation area is applicable. While cars and especially trains are
operating in a very little inclined range with respect to the horizontal plane, small
ships and aircraft are agile in roll and pitch angular range. In general smaller
vehicles move more agile than bigger ones.

An operation of the platform near the horizontal plane results in small roll and

pitch variations and gives open loop algorithms an advantage as only the azimuth
angle has to be determined, while antenna elevation is restricted to the location of

the mobile terminal with respect to the satellite.

Independently of the mobile environment, the tracking of a non-GEO satellite

with a terminal antenna would require the knowledge of the satellite position for
a pure open loop algorithm. This information can be calculated by the satellite

constellations parameters, which might change with the operational time of the

satellite system. Thus, open loop PAT are applicable for GEO satellite scenarios,

while PAT algorithms for non-GEO constellations incorporate at least a closed

loop component.

• Availability of Attitude and Position Sensors:

Typically aircraft (and especially airliners and business jets) are fitted with accu-
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 139

rate attitude and position sensors. Here, the implementation of an open loop PAT

is favourable, provided the sensor information is accessible.

• Mobile Channel Parameters:

Especially for channels with low shadowing probability (i.e. aeronautical or mar-
itime channels) PAT implementations allow a reduction of the open loop part and

an emphasis of the closed loop.

In conclusion, the vehicular environment poses the most challenging situation for a cost

and performance optimised PAT implementation, because of its fast moving mobile
platform, the lack of attitude information, and a fast fading and frequently shadowed

mobile satellite channel. The most unproblematic environment for a PAT implementa-
tion is the aeronautical one, opening the possibility of open and closed loop algorithms
in an efficient way. Table 4.8 is summarising the PAT challenges for different mobile
environments.

Table 4.8: PAT challenges for different environments (decreasing from top to bottom)
environment open loop PAT algorithm closed loop PAT algorithm
land mobile car typically no attitude sensors, fast fading shadowing channel
but GPS localisation
land mobile train no attitude sensors required, shadowing and tunnels
but GPS localisation and fixed
tracks
maritime no or slow attitude sensors re- slow fading channel
quired, GPS localisation
aeronautical accurate attitude and position slow fading channel
are available

Implementations of PAT solutions for different environments have been realised for this
work, of which three solutions will be presented in the following sections.

4.6.2 Open Loop Vehicular PAT Implementation

The land mobile EHF band channel measurements described in section 3.2, required

an implementation of a PAT algorithm for the mobile terminal antenna with a 3 dB

CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 140

beamwidth of 25◦ to be pointed towards the satellite emulated by an aeronautical plat-

form. The mechanical steering platform allows a tracking of the antenna with an az-
imuth speed of 70◦ /s and elevation speed of 40◦ /s.

The first implementation approach for the antenna steering was a differential GPS and
gyroscope based open loop algorithm. The aircraft reported its position through a
telemetry link to the measurement vehicle. The receiver was equipped with a steer-

ing processor calculating the pointing angle to the aircraft emulating the satellite. A
mechanical gimbal was used to steer the terminal antenna, cf. Figure 3.13.

In this geometrical relation the precise position of the aircraft was essential, thus an on-

line differential GPS position reporting was assessed, in order to achieve a high pointing
accuracy. A GPS reference station was therefore operated, leading to a position accuracy
of approximately 2 m of aircraft and measurement van. An INS, based on mechanical

three axis gyroscopes, accelerometers, several compasses and a travel pilot, provided an
attitude determination of better 0.25◦ in roll and pitch and better 0.5◦ in heading for the
measurement van. Evaluation of the pointing performance according to equation 4.9
results in an overall angular error of 0.6◦ .

The mechanical platform steered a camera besides the antenna (cf. Figure 3.13), thus
it was possible to measure the antenna PAT performance. The accuracy of the antenna

PAT was derived by evaluating the aircraft position in the video picture. Processing

all available data for the open loop algorithm leads to a resulting pointing error with a
standard deviation of less than 0.7◦ . Figure 4.19 shows the results of this evaluation.

This pointing accuracy was more than sufficient during the first trials, where a horn

antenna with 3dB beamwidth of 25◦ was used, resulting in signal loss below signal
measurement accuracy due to mispointing according to equation 2.20.

But there are two significant drawbacks of this algorithms. First, the use of antennas

with higher gain and smaller beamwidth rises the need for PAT algorithms with better
accuracy and second, the use of a high precision three axis gyro-scope is not applicable

for a commercial mass market implementation e.g. in cars.

CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 141

Figure 4.19: Histogram of pointing error for an open loop PAT implementation. Measured dur-
ing a land mobile channel characterisation campaign with satellite emulation by an aeronautical
platform.

4.6.3 Aeronautical INS Based Open Loop vs. Conical Scan Closed

Loop PAT Implementation

The aeronautical channel measurements described in section 3.4 have been performed
using different PAT algorithms.

The receiving antenna was a Cassegrain type with a beamwidth of 4.2◦ . For mechan-

ical steering the same gimbal was used as described above for the land mobile EHF
narrowband measurements.

Two control modes have been implemented for the antenna PAT

• open-loop tracking based on GPS and avionics inertial measurement unit (IMU)
data, and

• closed loop tracking algorithm based on conical scan.

For the open loop realisation, position information of the aircraft was determined by
GPS and the attitude information by the avionics INS. The avionics INS was the same

as used for the land mobile measurements described above.

CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 142

The aeronautical channel measurements have been performed using a GEO satellite.

No visualisation of the target could be used for PAT performance verification. But, the
mobile agility requirements for aircraft are less strict for aeronautical maneuvers than

for land mobile. The overall angular pointing error was less than 0.08 dB signal loss.

This was validated by evaluation of signal power losses during LoS and flight paths
with standard maneuvers i.e. holding pattern. No dominant signal loss during these

manoeuvres have been observed (cf. Figure 3.24 left side).

In general, most aircraft of the business jet and airliner category have accurate INS

installed to support an open loop PAT implementation. But, the PAT algorithm requires

an interface, typically an ARINC429 bus. A PAT implementation independent of this

aeronautical interface would be favourable.

A conical scan algorithm was designed and verified for the same aeronautical scenario

as above. Figure 4.20 shows the design of a mechanically steered antenna realised for
the aeronautical channel measurements discussed in section 3.4. This implementation

is based on the lateral offset of the sub-reflector rotated by a motor [HS99]. The solu-
tion requires a third motor, but does not impact the azimuth and elevation motor of the
mechanical gimbal by a continuous movement of the azimuth and elevation motors (cf.

Figure 4.20). Moreover, the azimuth and elevation motors would have been too slow for

a fast conical scan.

The receive signal was measured with a power diode at intermediate frequency 70 MHz
and fed into a data acquisition (DAQ) and digital signal processing (DSP) board. From

the received modulated signal a tracking control signal was extracted for the azimuth and

elevation motor and a phase locked loop (PLL) for synchronisation of the scan motor
was implemented. Figure 4.21 illustrates the setup of the conical scanning control setup.

The control signals for the azimuth and elevation controllers have been derived as de-
picted in Figure 4.22 with ϕscan and fscan being the conical scan angle and frequency,

respectively.
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 143

Figure 4.20: Conical scan antenna with an eccentrical rotating sub-reflector in a Cassegrain
setup.

Figure 4.21: Conical control system block diagram.

Satellite acquisition was realised by an arithmetic spiral scan of the hemisphere starting

at antenna zenith with an spiral arm distance of 10◦ , which was approximately the two

sided beamwidth 2ϑ3dB of the antenna. Acquisition times of about 20 to 30 seconds

have been observed. This was obvious, because the system was powered on ground on

a horizontal plane and the spiral needed about 4 rotations (with the given angular speed
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 144

Figure 4.22: Conical scan detection block diagram.

of the gimbal) to reach the antenna elevation angle of about 48◦ in southern Germany.
The algorithm was enhanced by starting the spiral scan at the last stored elevation angle
minimising the acquisition times.

Validation of the PAT algorithm was done as described above by evaluation of the re-
ceived power during LoS situations during standard flight manoeuvres. As for the open

loop implementation no signal loss during manoeuvres have been observed.

4.6.4 Vehicular Low Cost PAT Implementation

For applications targeting at mass market, a low cost implementation for the antenna

and PAT design is inevitable. Within the ESA project "Mobile Ku band Receive only

Terminal Demonstrator" such an application is developed and demonstrated. End of life

satellites, that might already have inclined orbits shall be used for data services in cars.

4.6.4.1 Required Pointing Accuracy and Terminal Angular Parameters

According to the PAT design methodology in Figure 4.2, the terminal antenna and satel-
lite system parameters are given first for evaluation of the required pointing accuracy

and terminal angular parameters.

CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 145

One proposed satellite for the service is e.g. ASTRA 1A. The orbital position of the

satellite is longitude 5.2◦ E with an inclined orbit. The operational service area is re-
stricted to mid Europe. The application is for receive only, thus no regulatory require-

ments for the antenna and pointing accuracy apply.

The terminal antenna has a diameter of 10 cm, resulting with equation 2.19 in a maxi-

mum gain of 21 dBi at 12 GHz, assuming an antenna efficiency of 0.7. Equation 2.21
gives a 3 dB beamwidth of approximately 9◦ . An overall loss of 3 dB is specified as

acceptable for mispointing of the antenna. The service is using spread spectrum sig-

nals, but a signal strength feedback with update rate of 10 ms is available. The signal is

linearly polarised without a cross polarised signal being present.

Vehicle movements are specified according to Table 4.1 and are typically smaller than

±5◦ in roll and pitch.

According to Figure 4.3 the angular elevation range of the terminal antenna could be

restricted from 25◦ to 40◦ with an additional ±5◦ for the roll and pitch angels of the car,

this results in an overall range of 20◦ to 45◦ . Obviously the angular azimuth range is
360◦ .

With the given values for antenna beamwidth and acceptable pointing loss, an angular
accuracy of approximately ±9◦ is resulting, according to equation 4.10.

As no cross-polarised signal is present, polarisation losses are considered negligible

according to equation 4.17 for the attitude changes of the car. Figure 4.17 is showing
the polarisation tilt angle due to mobile location movements in Europe. A worst case tilt

angle difference of ±15◦ is covering the service region and resulting in a receive signal

loss of 0.3 dB.

Equation 4.8 defines the maximum angular error due to roll and pitch changes resulting

in 7◦ for vehicular environments. This is less than the acceptable pointing error. There-

fore, it might not be necessary to countermeasure the attitude changes in the elevation
axis. Assuming that the elevation remains fixed for attitude changes, an azimuthal point-

ing error of 5◦ is acceptable according to equation 4.9.

CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 146

Summarising the resulting PAT requirements,

• an angular range of 360◦ in azimuth and 20◦ to 45◦ in elevation,

• an angular resolution of 5◦ in azimuth and elevation, and

• an elevation determination based on position information with accuracy of some

km and additional heading information for azimuth determination with accuracy

of 5◦

is necessary to achieve the required pointing accuracy.

4.6.4.2 Assessment of Sensors

In this section applicable sensor types and their accuracy will be discussed.

GPS Receiver

Although GPS is in principle not the only option to determine the location, it is by far
the most practical one in terms of speed, cost, and precision.

For a moving platform, a standard GPS receiver provides velocity information in the

e.g. ENU frame. This information is derived from the observed Doppler shift and can
be used to determine the heading.

Moreover, it is possible to determine not only the location but also the platform attitude
using multi-antenna GPS receivers. With three sufficiently separated GPS antennas it is

possible to measure yaw, pitch and roll with high accuracy and high rate. Depending

on the actual antenna separation an accuracy of approximately 0.1◦ can be achieved.

However, such GPS receivers are very costly and will thus not be considered suitable

for a low cost mass product.

Various error sources are occurring when vehicle position and heading is determined

via GPS. The most relevant error sources are:

CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 147

• loss or unreliable GPS information that cannot be detected via GPS reliability

information,

• misalignment of GPS heading and vehicle heading in curves, and

• low data rate of GPS.

GPS loss will of course inevitably result in temporarily unknown position and heading.
This might at a first glance not seem of importance, as satellite communication is also

not possible if the navigation satellites are shadowed, but reacquisition time matters
especially in the land mobile channel. The number of satellites in view and dilution

of precision information provided by the GPS receiver usually allows to estimate the

reliability and validity of the GPS position and velocity data.

Theoretically, GPS allows an update rate of up to 1 kHz, however, commercial products

offer update rate from 1 Hz to 20 Hz, only. This update rate is sufficient for position
determination of the mobile lateral movements. But, according to the expected max-

imum yaw rate of 40◦ /s as given in Table 4.1, the relatively low update rates of 1 Hz
may result in a heading error of up to approximately 40◦ between the GPS heading up-
dates. Increasing the number of heading estimates by extrapolation seems problematic

as it involves the derivation of the noisy GPS heading information, thus increasing the

variance.

GPS heading information is derived from the velocity information in the e.g. ENU

frame. Unfortunately, the vehicle heading and the GPS heading only coincide if the

vehicle is driving a straight line. In curves, this holds no longer because the measured

velocity components of the vehicle speed depend then on the location of the GPS an-
tenna, relative to the location of the centre of gravity around which the vehicle rotates

while passing through the curve. The resulting vehicle heading estimation error can be

minimised by placing the GPS antenna above the centre of the axle of the usually non-
steered rear wheels. However, because of the non-zero side-slip angle there remains an

inevitable estimation error.

CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 148

These errors are also most relevant because they may occur under line-of-sight condi-

tions and the magnitude being independent of velocity. GPS alone may be insufficient
due to the various error sources, in particular due to the inevitable heading estimation

errors in curves.

Compass

Determination of platform heading results in the need to determine the True-North di-

rection. The "classical" choice for North determination would be of course a magnetic

compass. A compass can in principle determine the magnetic North direction for a static
and for a moving platform. The local declination (i.e. the difference between magnetic
and True-North) is compensated using a map of the declination.

The main advantage of an magnetic compass is that it can provide heading information

for the driving as well as for the stationary vehicle. In the absence of local magnetic
anomalies and locally generated magnetic fields, and with proper compensation for bi-
ases (variation and deviation), the static performance of magnetic compasses can rival
that of much more expensive gyrocompasses.

However, due to the principle of operation, magnetic compasses are in general sensi-

tive to external sources of interference, like a platform tilt, the metal car body, elec-

trical cables, or even geological peculiarities. This demands for additional electronics,

namely tilt sensors and a 3-axis magnetic sensor cluster, to counteract such potential

error sources increasing the cost.

Low-cost compasses comprise a two-axis magnetic sensor cluster which can determine

magnetic North at an precision of approximately 2◦ . However this accuracy is achieved

only when the sensor is levelled. Tilting the sensor will result in a heading error due

to the vertical magnetic field component. The additional heading error amounts to ap-

proximately 1◦ -2◦ per degree of tilt, depending on the geographical location. This error

source is overcome using a three-axis magnetic sensor with tilt sensors. Such a sensor
system can provide accurate heading information and also roll and pitch angles at up
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 149

to around 100 Hz with an accuracy of approximately 0.5◦ to 1.5◦ , mostly under static

conditions. However, such devices usually cost in the range of some hundreds to a few
thousand Euro.

It can be concluded that the 3-axis compass systems are too expensive for a low-cost
system while the 2-axis system may not provide sufficient accuracy under dynamic driv-

ing conditions. However, due to its ability to provide quite accurate heading information

when the tilt is small, a 2-axis compass may be an interesting option to provide initial

heading information with the non-moving platform, when GPS heading information is

not available.

Inertial Navigation System

A device providing the required attitude, heading and location information is usually

referred to as an INS.

Two types of INS implementations are typically used. In a gimbaled INS a sensor
platform holding accelerometers and gyros is mounted in a gimbal allowing to maintain
a constant attitude of the sensor platform in space, counteracting the movements of the
mounting platform containing the gimbal. In a Strapdown INS, the accelerometers and

gyros are mounted on the moving platform without a gimbal, i.e. the sensor platform
moves as the mobile platform. In the beginning, Strapdown INSs were less accurate

than the gimballed counterparts, however, with the arrival of new sensor technology this

drawback was possibly overcome. Nowadays, sensor technology is at a stage where a

Strapdown INS can be implemented using low-cost sensors.

As indicated in Figure 4.23 a typical Strapdown INS consists of three nested building
blocks [TW05]:

Inertial measurement unit (IMU) comprises the inertial instrument cluster i.e. a

number of gyros and accelerometers, providing measurements of angular rate and

specific force to measure platform movements (translational and rotational) and

CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 150

instrument support electronics for pre-processing of the sensor output (e.g., in-

strumentation amplifiers, filters, ...). The output of the IMU are typically the
compensated body angular rates and body linear accelerations.

Attitude and heading reference system (AHRS) computes from the IMU output data
the platform attitude and heading with respect to a known reference coordinate

system, e.g. the local ENU coordinate frame. This is simply done by numerical

integration and suffers in accuracy from the inevitable sensor noise. Usually,

additional sensor data need to be deployed to counteract the random walk drift.

Inertial navigation system (INS) finally obtains also the position, velocity, and accel-
eration with respect to the navigation coordinate frame (e.g. the ECEF frame).

Figure 4.23: Schematic structure of an INS (cf. [TW05]).

The sensor cluster in a typical INS design comprises 3 gyros and 3 accelerometers to
measure the body angular rates and translational accelerations. Various physical effects

can be exploited to obtain this information, resulting in a number of sensor designs

differing in accuracy and cost. A multitude of different sensor technologies are available
to sense the angular body rates or the body attitude directly. Mechanical gyroscopes

are still the most precise devices, achieving a drift stability of 0.001◦ /h, however, they

are high-cost. Other common gyro technology are the fibre optic gyros being widely

used due to its high-precision in avionic INS but also in other applications. However,
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 151

this type of sensor is also costly. In particular in the recent years, advances in micro-

machined electromechanical system (MEMS) technology allow it nowadays to use low-

cost MEMS gyros, providing sufficient accuracy for many applications. The cost for a

MEMS gyro is in the range of some Euro to a few tens of Euro. For MEMS sensors,

drift stability is reported to be in the range of 3 - 30◦ /h [TW05].

Today many INS manufactures offer low-cost INS solutions based on MEMS sensors,

how-ever, the prices for such systems still are in the range of hundreds to thousands of
Euro and thus too expensive for the target application.

Car Navigation System

Navigation systems are more commonly build in cars, today. These systems augment

the GPS heading information by wheel sensors and matching of position information to
an onboard stored street map. Therefore, the system can provide more accurate heading
information, but still update rates are not sufficient fast for heading determination (cf.
paragraph "GPS Receiver" above). Moreover, with the similar arguments for avoiding

the need of an interface to the avionic bus, also here the information may not available
at an standard, easy to access interface. Thus, information of a car navigation system is
not considered appropriate as sole input for a PAT solution.

4.6.4.3 PAT Algorithm

The PAT system implemented for this particular application and required PAT accuracy

is utilising a strong open loop PAT components based on a GPS receiver for location

determination and an IMU (cf. Figure 4.24) consisting of one MEMS. The overall
costs of the used IMU sensors are less than 25 Euros (1000-5000 pieces) plus an GPS

receiver, yielding an acceptable price.

Figure 4.25 shows the results of a measurement during a test drive (approximately 1

hour length). The absolute heading error of the low cost IMU is compared to the head-
ing obtained by a high cost INS (utilising three fiber optical gyros and three accelerom-

eters and a two antenna GPS receiver). While the platform is not moving (e.g. at the
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 152

Figure 4.24: Low cost IMU hardware.

beginning of the measurement drive) random walk of the GPS heading information can
be observed.
0
3 yaw − yaw 10
P(heading error exceeds abscissa)

LC HC
heading estimation error in deg

GPS valid flag

2 −1
10

1 −2
10
0
−3
10
−1
−4
10
−2

−3
0 500 1000 1500 2000 2500 3000 3500 0 0.5 1 1.5 2 2.5 3
time in s absolute heading error in deg

Figure 4.25: Absolute heading error time series (left) and CCDF (right).

4.7 Conclusions

Mobile satellite terminals at higher frequencies have been used for channel measure-
ments and demonstrations or within military implementations without particular focus

on optimised antenna PAT algorithms and economic aspects of the terminal costs.
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 153

The thesis provides a methodology for an optimisation of mobile satellite terminal an-

tenna and PAT system design. The optimised PAT system design includes a trade off
between antenna size versus allowed pointing error. This is done by plotting the an-

tenna gain against the off boresight angle and interpreting the acceptable off boresight

angle as an allowed PAT error. For a gaussian beam contour it is derived that the 4.3 dB
beamwidth angle states the maximum allowed pointing error for the PAT system i.e.

less PAT accuracy can be traded off with increased antenna size up to a maximum of
4.3 dB mispointing.

The optimisation of the mobile satellite terminal antenna and PAT system design also

includes the evaluation of antenna beam agility requirements. The composition of these

agility requirements of platform attitude movements and platform to satellite geometry

are discussed within the thesis. Values for attitude movements are introduced for car,
train, ship and aeronautical platforms. It is derived that angular range and acceleration
requirements are highest for car heading (and in general for smaller platforms) whereas

angular range requirements are highest for aircrafts (and in general for platforms which
are operating worldwide).

Tradeoffs between mechanically versus electronically steered antennas for high gain
antennas show that mechanically steered beam axes are still less expensive especially

if a wide angular range has to be supported. The advantages of electronically steered

axes like multiple beams and active interference suppression can only be exploited once
fully digital beam forming networks are employed. If the angular elevation range is

restricted by the operational scenario an optimised mobile satellite terminal antenna and

PAT system designs consists of a mechanical steered azimuth axis and an electronically
steered elevation axis.

Tradeoffs between open versus closed loop PAT system show that an open loop PAT

system is best suited for aeronautical applications because high accurate attitude and
position information is typically already available within bigger aircrafts. For scenar-

ios which are utilising GEO satellites the accuracy and angular PAT requirements in
CHAPTER 4. Satellite Terminal Antenna Pointing, Tracking and Acquisition 154

azimuth is typically higher due to the neighbouring satellite interference and platform

attitude dominated by heading movements. In this case the optimal solution may be a
combined closed loop open loop algorithm for azimuth whereas and open loop pointing

based on position data is sufficient for elevation control.

Chapter 5

System Dimensioning for Collective

Mobile Satellite Terminals

Satellite Spectrum and Term. Wireless Channel

Satellite Radio Link
Constellation Antenna Regulation Characteristics
Chapter 2

Mobile Satellite
Channel
Chapter 3
Mobile Platform Mobile Terminal
Characteristics Design
Chapter 4

Mobile Group Capacity and Handover

Mobile Services
Characteristics Analysis

Chapter 5

The generalised baseline for a systematic satellite system design process includes the

dimensioning of the communication systems for the services which are to be provided

by the system to the mobile user group.

This chapter presents in section 5.1 a general methodology for dimensioning of satellite

systems serving collective mobile terminals and details in section 5.2 the process at the
example of aeronautical passenger communication.

The hierarchical approach is generic for any type of communication system used by col-
lectively mobile groups and allows at any stage an adaptation of the models by exchange

of factors like service mix, user group size or terminal temporal and local distribution.

155
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 156

The analysis of the capacity requirements for one single mobile group terminal as well

as for spot beam and satellite constellation have to be considered for an optimised satel-
lite system design. Moreover, design for truly global coverage against capacity assign-

ment fitted to the market needs for collective mobile satellite terminals have to be care-

fully balanced.

Continuous service provisioning to the user is necessary for service acceptance, but sets
handover requirements especially for those systems using spot beams or being originally

designed for fixed usage.

Finally, the service usage which is on the one hand an essential basis for a capacity

requirement estimation, can also be exploited to perform a revenue estimation for the

system operation.

5.1 Methodology of Hierarchical System Dimensioning

As a starting point, a linearly scalable multi-service traffic model was introduced in

[WHS97]. Figure 5.1 depicts a schematic of this model, which calculates the aggregate
busy hour source traffic in a particular spot beam or satellite as the sum of products of

the number of users, service traffic parameters (call arrival rate; mean call holding time;
maximum bit rate; burstiness), and a very basic set of dimensionless control factors

capturing various effects such as busy hour factor, multi-service correlation, and group
terminal behaviour.

This approach was originally developed for a fixed multimedia satellite system. Multi-

ple services are characterised by applying different traffic parameters to a model based
on arrival rate and holding time. The multi-service model scales linearly with the num-

ber of users and a “multiplexing” gain by is included as burstiness factor at service

level.

For an investigation of collective mobile terminals, an enhancement of the model is

CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 157

Figure 5.1: Multi-service model schematic [HHJW04].

needed, because multi-service traffic characterisation for collective mobile terminals

has to reflect:

• the different traffic models, which are varying per service of the service mix,

• the different usage patterns and resulting traffic parameters of the services, which
are varying per group with the travel characteristics,

• the different group sizes,

• the local and temporal distribution of groups within the service area.

Introducing these aspects into the model allows characterisation of the terminal require-

ments depending on group size, a gross traffic assessment including statistical multi-

plexing effects on group, and user level and an analysis of handover requirements.

In general, the methodology follows an hierarchical and modular network dimensioning,

starting with single user traffic and ending up at the satellite constellation. Abstract
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 158

traffic models could be introduced at the different layers of the model hierarchy with

minor loss of precision, i.e., the modelling of single user web traffic can be replaced at
group or spot beam level by a model with less computational performance requirements.

sec. 5.3

Total system traffic Constellation optimisation

system

methodology: sec. 5.1.3 methodology: sec. 5.1.4

numerical example: sec. 5.2.3 numerical example: sec. 5.2.4 and 5.2.5

Satellite scenario Spot beam / satellite capacity

Spot beam / satellite traffic
Distribution of users/groups and handover requirements

satellite / spot beam

methodology: sec. 5.1.2 methodology: sec. 5.1.2

numerical example: sec. 5.2.2 numerical example: sec. 5.2.2

Terminal data throughput

User group size Mobile terminal traffic
requirement

group / terminal

methodology: sec. 5.1.1

numerical example: sec. 5.2.1

Service mix and usage model Multi-service traffic

single user

Figure 5.2: Required steps for hierarchical system dimensioning.

The complete system dimensioning process is structured in the following steps (cf. Fig-
ure 5.2):

• The identification of the applicable service mix (with associated service traffic

models per service) and the definition of service usage (traffic parameters per
service) leads to the multi-service traffic per single user.

• In a next step the single user traffic together with the user group size leads to

determination of gross traffic per collective mobile terminal. This gives input for
the terminal design and is directly setting the minimum throughput requirement

of the collective mobile terminal.

• Determination of the temporally and locally varying traffic is based on the user
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 159

group distribution. Identification of potential serving satellites and their coverage

areas and accumulation of collectively mobile networks within footprints leads to

capacity and handover analysis.

• Based on the aggregate traffic, a system optimisation for coverage layout adapted

to the market and minimised handover requirements can be established. Service

mix and usage together with the cumulative number of users leads to a revenue

estimation for the system operation.

These steps are detailed in the remainder of this section with focus on a general method-
ology applicable for all type of collective mobile groups. Section 5.2 provides numerical
examples for aeronautical passenger communication.

5.1.1 Methodology - Service Mix, Models and Usage

The mix of services which are provided to the users of a collective mobile terminal can

be classified in multiple ways, for instance by user type (passenger, crew, machine), by

direction of traffic flow (uni- or bidirectional, symmetrical, asymmetrical), by billing

type (free of charge, via existing service provider, pay per use), by distribution level
(individual, group, terminal, fleet), and many more. Table 5.1 summarises a set of

collective mobile group services classified by the typical use types “passenger” and
“mobile vehicle operation”.

For each of the service a single service traffic model may apply, e.g., Erlang model for
voice, ETSI model for web, constant bit rate for system health monitoring.

The sum of services for a collectively mobile user group result in a stochastic data mix

with various requirements on the transport links. The aggregate bit rates on up- and

downlink are certainly statistically time-dependent and might be asymmetric between

forward and return direction. Furthermore, the traffic is affected by the available bit rate

for the mobile group terminal and shaping or blocking will equalise the peak loads.
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 160

Table 5.1: Service classification by use type

Passenger Vehicle operation
Personal communication: voice, messag- Crew: navigation, vehicle and passen-
ing service, web , e-mail, ... ger management, manuals, traffic and
weather information, ...
Office: virtual private network (VPN),
voice/video conference, ... Vehicle information: system health
monitoring, maintenance data, ...
Entertainment: games, trip information,
e-magazines, video audio broadcast, Emergency: tele-medicine, tele-
video/audio on demand surveillance ...

Traffic generated and received by the users of a collectively mobile network can be

described as superposition of the traffic generated by each user according to the char-

acteristics of the desired services like data rate (symmetric, asymmetric, peak rate, sus-
tainable rate) and quality of service (QoS) parameters (delay, delay variation, loss rate).

For system dimensioning those services having potentially high influence on the peak
or average data rate of the transport link are of relevance. These are the “individual”

passenger services like voice, email, and web. Detailed traffic models of these service

types are provided in Annex D, as these models are used for later numerical analyses.

Traffic models for “non-individual” services ,e.g., such as TV, have to reflect the fact,
that the service is broadcast (and possibly cached) to all terminals independent from the

user behaviour. But, the language specific channels in a mobile scenario differ with the
operational region of single terminal or by terminals belonging to one fleet.

Services for “mobile vehicle operation” generate valid revenues. The services intro-

duced today are build upon voice and narrowband data transmission. This may indicate
that these services are of particular interest in the business case, because of creating

revenues on the one side and for their small bandwidth consumption at the satellite

transport on the other. For the data traffic estimation they are assumed to be of minor

impact, however the availability of a broader bearer service may generate new services
in this field.

Usage of the services – and thus traffic parameters of the associated single service model
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 161

– is varying per group with their travel characteristics. The travel duration may influence

the availability and acceptance of the different service types. While for short trips voice
and SMS may have been dominating the service mix, the data and internet services may

prevail on longer term travels. With the introduction of smart devices and the acceptance

of social network services this fact may have changed today. The (physiological) travel
time influences all service acceptance equally and is according to the classical daily busy

hour cycle. The different travel intention of business and leisure passengers weights the
acceptance of the services, according to their budget and interest.

5.1.2 Methodology - User Group Size and Mobile Terminal Traffic

Segmentation of the collectively mobile groups into categories of different size is nec-
essary to give an indication about the number of potential users per terminal. Baseline

for this segmentation is at a first place the vehicle size reflecting the passenger capac-
ity. In some cases a further classification in travel categories like first, second, business,
economy class is useful, as this classes indicate different travel intention. Besides the
passenger capacity also the aircraft load factor of the mobile vehicle has to be taken into

consideration for determination of the potential user number. Table 5.2 gives a typi-

cal segmentation of aeronautical, maritime, and railway with their associated passenger
capacity.

At this level the hierarchical dimensioning (cf. Figure 5.2) allows an identification of
the per terminal traffic and gives input for the terminal design as a minimum overall

throughput requirement of the terminal. The aggregate group terminal traffic can be

determined by simulation of the mix of single service traffic models taking usage pa-
rameters and groups size into account.

By simulating the superposition of single user traffic per service, the multiplexing ef-

fects on the group traffic is taken into consideration. A numerical example, e.g., for
multiplexing gain of aggregate voice traffic is given in Annex D.3 showing that the ac-

cumulated voice data rate which is not exceeded with a probability of 99.9% is 96 kbps
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 162

Table 5.2: Example vehicle categories and size

Aeronautical type of vehicle approx. mean passenger capacity
Jets 50
Single aisle 135
Widebody 270
Jumbo 500
Train type of vehicle approx. mean passenger capacity
Commuter 600
Long Distance 1200
High speed 800
Maritime type of vehicle approx. mean passenger capacity
Cruise ships 1500
Ferries 600
Yachts 20
Fishing, Cargo 5

for a group of 100 users and only 297.6 kbps for a group of 500 users assuming a user
activity of 0.033 Erlang.

5.1.3 Methodology - Satellite Scenario and User Group Distribu-

tion

Unlike user distribution models for terrestrial traffic dimensioning, the users of a collec-

tive mobile terminal are part of the network for a limited time and are forming a mobile

group. Thus, the temporal and local distribution of the collective mobile group as well
as gross number of users – or more precisely, the timeshare the users stay within the

satellite – is of relevance for system dimensioning.

The distribution of groups is important in order to evaluate the need of a carrier assign-

ment in a certain region as well as the cumulative passenger travel time for the multi
service traffic analysis.

Valuable input for numerical evaluation of the user distribution are the regular schedules

of public transportation systems. These schedules are available as large area databases
for aeronautical, train, ship or even intermodal travels, because they are used since a

long time for journey planning.

CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 163

A set of information is typically associated with each single travel of a journey planner

database, such as time and location of departure and arrival as well as information about
the vehicle, e.g. identifier, category, and size.

Based on this input, the temporal and local distribution can be derived as a 3-
dimensional matrix with the dimensions latitude, longitude, and time (altitude is of no
relevance and thus neglected); in order to determine at some time instant the position

of a vehicle serving a given travel, it is assumed, that the vehicles follow the shortest

route between the connected departure and arrival locations, i.e., along a great circle,

and travel at constant speed, which is determined by the total travel time and the great
circle distance. This introduces some inaccuracy into the determination of the vehicle

position since the actual route deviates from the theoretical one (great circle).

For the aeronautical case the assumption introduces very little inaccuracy in comparison

to satellite spot beam sizes. Figure 5.3 shows a comparison between theoretical and
actual flight routes, indicating that the deviation from the theoretical shortest route may
amount to a few hundreds km, while the spot beam size of today operational satellite

systems is minimum 800 km (cf. Figure 2.5).

Figure 5.3: Theoretical aeronautical (great circle) routing (left). Actual routing (right). [Eur05]

For railway, direct interconnection between intermediate stops of a travel route – which

can be typically extracted from the journey databases – gives also sufficient resolution
with respect to typical satellite spot beam sizes. For the maritime scenario a dedicated

model has to be applied which calculates the shortest route between two stops based on
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 164

the precondition that at no time the ships can pass landmass.

An array of travels is associated to each element of the distribution matrix. Each trip
is uniquely identified, e.g., by a flight number, and the associated vehicle information,

which may lead to classification of the travel group.

Mapping of the local distribution to certain regions can be applied in a next step and

allows an analysis in total or per group classification, per spot beams, satellite coverage,

or any kind of other area, like landmass, national and international water, air traffic

surveillance zones, and many more.

5.1.4 Methodology - Capacity and Handover Analysis

Figure 5.4 summarises the system dimensioning methodology as a schematic as imple-

mented for simulation for the numerical example of communication for aeronautical

passengers detailed in the next section. The simulation is divided into a pre- and post-
processing.

• Preprocessing steps are:

– the simulation is starting with the segregation of a journey planner database

into group characteristics and temporal and local group distribution matrix;

– the traffic models and associated usage parameters are applied to the chosen
group segmentation;

– a region mapping to satellites and spot beam is applied to the temporal and

local distribution matrix.

• Postprocessing, allows analysis of

– the number of active terminals by satellite and spot beam;

– the aggregate traffic of each satellite and spot beam;

CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 165

Journey planner database

Departure Location
Time

Arrival
Travel ID Location
Time

Type
Vehicle
Number
Fleet

Group characteristics Time and local group distribution

PAX Vehicle Duration

Travel ID Fleet …
number type category
lat
1 135 SA SH A tid1,1,t… tid1,long,t
2 278 WB LH B tid1,1,1 … tid1,long,1

... . .

...
.. .
3 455 JB LH A

...

...
M lat,long,t=
4

.
409 JB LH C tidlat,1,t… tidlat,long,t
5 254 WB SH C tidlat,1,1 … tidlat,long,1
6 140 SA SH A time
7 270 WB LH B
... ... ... ... ...
SH: short haul; LH: long haul long
SA: single aisle; WB: wide body; JB: jumbo

Service parameters Region mapping

Group class service parameters lat

SH voice 20%, data 10%
LH voice 20%, data 20%
... ...
Sat1 Sat2 Sat3

Capacity and Handover Analysis

long

Figure 5.4: Mobile group system dimensioning schematic.

– the terminal handovers from satellite and spot beam point of view;

– the satellite and spot beam handovers from terminal point.

CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 166

Travel ID 1: C 1, N1 Travel ID 2: C2 , N2 Travel ID M: C M , NM

L C1 L C2 L CM

Aserv1,C1 Aserv2,C1 A servK, C1 Aserv1,C2 Aserv2,C2 A servK, C2 Aserv1,CM Aserv2,CM AservK, CM

Fserv1 Fserv2 FservK Fserv1 Fserv2 FservK Fserv1 Fserv2 FservK

A servK,CM Acceptance Factor of Service Type K for Group Class C M

FservK Traffic Model for Service Type K
LCM Load factor for User Group Class C M
N1...M User Group Size Time Series of Cumulative Bit Rates
C1...L User Group Class

Figure 5.5: Data traffic simulation schematic.

Figure 5.5 details the simulation for generation of the commutative bit rates for the
capacity analysis. For each travel ID the associated group size and group class is known.

The load factor and single service acceptance per group class is applied. The activity
and bit rate series of each single traffic model can be generated and summed for a group

class (or each single travel ID if this level of detail is desired), spot beam, satellite and
system. Multiplexing gain can be applied on all layers of the simulation. But, this would

assume ideal multiple access on each layer.

For the later numerical example the simulation assumes, that multiplexing gain can be

achieved on group (cf. Figure 5.6 of the numerical example) and on spot beam level (cf.
Figure 5.14 of the numerical example). The satellite bit rates are generated by summing

up the cumulative spot beam bit rates.

Adding information about pricing to each service the evaluation can be extended to a
revenue estimation by multiplication of the service usage, the number of total users, and

price per service. A revenue analysis for passenger AirCom considering a service mix
of voice and data using the model given within this section is given in [BWH03].
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 167

5.2 Numerical Example - Aeronautical System Dimen-

sioning

Communication for passengers of an aircraft represents the most evident example for

a satellite connection to a closed mobile user group. The wide operational area, the

fairly friendly channel properties, and the required data rate calls for satellites at higher

frequencies and the distribution of the services via a multiple of wireless and wired

standards to passengers and crew.

5.2.1 Aeronautical Example - Service Mix, Models and Usage

5.2.1.1 Service Mix

Typically aeronautical communication services are classified in four groups, referring

to the RTCA/DO215 specification of Aeronautical Mobile Satellite Services:

Table 5.3: Regulatory Aeronautical Service Classification

Cockpit services Cabin services
Air Traffic Services (ATS): Airline Administrative Communications (AAC):
air traffic control, flight information cabin provisioning and inventory, seat assign-
services, alerting services ments, passenger travel arrangements
Airline Operational Control (AOC): Airline Passenger Correspondence (APC):
crew communication with the air- personal passenger communication, e.g. voice,
line operation center web, email ...

ATS, AOC and AAC will remain to be more or less narrowband services while APC will

require broadband communication systems, to ensure similar communication usability

for aircraft passengers like in terrestrial mobile networks.

The services available for aircraft passengers and crew are considered as a subset of

the list of services detailed in Table 5.1. Market studies verified that telephone, web,
and email services are the most likely accepted services for passengers and it has been

shown that these services will dominate the transport link capacity requirements.
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 168

5.2.1.2 Traffic Models

The traffic models and parameters used throughout the following numerical studies for
the three most important user services (voice, web browsing, and email) are detailed

in Annex D. For the sake of tractability and simulation performance, system feedback

effects such as user reaction to blocking, delays, load situation and throughput are ne-

glected, and higher-layer protocol influences on real traffic – especially the effects of
the various satellite transmission control protocol (TCP) solutions – are not taken into

account in the modelled traffic.

For each service a packet level and call/page/email event level model is detailed and

compared with respect to accuracy. Packet level simulation is useful, e.g., in assessing

the efficiency of encapsulation schemes or if packet scheduling for QoS provision shall
be investigated. Of course, this level of detail would also be desirable for investigating

the capacity requirements, both on a per terminal level, as well as on the satellite system
level. However, considering the large number of terminals, the little accuracy gained
in comparison to call/page/email event models is not justifying the computational com-
plexity of simulating the traffic carried in the satellite system on packet level.

Thus, the three relevant user services are modelled according to the following level of

detail.

• Each voice call is characterised by the call arrival time (user accepts or initiates

call), the call duration, and a constant bit rate.

• Each email download/upload is characterised by the time the user initiates the

download/upload and the email volume.

• Each web session is characterised by the session start time and the page down-

loads within the active session. The page downloads are in turn characterised by
the number, and the time instances of page downloads and, finally, by the volume

of each page.
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 169

By modelling the services on this level it is possible to obtain reasonably accurate ca-

pacity estimations including possible statistical multiplexing gains due to the statisti-
cal access to the services, while the high complexity of the packet level simulation is

avoided.

Depending on the type of evaluation, these traffic parameters might be adjusted carefully

with a pessimistic acceptance for worst case business case modelling vs. an optimistic

acceptance for worst case capacity dimensioning.

For web services the outdated models of ETSI/UMTS [ETS98] and Paxson [Pax94]
have been adapted accordingly. In order to guaranty service quality comparable to

terrestrial mobile services the QoS performance requirement recommendation of ITU

(summarised in Table 5.4 for the services of relevance) is taken into consideration.

Table 5.4: Application QoS performance requirements (from ITU1010)

Service Data amount Mean delay Average bit rate

WWW 10 kbyte a <2-4 s/page 20-40 kbps
Email 10 kbyte b <2-4 s/email (POP/IMAP) 20-40 kbps (POP/IMAP)
minutes/email (SMTP) 1 kbps (SMTP)
Voice N/A <150-400 ms c 4-64 kbps
a
An average page size of 120 kbyte is assumed in [BUWH04].
b
Obviously, emails with large attachments are not considered here.
c
For a satellite link 350 ms can be assumed as the maximum delay (one-way).

In line with this recommendation an average bit rate of 24 kbps for web services is con-
sidered as sufficient. Today web service providers typically use the contention ratio as

service measure. The contention ratio describes the number of users sharing the over-
all transport link capacity. Values between 1:20 and 1:100 are typically describing the

share of a digital subscriber line (DSL) (2 Mbps) link capacity among users. Adopting

this model would mean, that the ratio of transport link capacity dedicated for “Internet

Access” to the number of users of a mobile group accepting the service should be in the
same order of magnitude.

A group terminal with 500 kbps satellite return link shared by a group of 140 passengers

of which 10% are accepting the service in a SH flight leads to 35 kbps mean data rate.
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 170

Respectively, a group terminal with 1 Mbps satellite return link, shared by a group of

270 passengers of which 20% are accepting the service in a LH flight leads to 19 kbps
mean data rate.

No proven values are available yet, for modelling of Erlang usage for mobile telephony
in an aircraft. The definition of Erlang generated by one single user during a flight gives

the percentage of time this user is making phone calls related to the time being on the

plane. Note that due to the overall scaling of service acceptance only 16% of cumulative

passenger flight hours are baseline for the following numerical analysis.

A reasonable example is the assumption that in average one user accepting the service is

making (incoming or outgoing) one call with average length of 180 seconds per service

availability period of an average short and medium haul flight (SH) flight. This results
in 0.033 Erlang per user accepting the service.

5.2.1.3 Service Usage

Data traffic modelling as described in section 5.2.4 requires a usage model for the single
services available to the passengers in the aircraft. This service usage is influenced by
several factors.

Flight Class and Intention A clear way of segmenting the market for AirCom ser-

vices is the consideration of the class the passengers fly and the scope of their journeys,
i.e., according to the passengers that have the requirement for and the economic capacity

to purchase a service of the AirCom ones.

In [BWH03] a model assumes that distinguishes the service acceptance of all business
and first class passengers in comparison to economy passengers. But in essence, there is

no noticeable effect on the system dimensioning process. Summing up over the typical

cabin layouts for different A/C types results in a more or less equal scaling of service
acceptance of the overall A/C capacity.
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 171

Flight Time Also the psychological flight time is influencing the service usage. For

SH flights the services will be influenced accordingly to the availability of communi-

cation partners and activity in the cabin. For SH flights the according terrestrial daily

utilisation phases is applicable.

It is more likely that passengers flying from East to West on long haul flight (LH) flights

will make use of communication services, than passengers flying from West to East,

who would tend to sleep during most of the flight duration in order to minimise the

jetlag effects. Connexion by Boeing reported during the first service test flight early

2003 on average 80 users on the westbound, but 50 users on the eastbound flights 1 .

The physiological flight time will be considered for the service assessment based on the

daily distribution of flights.

Flight Phase Assuming that cabin communication services are only available at cruise

level, only a certain percentage of flight time will be used for the service offer. Firstly,
the time periods after take-off and before landing during which no satellite service is
offered, the service off-times, are identified. Generally, each flight profile can be divided
into about 7 phases: taxi-out, take-off, climb, cruise, descent, landing, taxi-in (cf. Figure

3.26).

It is assumed that communication services are offered in an aircraft only during times

at cruise altitude. Typically it takes 30 min after take-off to reach cruise altitude and,

further, cruise altitude is left 30 min before landing. A service off-time of two times
30 minutes at start and landing will be subtracted from flight duration to obtain service

availability time.

Flight Duration Looking to the support of aircraft passenger communication, differ-

ent categorisation for flight duration, like continental, intercontinental, very short or
1
During the test period the service was offered for free, thus, these figures may not reelect the accep-
tance of the commercial service
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 172

commuter flights, short haul, medium haul, long haul may be applied to the flights.

The usage model and thus the relating aggregate traffic mix is influenced by the flight
duration. This prevents a pure multiplication of one aggregate traffic mix with the ac-

cording flight duration. In general data services may become more dominant during

intercontinental flights.

A differentiation in the two categories SH and LH with a limit of 3.5 hours of flight
duration seems to be the most suitable. Very short haul flights of less than 60 minutes

are not of main focus for broadband communication services.

For the further numerical studies it will distinguish between medium/short and long haul

flights, with the following mean duration 2 and influence on the usage of services:

• short haul (SH): 2.5 hours; voice, messaging, and email may prevail data/internet
services; some services are not part of the service mix;

• long haul (LH): 7 hours; data/internet services may become more dominant and
are scaled accordingly in the usage pattern.

To respect the influences of the travel characteristics for AirCom, different service usage
due to flight duration will be taken into consideration. The acceptance of the service mix

by the passengers is assumed to be 16% for voice services and 10% for internet services
in SH flights and 16% for voice services and 20% for internet services in LH flights. No
differentiation due to flight class and intention is applied in this numerical example.

5.2.2 Aeronautical Example - User Group Size and Mobile Termi-

nal Traffic

5.2.2.1 User Group Size

Different categorisations for A/C types are applicable, e.g., single aisle, narrow-body,

widebody, jumbo, long range, short range, extended range, high capacity, twin-deck.
2
Please note that for later numerical studies the flight duration is not based on the here mentioned
average, but on the duration of each single flight.
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 173

For the further study the three different A/C types single aisle A/C (SA), widebody

A/C (WB), and jumbo A/C (JB) will be distinguished. The total seating capacity of
A/Cs is differing by cabin layouts of the different airlines.

The considered aircraft categories 3 are:

• single aisle (SA): ≤ 140 seats

• wide-body (WB): 140 < seats ≤ 370

• jumbo (JB): > 370

Additionally also the average load factor has to be considered to estimate the address-
able number of passengers per A/C. For later calculations an average load level of about

70% will be taken into consideration.

This value is based on publicly available statistics from several airlines (examples:
Lufthansa 72% in 2003; British Airways 73,1% in 2003, British Airways 73,5% in
2002.) No differentiation between load levels for LH and SH flights is applied, even
if some publications indicate, that load levels on long haul flights tend to be higher

compared to short haul.

5.2.2.2 Mobile Terminal Traffic

Using the traffic models described in Annex D, the terminal bandwidth requirements for

a single aircraft can be determined by simulation. This is done according to four typical

categories i) short/medium haul single aisle (SHSA), ii) short/medium haul, wide body

(SHWB), iii) long haul, wide body (LHWB), iv) and long haul, jumbo (LHJB).

Figure 5.6 shows the CCDF of the aggregated downlink data rate for the four categories.

Table 5.5 summarises the aggregate mean bit rate and the data rate per aircraft/flight

class which is not exceeded with a probability of 99.9%.

3
Please note that for later numerical studies the number of passengers per aircraft is not based on the
here mentioned average per A/C category, but on the seating layout per single A/C type.
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 174

Downlink Group Datarate

0
10

P(data rate exceeds abscissa)

−1
10

−2
10

−3 SHSA SHWB LHWB LHJB

10

−4
10
0 50 100 150 200 250 300 350 400 450
data rate (to A/C) in kbps

Figure 5.6: Comparison of total downlink data rates (voice, email, web) for the aircraft/flight
class SHSA (red), SHWB (green), LHWB (blue) and LHJB (pink) (dashed: mean values).

Table 5.5: Aggregate mean bit rate and data rate at P=0.1% per aircraft/flight class
Bit rate in [kbps] to A/C to A/C from A/C from A/C
Scenario mean at P=0.1% mean at P=0.1%
Short haul, single aisle 23 150 7 45
Short haul, wide body 46 190 14 60
Long haul, wide body 80 270 18 68
Long haul, jumbo 162 400 37 100

5.2.3 Satellite Scenario and User Group Distribution

5.2.3.1 Satellite Scenario

Baseline for the further numerical studies of the aeronautical system dimensioning is
a 3 GEO satellite system with satellites at a regular 120◦ orbital spacing. This is the

most commonly used constellation to provide nearly – neglecting the poles – global

coverage. The three satellites will be named and located as follows: satellite #1 at
20◦ E, satellite #2 at 100◦ W (or -100◦ E), and satellite #3 at 140◦ E. The coverage of this

reference constellation is shown in Figure 5.22.

The spot beam pattern of the reference satellite antenna consists of 233 beams arranged
at 8 concentrical rings adding additional spot beams of the 9th ring to serve the outer
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 175

regions. Figure 5.7 depicts the spot beam pattern, the spot beam numbering and the

transformation of the pattern onto the earth surface for satellite #2.

179
221 217 180 230
220 216 148 181 231 80
219 215 139 149 182 232
218 214 138 106 150 183 233 221
213 137 127 107 151 184 217 180
212 136 126 74 108 152 185 60 220 231
219 216 148 181 232
211 135 125 91 75 109 153 186 215 182
218 139 149
196 134 124 90 48 76 110 154 187 214 183
138 106 150
163 123 89 60 49 77 111 155 40 213 137 127 107 151 184
197 112 88 61 29 50 78 105 195 212 136 126 74 108 152 185
211 135 125 91 75 109 153 186
164 79 59 30 28 51 73 162 196 134 124 90 48 76 110 154
198 113 52 31 14 27 47 104 194 163 123 89 60 49 77 111 155
20 197 112 88 61 29 50 78 105 195
165 80 32 15 13 26 72 161 164 79 59 30 28 51 73 162
199 114 53 16 5 12 46 103 193 198 113 52 31 14 27 47 104 194
165 80 32 15 13 26 72 161

lat. in deg
166 81 33 6 4 25 71 160 199 114 53 16 5 12 46 103 193
166 81 33 6 4 25 71 160
200 115 54 17 1 11 45 102 192 0 200 115 54 17 1 11 45 102 192
167 82 34 7 3 24 70 159
167 82 34 7 3 24 70 159 201 116 55 18 2 10 44 101 191
201 116 55 18 2 10 44 101 191 168 83 35 19 9 23 69 158
202 117 42 36 8 22 43 100 190
168 83 35 19 9 23 69 158 169 67 41 37 21 56 68 157
20 203 98 66 40 20 58 84 99 189
202 117 42 36 8 22 43 100 190 147 97 65 39 57 85 118 156
146 96 64 38 86 119
169 67 41 37 21 56 68 157 178 145 95 63 87 120 129128
203 98 66 40 20 58 84 99 189 177 144 94 62 121 130 204
176 93 122 131 205
147 97 65 39 57 85 118 156 40 175 143 206
174 142 92 132 207
178 146 96 64 38 86 119 128 188 229 141 133
173 208
177 145 95 63 87 120 129 204 228 223
172 140 209
176 144 94 62 121 130 205 60 227 224
175 143 93 122 131 206 171 210
226
229 174 142 92 132 207 222
228 173 141 133 208 223
227 172 140 209 224 80
226 171 210 225
170 180 160 140 120 100 80 60 40 20
lon. in deg

Figure 5.7: Regular cell pattern and beam numbering (left; spot beam pattern of satellite #2 at
100◦ W (right).

5.2.3.2 Aeronautical Example - User Group Distribution

The basis for the investigation is a database of all (i.e. worldwide) scheduled flights for

May 2002. One day (Wednesday, 1. May), is considered which is a reasonable simpli-

fication for the analysis, since it has been verified that the traffic volume does not vary
much among different week days (weekends show lower numbers). The investigation is

based on great circle routing with immediate HO to nearest satellite.

Figure 5.8 shows the temporal distribution of aircraft in flight over universal time clock

(UTC) comparing the effects of service availability time. The effect of service off-time

will be reflected by the wording aircraft in service.

Besides the local and temporal distribution also the total number of aircraft is of interest,

e.g., for evaluation of installation costs per terminal. Today there are approximately
9000 SA, 2100 WB, and 900 JB (currently Boeing 747 only) aircraft operational for

commercial passenger traffic.

CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 176

no service off time 2x30 min service off time

4500 SA 4500 SA
WB WB
4000 JB 4000 JB
3500 3500
number of A/C in flight

number of A/C in flight

3000 3000

2500 2500

2000 2000

1500 1500

1000 1000

500 500

0 0
0 5 10 15 20 0 5 10 15 20
UTC time in h UTC time in h

Figure 5.8: Number of A/C in service vs. UTC time. Left: 0 min service off-time. Right: 2x30
min service off-time.

The two different values for user group distribution within the satellite reference con-

stellation which are needed for a numerical analysis are: i) the distribution of aircraft

in order to evaluate the need of a carrier assignment in a certain region, and ii) the
cumulated passenger flight time for the multi service traffic analysis.

Firstly numbers are given with no restriction in respect to minimum flight duration.
The passenger flight hours (PFH) are calculated by summing up the number of seats in
aircraft multiplied by the load factor of 70% and the flight duration in hours. Table 5.6

shows the target market of a total of 83336 flights with 15.84 million of passenger flight

hours per day. A pronounced focus of domestic flights can be observed with nearly

40000 continental US and 30000 continental European flights.

Table 5.6: Total market size for aeronautical traffic

Total number of flights per day 83336
Total number of airlines serving these flights 662
Largest 20 airlines (w.r.t. flight number) serve 45528 flights per day
Largest 20 airlines (w.r.t. flight number) serve 54% share of global
Total global passenger flight hours per day 15.84 million
Largest 20 airlines (w.r.t. PFH) serve 9.3 million PFH
Largest 20 airlines (w.r.t. PFH) serve 59% share of global

In the aeronautical case the service unavailability time during ascent and descent has
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 177

to be considered. Thus, only flights lasting longer than 1 hour are considered in the

following. Table 5.7 summarises the aeronautical target market with a total of 59114
flights and 14.8 million of PFHs per day. It can be realised that this reduction in service

availability time results in a large reduction of very short haul flight, whereas the PFH

are dominated by the short/medium and long haul flights.

Table 5.7: Total group size for aeronautical traffic accounting only flights > 1 h
Total number of flights > 1 h per day 59114
Total number of airlines with flights > 1 h 559
Largest 20 airlines (w.r.t. flight number) serve 36671 flights per day
Largest 20 airlines (w.r.t. flight number) serve 62% share of global
Total global PFH per day 14.8 million
Largest 20 airlines (w.r.t. PFH) serve 8.9 million PFH
Largest 20 airlines (w.r.t. PFH) serve 60% share of global
histogram (bin size is 100 km)

8000
histogram (bin size is 0.25h)

10000 6000

4000
5000
2000

0 0
0 5 10 15 0 5000 10000
flight duration in h travelled distance in km

Figure 5.9: Histogram of flight duration (left) and flight distance (right) for flights > 1h.

The consideration of a 2x30 minutes service off-time has a major impact on the number

of instantaneous aircraft in service, especially on the single aisle aircraft. Without ser-
vice off-times, the peak is around 4450; while with service off-times, the peak is around

2200. Figure 5.9 shows the histogram of flight durations and travel distance for flights

longer one hour. The majority of flights are less than two hours duration and 1000 km

travelled distance. The peak number of aircraft per spot beam during a day is shown in
Figures 5.10 (satellite #1), 5.11 (satellite #2), and 5.12 (satellite #3).

It can be observed, that the number of aircraft/passengers over spot beams is extremely
unbalanced. The spot beams of satellite #2 which are serving parts of CONUS (i.e. spot
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 178

Figure 5.10: Peak number of A/C in service per spot beam for satellite #1.

Figure 5.11: Peak number of A/C in service per spot beam for satellite #2.

beam #107, #108 #150 and #151) are heavily loaded while large parts of the Pacific

Ocean are not crossed by a flight route, and some spot beams exist where no capacity
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 179

Figure 5.12: Peak number of A/C in service per spot beam for satellite #3.

assignment at all is needed. The peak number of aircraft in service per spot beam is 285
in satellite #1, 535 aircraft in satellite #2 and 64 aircraft in satellite #3.

5.2.4 Aeronautical Example - Data Traffic Analysis

Based on the traffic models given in the Annex D, the down- and uplink data require-
ments on satellite and spot beam level are derived in the following sections. Above

findings and remarks on asymmetric demand over satellites and spot beams remain in
principle valid for data traffic and bit rates. But, the different weights on data and voice

usage in SH and LH flights are influencing the traffic loads of the spot beams which

serve distinct LH routes.

5.2.4.1 Per Satellite

The total traffic within the satellite of the reference scenario is shown in Figure 5.13 for

the down- and uplink. Due to the asymmetric web and email and the symmetric voice
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 180

characteristics, the downlink traffic is dominated by web browsing and the uplink traffic

is dominated by voice. The peak traffic demand values for each satellite are summarised
in Table 5.8.

Table 5.8: Peak traffic demand per satellite.

Satellite Peak Down Mbps Peak Up Mbps
#1 96.57 26.76
#2 155.55 42.36
#3 80.13 22.35

all airlines, 2x30 min service off−time all airlines, 2x30 min service off−time
50
160 sat 1 sat 1
sat 2 sat 2
140 sat 3 40 sat 3
downlink bit rate in Mbit/s

uplink bit rate in Mbit/s

120

100 30

80
20
60

40
10
20

0 0
0 5 10 15 20 0 5 10 15 20 25
UTC time in h UTC time in h

Figure 5.13: Total up link (left) and down link (right) data traffic per satellite.

With equation 2.16 and applying a frequency reuse of cluster size K = 7 to the spot

beam number of nsb = 233 and a transponder bandwidth of 30 MHz (according to

Figure 2.13, 72 MHz downlink and 120 MHz uplink are allocated to all GEO systems

providing mobile services) the total frequency resources of one satellite is theoretically
1 GHz. Assuming an efficiency ηmodcod of 1 bit/sec/Hz such a satellite would reach a

theoretical upper cluster capacity limit of 30 Mb/s and satellite capacity limit 1 Gb/s

(cf. equation 2.16). An Inmarsat I-4 satellite incorporates a total of 588 channels with
200 kHz accumulating to a total of 117.6 MHz.
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 181

5.2.4.2 Per Spot Beam

The down- and uplink data requirements per spot beam over time is shown in Figure
5.14 and Table 5.9 summarises the peak data rate of the worst case spot beams.

sat 1, all airlines, 2x30 min service off−time, downlink sat 1, all airlines, 2x30 min service off−time, uplink

220 220
8
200 200
7 2
180 180
160 6 160
spot beam index

spot beam index

140 140 1.5
5

Mbit/s

Mbit/s
120 120
4
100 100
1
80 3 80
60 60
2
0.5
40 40
1
20 20
0 0
0 5 10 15 20 0 5 10 15 20
UTC time in h UTC time in h

sat 2, all airlines, 2x30 min service off−time, downlink sat 2, all airlines, 2x30 min service off−time, uplink

220 220
14 4
200 200
180 12 180 3.5

160 160 3
spot beam index

spot beam index

10
140 140
2.5
Mbit/s

Mbit/s
120 8 120
100 100 2
6
80 80 1.5
60 4 60
1
40 40
2 0.5
20 20
0
0 5 10 15 20 0 5 10 15 20
UTC time in h UTC time in h

sat 3, all airlines, 2x30 min service off−time, downlink sat 3, all airlines, 2x30 min service off−time, uplink

220 220 1
3.5
200 200
0.9
180 3 180
0.8
160 160
spot beam index

spot beam index

2.5 0.7
140 140
0.6
Mbit/s

Mbit/s

120 2 120
0.5
100 100
1.5 0.4
80 80
60 60 0.3
1
40 40 0.2
0.5
20 20 0.1

0 5 10 15 20 0 5 10 15 20
UTC time in h UTC time in h

Figure 5.14: Peak bit rate per spot beam downlink (left) and uplink (right) for sat #1 to #3.
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 182

Table 5.9: Peak traffic demand in worst case spot beam per satellite.
Satellite Peak Down Mbps Peak Up Mbps Beam ID and UTC
#1 8.70 2.13 215 at 10:45
#2 15.63 4.48 151 at 22:45
#3 3.65 0.90 91 at 10:30

The ten most loaded beams carry more than 50% of the total satellite traffic. Such

pronounced peak loads on spot beams may turn out to be the limiting capacity bottle-
necks and have to be taken into consideration for resource assignment capabilities of

the satellite system. Much more as these pronounced peak loads are occurring for the

aeronautical example in neighbouring spot beams.

Two effects limit the dynamic assignment of capacity resources to spot beams i) the
bandwidth available for the satellite, respectively cluster size, and ii) technological con-
straints of the satellite like channelisation granularity and total channel number avail-

able.

Even if the available total capacity can be increased with frequency reuse, the associated

cluster capacity is limited by the available bandwidth. The total spot beam capacity is
determined by the cluster capacity divided by cluster size. Assuming the same values
as given above for the satellite considerations, this results in a spot beam capacity of

approximately 4 Mbps. Unequal allocation of the frequency resources to spot beams

of one cluster can be used to increase the capacity of one spot beam, but restricts the
allocation of the resources within the different clusters.

5.2.5 Aeronautical Example - Handover Analysis

As the GEO systems used today for mobile broadband satellite communication do not

allow seamless satellite handover, minimisation of HO occurrence is one of the opti-

misation goals for service acceptance from user perspective. Determination of the spot
beam HO numbers is even more of importance for calculation of the signalling overhead

and ensure capacity dimensioning with minimised service dropping or blockage.

CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 183

5.2.5.1 Satellite Handover

Handover between satellites without interruption of the ongoing applications is not im-
plemented for any GEO satellite system today. A satellite handover involves change of

pointing of a directed antenna, acquisition of the new satellite signal, modem synchro-
nisation and typically an IP layer HO for the applications. IP layer HO can be realised

by switched IP tunnels or border gateway protocol. Switched IP tunneling is faster than

border gateway protocol and could remain transparent IP addresses for the application,

but introduces an intense overhead to the scarce satellite bandwidth. Handover dura-

tion amounts in a range of some minutes and typically brakes the connection of the

applications.

A minimisation of satellite HO occurrence is therefore the most vital design goal for a
satellite system design for service acceptance.

Satellite perspective The number of satellite handovers over time is shown in Figure

5.15. The values are given for time periods of 15 minutes and are shown for all satellite
combinations in both directions. Table 5.10 summarises the peak and sum of handover
between the satellites.

Especially for the handovers between satellite #1 and satellite #2 which reflect the inter-

continental Atlantic flights the temporal distribution of east- and westbound flights can
be verified.

Table 5.10: Number of handovers between satellites of the reference constellation.

Satellite Peak number of handovers Total sum of handovers per day
#1 ⇒ #2 35 / 15 min 549
#2 ⇒ #1 27 / 15 min 551
#1 ⇒ #3 12 / 15 min 218
#3 ⇒ #1 6 / 15 min 203
#2 ⇒ #3 9 / 15 min 95
#3 ⇒ #2 8 / 15 min 128
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 184

Number of aircrafts moving between satellites 1 and 2

40
(1−>2)
(2−>1)
20

0
0 5 10 15 20
Time (hours)

Number of aircrafts moving between satellites 2 and 3

10
(2−>3)
5 (3−>2)

0
0 5 10 15 20
Time (hours)

Number of aircrafts moving between satellites 3 and 1

15
(3−>1)
10 (1−>3)

0
0 5 10 15 20
Time (hours)

Figure 5.15: Number of satellite handovers between satellites of the reference constellation in
15 minutes time periods.

Terminal perspective Maximum one satellite handover per aircraft during a flight

is occurring as flight range is less than twice the distance between the intersection of
coverage areas of a 3 GEO constellation and great circle routing is assumed. Of all
flights per day only 1744 are affected by a satellite handover, which seems to be an

acceptable number.

5.2.5.2 Spot Beam Handover

Satellite perspective The total number of spot beam handovers for each satellite is
given in Figures 5.16 and 5.17. Again a 15 min time period is used for the simulation

and only inbound HOs are taken into account as simulation validate similar values for

in- and outbound traffic.

Likewise to the unbalanced distribution of aircraft discussed above, a pronounced focus

of spot beam handovers during the daylight hours of the associated satellite coverage
can be seen between the spot beams which are loaded by aircraft/passengers (e.g. 6:00

to 21:00 UTC for beam 139 of satellite #1, which is serving Europe; cf. Figure 5.10).
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 185

Sum of spotbeam handovers satellite 1

300
200
100
0
0 5 10 15 20
Time (hours)
Sum of spotbeam handovers satellite 2

500

0
0 5 10 15 20
Time (hours)
Sum of spotbeam handovers satellite 3
200

100

0
0 5 10 15 20
Time (hours)

Figure 5.16: Total number of inbound spot beam handovers for each satellite of the reference
constellation in 15 minutes time periods.

Satellite 1 total handovers per spot beam Satellite 2 total handovers per spot beam
40
200
80
spot beam index

30
150 60

100 20
40

50 10 20

0 5 10 15 20 0 5 10 15 20
UTC time (hours) UTC time (hours)

Satellite 3 total handovers per spot beam

14
200
12
spot beam index

150 10
8
100 6
4
50
2

0 5 10 15 20
UTC time (hours)

Figure 5.17: Number of inbound spot beam handovers per spot beam for each satellite of the
reference constellation in 15 minutes time periods.
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 186

The peak number of spot beam handovers related to the worst case beam and the total

number of spot beam handovers for each satellite is shown in Table 5.11. The total rate
of spot beam handovers is up to 98 per 15 minutes for the worst case beam and 645 per

15 minutes for the worst case satellite. The total number of spot beam HO during a day

is in the range of the total number of flights.

Table 5.11: Peak number of spot beam handovers per satellite.

Satellite Peak number of Beam ID and Total number of Total number of
spot beam han- UTC spot beam han- spot beam han-
dovers in worst dovers per satel- dovers per satel-
case beam lite per 15 min lite during 1 day
#1 43 / 15 min 139 at 05:45 276 / 15 min 16531
#2 98 / 15 min 150 at 00:00 645 / 15 min 32398
#3 15 / 15 min 124 at 07:15 197 / 15 min 10025

Terminal perspective Figure 5.18 shows the spot beam handover occurrence from

the aircraft perspective for all flights and the associated HO rate.

30.000 5000

25.000
Histogram (bin size 5 min)

4000

20.000
Histogram

3000
15.000
2000
10.000

1000
5.000

0 0
0 5 10 15 20 0 0.5 1 1.5 2
Number of HO occurrence from terminal perspective HO rate from terminal perspective in HO per h

Figure 5.18: Histogram of spot beam handover occurrence (left) and handover rate in HO per
hours (right) from terminal perspective.

Spot beam HO occurs only during service on time, i.e, the service-on time is 1 hour less

than the flight duration and only flights longer one hour are taken into consideration.

Many flights exist with a duration up to two hours and a flight distance of less than
1000 km (cf. Figure 5.9). These flights in service may stay within one spot beam. This

effect can be seen in Figure 5.18. Approximately half of the flights are not affected by
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 187

a handover. The mean HO rate for flights affected by one or more handovers amounts

to approximately one HO per 40 min. These numbers of HO occurrence proves the

fact that seamless spot beam handover is essential for the provisioning of aeronautical

satellite communication when the satellite system incorporates narrow spot beams.

Impacts on signalling overhead Incorporation of a seamless handover functionality

introduces signalling overhead into a system. The scheduled movement of the mobile

group terminals enables nearly deterministic planning for channel allocation due to han-
dovers. The mean data required for signalling during the phases information gathering
and handover initialisation/execution can be neglected with respect to the terminal data

rates. Information gathering can be performed, e.g., by position reporting of the ter-
minal with an update rate of about one minute. The initialisation and execution may
require about 6 to 10 packets (c.f. [LWJ00] GSM-like signalling flow) with a mean
spot beam HO rate of approximately 0.7 HO/s (extracted from the peak number of spot

beam handovers in the worst case beam according to Table 5.11). Assuming an average

packet size of 0.8 kbit the amount of signalling overhead per satellite can be estimated
by 66 kbps4 for position reporting and 5.6 kbps5 for spot beam handover signalling.

Impacts on channel allocation The local and temporal changes of aircraft introduces

dynamics to the channel allocation within the satellite. In contrast to the signalling

overhead which is depending on the amount of occurring handovers, the number of

channels needed within a spot beam is influenced by the changes in number of aircraft

within the spot beam. In order to give an estimate of the dynamics of the resource

allocation, the decrease (respectively increase) of aircraft which are active in a certain

spot beam is analysed. An allocation of one channel per aircraft is assumed and that
the system is dimensioned to cope with the maximum number of aircraft active during a
P
day maxt s nac,sb in the associated satellite. For all other times the system is not fully
4
maximum 5000 aircraft active * 0.8 kbit / 60 seconds
5
10 packets * 0.8 kbit * 0.7 HO / second
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 188

loaded.

For satellite #2 the absolute difference of active aircraft in a spot beam nac,sb between
two time instances is calculated. This number is summed over all spot beams at one time
P
instant and is normalised by the maximum active aircraft during a day maxt s nac,sb .

Figure 5.19 depicts the histogram of these values over one day according to Equation
5.1.

P
sb (|nac,sb (t P
+ 1)− nac,sb (t)|)
(5.1)
maxt sb nac,sb

It can be seen that approximately a maximum of 11% of the channels and mean 7% of
the channels have to be dynamically changed within the 15 minutes time instances.

3
Histgram

0
0 0.02 0.04 0.06 0.08 0.1 0.12
sum over all spot beams of ∆ number of aircraft in spot beam
normalised to max number of aircraft in the satellite

Figure 5.19: Satellite #2: Histogram of sum of aircraft difference per spot beam in 15 min time
intervals normalised to the max number of aircraft of the satellite).

5.3 Assessment of Satellite System Concepts for Aero-

nautical Communication

5.3.1 Coverage Optimisation

Only GEO systems are used today for mobile broadband satellite communication, thus

– as a first element towards coverage analysis for a (global) constellation – the case
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 189

of two neighbouring satellites at longitude positions lon1 and lon2 , respectively, with

a longitude offset ∆lon = lon2 − lon1 is assessed. As illustrated in Figure 5.20, the
crossing point of the EOC circles for the considered minimum elevation yields a critical
latitude. This is the maximum latitude up to which 100% coverage availability in the

longitude sector between lon1 and lon2 can be guaranteed. Table 5.13 lists critical
latitudes versus longitude offset for the chosen reference minimum elevation angles.

Figure 5.20: Footprints and critical latitude for a 2-GEO case.

Table 5.12: Critical latitude values for the 2-GEO case.

Critical latitude in ◦ Longitude offset ∆lon in ◦
10 20 30 40 50 60 70 80
◦
εmin =0 81.3 81.2 81.0 80.8 80.4 80.0 79.4 78.6
εmin =5◦ 76.3 76.1 75.9 75.5 74.9 74.2 73.3 72.1
εmin =25◦ 57.0 56.6 55.8 54.7 53.2 51.2 48.5 44.9
Critical latitude in ◦ Longitude offset ∆lon in ◦
90 100 110 120 130 140 150 160
◦
εmin =0 77.7 76.4 74.7 72.5 69.1 63.8 54.4 29.7
εmin =5◦ 70.5 68.5 65.7 61.8 56.0 46.4 24.3 0.1
εmin =25◦ 39.9 32.4 18.9 0 0 0 0 0

In straightforward extension of the inhomogeneous 2-GEO case the relevant results for

the case of homogeneous n-GEO satellite constellations are presented, in the sense as

defined above. It can be seen that, e.g., no GEO constellation can provide coverage
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 190

with εmin = 25◦ at north European latitudes as it would be required taking a standard

roll angle like depicted in Figure 3.25 into account. 6 GEOs are required to achieve a
coverage with > 0◦ elevation at 80◦ .

Table 5.13: Critical latitude values for the homogeneous n-GEO case.
Critical latitude in ◦ Number of satellites
1 2 3 4 5 6 7 8
◦
εmin =0 0 0 72.5 77.7 79.3 80.0 80.4 80.6
εmin =5◦ 0 0 61.8 70.5 73.1 74.2 74.8 75.2
εmin =25◦ 0 0 0 39.9 47.9 51.2 53.0 54.0

An evaluation of the aircraft distribution statistics is summarised Table 5.14. In order

to cover 100% of global flight time (please recall that great circle routing is assumed) a

latitude coverage to 85◦ is necessary. Even though this is not completely achievable with
a GEO constellations, already a 4 GEO constellation can serve 99.99% of all scheduled
flights, assuming εmin = 0◦ .

Table 5.14: Critical latitudes versus aircraft distribution.

highest latitude to cover 100% of global flight time 85◦
highest latitude to cover 99.99% of global flight time 77◦
highest latitude to cover 99% of global flight time 63◦
highest latitude to cover 95% of global flight time 55◦

Besides a “truly” global coverage, also availability and reliability is an issue especially

for an integrated AirCom and air traffic management satellite approach. Thus, non-GEO

systems, may gain interest for a constellation design with multiple satellite visibility.

5.3.2 Handover vs. Capacity Optimisation

A placement of the satellite constellation in a way that the heavily loaded regions are

served by the overlap of two satellites, would allow to balance peak loads of one satel-
lite, but increases in parallel satellite HO occurrence. In order to numerically investigate

this trade-off, the reference constellation of Section 5.2.3.1 is shifted by 30◦ , 60◦ , and
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 191

all airlines, 2x30 min service off−time all airlines, 2x30 min service off−time
20 140 260 50 170290
downlink bit rate in Mbit/s

downlink bit rate in Mbit/s

sat 1 sat 1
150 150
sat 2 sat 2
sat 3 sat 3
100 100

50 50

0 0
0 5 10 15 20 0 5 10 15 20 25
UTC time in h UTC time in h
all airlines, 2x30 min service off−time all airlines, 2x30 min service off−time
80 200 320 110 230 350
downlink bit rate in Mbit/s

downlink bit rate in Mbit/s

sat 1 sat 1
150 150
sat 2 sat 2
sat 3 sat 3
100 100

50 50

0 0
0 5 10 15 20 25 0 5 10 15 20 25
UTC time in h UTC time in h

Figure 5.21: Comparison of downlink traffic demand for satellite constellation with 3 GEO
satellites at: reference (upper left); shift 1 (upper right); shift 2 (lover right); shift 3 (lower left)

90◦ and compared in terms of overall traffic and satellite HO values. The four constel-
lations will be named according to Table 5.15 in the following.

Table 5.15: Constellation Names

Constellation Name Sat #1 position Sat #2 position Sat #3 position
Reference 20◦ -100◦ 140◦
Shift 1 50◦ -70◦ 170◦
Shift 2 80◦ -40◦ -160◦
Shift 3 110◦ -10◦ -130◦

Figure 5.21 shows the total downlink traffic of the four satellite scenarios. The capacity

requirement of a constellation – assuming identical satellites – is determined by the

peak load of one single satellite. The peak load can be observed at the satellite which
is serving the domestic flights of continental US, cf. peak number of aircraft in Figure

5.11. In constellation “shift 2”and “shift 3” this area is served by two different satellites,

resulting in peak traffic load balancing.

Figure 5.22 shows on the left side the coverage areas of the satellites for all four satellite
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 192

Number of aircrafts moving between satellites 1 and 2

40 (1,2)
(2,1)
20
Sat 1: 20˚; Sat 2: -100˚; Sat 3: 140˚ 0
0 5 10 15 20
Time (hours)
Number of aircrafts moving between satellites 2 and 3
10 (2,3)
50
5 (3,2)
Latitude

0
0 5 10 15 20
0 Time (hours)
2 1 3
Number of aircrafts moving between satellites 3 and 1
15 (3,1)
50 10 (1,3)
5
0
150 100 50 0 50 100 150 0 5 10 15 20
Longitude Time (hours)
Number of aircrafts moving between satellites 1 and 2
40 (1,2)
20 (2,1)

Sat 1: 50˚; Sat 2: -70˚; Sat 3: 170˚ 0

0 5 10 15 20
Time (hours)
Number of aircrafts moving between satellites 2 and 3
50 20 (3,2)
10 (2,3)
Latitude

0
0 5 10 15 20
0 Time (hours)
2 1 3
Number of aircrafts moving between satellites 3 and 1
20 (3,1)
50 (1,3)
10

0
150 100 50 0 50 100 150 0 5 10 15 20
Longitude Time (hours)
Number of aircrafts moving between satellites 1 and 2
30 (1,2)
20 (2,1)
10
0
0 5 10 15 20
Sat 1: 80˚; Sat 2: -40˚; Sat 3: -160˚ Time (hours)
Number of aircrafts moving between satellites 2 and 3
100 (3,2)
50 50 (2,3)
0
Latitude

0 5 10 15 20
0 Time (hours)
3 2 1 Number of aircrafts moving between satellites 3 and 1
10 (1,3)
50 (3,1)
5

0
150 100 50 0 50 100 150 0 5 10 15 20
Longitude Time (hours)
Number of aircrafts moving between satellites 1 and 2
20 (1,2)
10 (2,1)

0
Sat 1: 110˚; Sat 2: -10˚; Sat 3: -130˚ 0 5 10 15 20
Time (hours)
Number of aircrafts moving between satellites 2 and 3
40 (2,3)
50
20 (3,2)
Latitude

0
0 5 10 15 20
0 Time (hours)
3 2 1
Number of aircrafts moving between satellites 3 and 1
15 (1,3)
50 10 (3,1)
5
0
150 100 50 0 50 100 150 0 5 10 15 20
Longitude Time (hours)

Figure 5.22: Coverage and comparison of handover from downlink traffic demand for satellite
constellation with 3 GEO satellites at: reference; shift 1; shift 2; shift 3 (from top to down)
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 193

scenarios. Associated to these coverage plots on the right side, the number of satellite

handovers in 15 minutes periods is shown for all satellite combinations, similar to the
analysis of Figure 5.15.

Table 5.16: Comparison of satellite HO and capacity requirements of reference and shifted
constellations
Constellation Max. number of cumu- Max. number Max. downlink data
lative satellite HO be- of cumulative rate (involved sat.)
tween two satellites one satellite HO all
day (involved sat.) satellites of the
constellation per
day
Reference 1100 (Sat#1 ⇔ Sat#2) 1744 155 Mbps (Sat#2)
Shift 1 1192 (Sat#1 ⇔ Sat#2) 2705 162 Mbps (Sat#2)
Shift 2 4209 (Sat#2 ⇔ Sat#3) 6018 135 Mbps (Sat#2)
Shift 3 1054 (Sat#2 ⇔ Sat#3) 2047 136 Mbps (Sat#3)

Table 5.16 summarises the peak traffic and the total number of satellite HO per day.
The minimal number of satellite HOs for both values, peak within 15 minutes and total
sum over a day between two satellites, occurs in constellation “shift 3”. The minimal
number of total satellite HOs per day is observed in the reference constellation. In both

constellations two areas of pronounced continental flights (Europe and US) are served
by one distinct satellite. In contrast, the HO occurrence in the constellation “shift 2” is
maximal, because these two areas of continental flights are intersected by satellite #2

and #3 and satellite #1 and #2, respectively.

The balancing effect on the data rate demand can be seen as well as the HO minima in

case the satellite footprints of a constellation do not intersect the two pronounced areas
of continental flights.

Although orbit positioning for GEO constellations is influenced by the availability of

free positions in the GEO arc, this comparison shows the potential gain in traffic bal-

ancing and minimisation of satellite HO.

CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 194

5.4 Conclusions

Various end user services and the mobility of passenger groups pose additional traffic
dynamics to satellite systems compared to static single user services. The thesis pro-
vides a methodology for a hierarchical traffic simulation of a satellite system shared

by mobile groups of users and a numerical example for commercial aircraft passenger
communication.

The target market of aeronautical passenger communication sums up to approximately

15 million passenger flight hours a day. All fights of less than 1 hour flight duration are

excluded in this number, because these flights are not considered to be of relevance for
the use of passenger communication services. 60% of the total passenger flight hours
are carried by the 20 largest of approximately total 550 airlines.

At peak times approximately 2200 aircrafts are simultaneous in flight and have to be
served by the satellite system. The peak number of aircrafts flying in one spot beam is

extremely unbalanced and is up to 285 aircrafts in satellite #1, 535 aircraft in satellite
#2 and 64 aircraft in satellite #3 of the reference satellite system.

Mean and peak data rates for the group terminal can be derived from the multi service

model and group categorisation. A peak data rate of 400 kbps to the aircraft is not
exceeded with a probability of 99.9% and 100 kbps from the aircraft. These results

define the data rate requirements of the group terminal.

Peak data rates at spot beam level are resulting by applying the local and temporal
user/group distribution to the multi service model. In the given example peak data

rates per spot beam are 16 Mbps downlink and 5 Mbps uplink. Resulting from the very

unequal distribution of aircrafts a very unequal distribution of the aeronautical passenger

data traffic becomes visible at spot beam level; 10 of 233 spot beams carry more than

50% of the satellite traffic.

Peak data rates at satellite level show approximately 150 Mbps in downlink and 40 Mbps

in uplink.
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 195

These numbers are input for satellite system dimensioning and the unequal traffic distri-

bution has to be reflected by flexibility of satellite resource allocation. Approximately a

maximum of 11% of the channels and mean 7% of the channels have to be dynamically

changed within 15 minutes time instances.

The model has also been used to provide service usage and traffic parameters to service
operators and helped to build up business models i.e. define end user service prices and

estimate cost for satellite capacity.

The hierarchical model allows besides traffic simulation also handover analysis based

on the local and temporal distribution of the user groups.

Spot beam handover peak rate from a satellite perspective results in 98 handovers in 15

minutes in one spot beam and 645 handovers in 15 minutes in one satellite.

From a terminal perspective about 50% of flights are not affected by a spot beam hand-
over. For flights which are affected by a spot beam handover the mean handover interval

from a terminal perspective is approximately 40 minutes. These numbers of handover

occurrence prove the fact that seamless spot beam handover is essential for the provi-
sioning of aeronautical satellite communication when the satellite system incorporates

narrow spot beams. The average spot beam signalling overhead per satellite can be esti-
mated by 66 kbps for position reporting and 5.6 kbps for spot beam handover signalling.

Satellite handover is typically not seamless and a service outage is visible to the end user.

Peak handover is occurring for the handovers between satellite #1 and satellite #2 which

reflect the intercontinental Atlantic flights. The peak handover rate is approximately 30

handovers in 15 min and 550 handovers per day. From terminal perspective a maximum
one satellite handover per aircraft during a flight is occurring and only 1744 are affected

by a satellite handover, which seems to be an acceptable number.

A system traffic versus handover analysis shows that the example satellite system has an
optimum w.r.t. the total number of satellite handovers and the number of handovers can

be drastically reduced if the satellite locations are chosen in way that Europe and North
CHAPTER 5. System Dimensioning for Collective Mobile Satellite Terminals 196

America is covered by one single satellite each. This is to the cost of a increased peak

traffic demand of the satellite, because the distinct peak loads cannot be shared between
two satellites.
Chapter 6

Summary

This thesis deals with the collective use of broadband satellite terminals and the issues

imposed by the mobility of the user group. A generalised baseline for a systematic
system design process is followed, highlighting the substantial topics of mobility as
there are:

• Satellite channel characteristics for mobile environments: Mobility of a ter-

minal affects satellite visibility and influences fundamentally the performance of
a satellite communication in multiple ways, which have been addressed in Chap-

ter 3. The effects on signal availability demands for segment handover in shad-

owed areas as well as for optimised algorithms for antenna re-acquisition of the

satellite (resp. modem re-synchronisation) after signal loss.

• Terminal antenna pointing, acquisition and tracking (PAT): Broadband satel-

lite communication requires directive terminal antennas. Thus, the thesis dis-

cussed terminal mobility, which introduces variations in the satellite to terminal

geometry and requires continuous pointing of terminal antennas towards the satel-

lite.

• Hierarchial system dimensioning, including capacity and handover analysis:

Temporal and spatial distribution of the collectively mobile group and the char-

197
CHAPTER 6. Summary 198

acteristics of multiple services used by the group are introduced into the system

dimensioning process.

This thesis is not only based on theoretical analysis, but the author was personally in-
volved with key responsibility in system design, development, and operational introduc-

tion of the first systems for collectively use which are operational today.

In the following, the core areas of investigation are recapitulated, the main results are
summarised chapter-wise.

Chapter 2 summarises relevant issues of mobile satellite communications and lays a

basis for a common understanding of mobility issues of satellite terminals.

The chapter includes the relevant parameters of satellite to mobile geometric re-

lations in a unified description, i.e., earth spherical coordinates of the terminal,

being of importance for the calculations in the subsequent chapters and in general

for any evaluation with impact from the mobile to satellite geometry like Doppler
calculation, PAT algorithms.

An overview of the current status of regulation for mobile satellite terminals at

Ku and Ka band is given, because these frequency band attracted interest for col-

lective mobile terminals. The requirements of aeronautical earth stations (AES),

earth stations on vessels (ESV) and earth stations on trains (EST) are compared.

The influences of terminal mobility onto the radio propagation path are given and

a model for channel characterisation used in chapter 3 for statistical evaluation of

channel measurements is reviewed.

Chapter 3 is dedicated to the peculiarities of the mobile satellite channel. Statistical

evaluations of measurement campaigns as well as analytical assessments of chan-

nel parameters are given for the LMS and the aeronautical channel.

Exhaustive channel measurements have been performed for the LMS channel at S

and EHF bands and for the aeronautical satellite channel at Ka band. With these
CHAPTER 6. Summary 199

channel measurements a general categorisation of the mobile environments and

operation scenarios has been introduced.

Land mobile channel measurements at S band have been performed for the satel-
lite system ICO. The measurement setup consisted of two receive and transmit
paths to investigate diversity aspects. A Zeppelin airship was used to emulate the

slow angular changes of the MEO constellation. Channel data have been collected

in down- and uplink of two diversity paths and are presented for selected user en-

vironments and operational scenarios. These measurements have found a recent

interest revival by the emerging AtG plans discussed within CEPT working group
FM48.

Within the projects “Satellite EHF Communications for Mobile Multimedia

Services” (SECOMS) and ABATE, this thesis performed and evaluated land mo-
bile channel measurements at EHF band and aeronautical satellite channel mea-

surements at Ka band being of major interest for all broadband aeronautical MSS
systems specifically Inmarsat GlobalXpress.

An overview on the LMS EHF band measurement campaign is given and the
derived channel parameters for narrowband and wideband models are presented.
The LMS channel measurement results have been verified by an outage analysis

based on very precise, ’real world’ 3D architectural and vegetation models for

urban environment. Important work for generation of 3D building, roof and veg-
etation shapes has been performed in the late 90s. For the first time such models

have been used for a deterministic verification of the statistically generated LMS
data in this work, whereas today the models are the predominant network planning

tools for terrestrial mobile network operators. The LMS modelling confirmed the

expected on/off (line of sight (LoS)/non-LoS) behaviour of satellite channels at

higher frequencies.

The performed aeronautical channel measurements confirmed an on/off and (more

or less) deterministic behaviour of aeronautical satellite channel at higher fre-

CHAPTER 6. Summary 200

quencies, based on shadowing by the aeroplane structure during flight maneu-

vers. It is shown that the overlaying attenuation effects, like attenuation due to
rain, clouds and water vapour which are of interest especially for higher frequen-

cies, become of less importance for aeronautical satellite communication, because

cruise heights are mostly above the clouds.

Besides the characterisation of channel parameters, the aeronautical demonstra-

tion anticipated the need of new mobility concepts for mobile user groups.

Chapter 4 examines the mobility aspects introduced on directive terminal antennas.

All channel measurements and service demonstration campaigns at higher fre-

quencies which have been performed in the frame of this thesis and related
projects, required steered satellite terminal antennas with accurate PAT.

Different antenna technologies have been used for the investigation of PAT algo-
rithms. While observing the attempts to overcome the technological constraints
of electronically steered beams at a reasonable price, this thesis discusses the

tradeoffs between mechanically and electronically steered beams. The common

problem of an optimised antenna PAT for different environments and vehicles is

addressed in depth.

A methodology for the design of an optimised antenna PAT system is presented.

This methodology takes the relevant factors into account as there are i) movement

characteristics of mobile and satellite, ii) the antenna parameters and regulatory
aspects, iii) the satellite channel characteristics, and iv) the availability resp. price

and accuracy of position and attitude parameters. All these aspects are discussed
thoroughly within the thesis.

The characterisation of movements for vehicle types of all mobile scenarios, i.e.,
car, train, ship and aircraft has been derived either by evaluation of publicly avail-

able data or by measurements campaigns. Generally it can be stated that the

attitude changes of the mobile platform prevails the effects of position changes
CHAPTER 6. Summary 201

of satellite and mobile. Smaller vehicles of any mobile environment, i.e., land,

maritime, and aeronautical tend to faster attitude changes, and limitation of those
are given by the comfort aspects of the passengers. The results show also, that

PAT implementation for cars are worst case with respect to both the agility of the

mobile vehicle as well as the cost constraints in a mass market.

Both beam steering mechanisms, mechanically and electronically, are detailed

and discussed in their applicability for the three axes azimuth, elevation, and po-

larisation. Mechanical beam steering supports typically full gain in all directions

and is less expensive, but requires a protuberant implementation. Electronical

beam steering in contrast, inheres the possibility of multiple beam forming net-

works, being of major importance if seamless satellite handover has to be sup-

ported. A drawback here is certainly the decrease of the antenna gain with in-
creased steering angle and the usually more complex implementation resulting in
increased costs.

The functionality of “open loop” and “closed loop” based PAT solutions are de-

scribed and the PAT algorithms are studied under the aspect of applicability for
different mobile scenarios. While the open loop PAT performance is not influ-
enced by signal shadowing, its accuracy depends on the attitude and position in-

formation available for pointing vector calculation. Sensors which are applicable

for open loop PAT are assessed in accuracy and price. Closed loop PAT is inde-

pendent from attitude and position data, but has drawbacks in environments with
high signal shadowing probability and in case polarisation mismatch has to be

compensated by the antenna. Examples for advantageous combinations of both

algorithms are given within the thesis.

A description for calculation of azimuth, elevation and polarisation PAT errors

and associated signal loss with typical values is presented. An optimisation of the

allowed pointing error versus antenna gain for receive-only antennas is derived,

yielding a PAT system with a maximum tolerated antenna off boresight angle of
CHAPTER 6. Summary 202

4.3 dB.

Based on these analyses and measurement campaigns, requirements and techni-

cal limitations for the different mobile scenarios have been identified, and exam-
ples of prototypal implementations for aeronautical and land mobile environments
conclude this chapter of the thesis.

Chapter 5 finally introduces a general hierarchical multi-service system dimensioning

methodology for collectively mobile user groups.

The layered approach starts with the identification of the applicable service mix

and usage patterns based on the travel characteristics of the group. Together with
the group size this leads to the aggregate traffic mix of one group terminal and is

input for the terminal design. Local and temporal distributions of user groups and
mapping of the distribution on coverage areas of satellite networks are the basis
for a capacity and handover requirement analysis for system dimensioning.

For passengers communication of an aircraft – which represents the most evident

example for a satellite connection to a closed mobile user group – this system

dimensioning is detailed.

Aggregate mobile terminal traffic of four group classes (i.e. short haul single
aisle, short haul wide body, long haul wide body, and long haul jumbo) and sys-

tem capacity requirements are calculated on basis of the three services with most

significant impact onto the transport link: passenger voice, web, and email.

The market size is given on the one hand as a total number of aircraft, flights per
day and on the other hand as passenger flight hours. For a reference GEO satellite

scenario the user group distribution is presented based on worldwide scheduled

commercial flights.

By simulation of global fight traffic the impact of service off-time on mobile

groups being “active”, i.e., aircraft being in service, has bee shown. Without the

service off-times, the peak of aircraft in service is around 7500 (4450 for single
CHAPTER 6. Summary 203

aisle); while with consideration of service off-times, the peak of aircraft in service

is around 4200 (2200 for single aisle). Considering flights with a duration longer
than 1h, the largest 20 airlines serve 60% of all global PFHs.

A clear focus of the user distribution on the northern hemisphere and especially
the continental US and Europe can be seen. The number of aircraft/passengers

over spot beams is extremely unbalanced. The beams serving continental US

and Europe are extremely loaded, while the majority of beams face virtually no

demand. In the given example, the 10 most loaded spot beams have to carry

more than 50% of the data demand. The pronounced peak load stems mainly

from short haul flights due to the dominant domestic US flights and continental
European flights, while long haul flights have some balancing effect.

Simulation of aggregate traffic shows, that in total sufficient satellite capacity

could be provided to cope with the data demand. But, the focused ’hot spots’
involve neighbouring spot beams and data rates of up to 15 Mbps in the worst

case cannot be handled by a typical spot beam concept and cluster bandwidth
available today.

Handover analyses show the vital need for system inherent seamless (uninter-

rupted for the end user application) spot beam handover for satellite systems util-
ising spot beams, while the data rate needed for signalling overhead due to spot

beam handover is negligible in comparison to the user data rates. The dynamics

in channel allocation is about 10% of the total channel capacity.

Aeronautical data traffic analyses as well as number of handover occurrences are

presented for 3-GEO satellite constellations at different orbital positions. Optimal

constellations can be found minimising satellite HO and traffic balancing.

Summarising, the core contribution of this thesis are:

• Introduction of a systematic satellite communication system design process for

collective mobile terminals;

CHAPTER 6. Summary 204

• Statistical land mobile satellite (LMS) channel characterisation by measurements

at S band and EHF band as well as analytical determination of LMS shadowing

parameters by 3-dimensional urban models;

• Statistical aeronautical satellite channel characterisation by measurements at

Ka band as well as analytical determination of aeronautical shadowing by aircraft

attitude mobility and geometry;

• Methodology for optimised PAT system design including characterisation of mo-

ments for vehicle types of all mobile scenarios i.e. car, train, ship and aircraft;

• Tradeoff analysis between mechanically and electronically steered beams and ap-

plicable PAT algorithms;

• Description of azimuth, elevation and polarisation PAT errors and optimisation of

pointing error vs. antenna gain;

• Introduction of a general hierarchical multi-service dimensioning methodology

for collectively mobile user groups, including voice, data, and multimedia ser-

vices;

• Aeronautical system dimensioning scheme with (capacity and handover) require-

ments analysis and evaluation of results for different satellite scenarios.

Applicability today and future work

The channel model parameters which have been derived within this PhD work are time-

liness valid and gained recent interest especially for future aeronautical direct air to
ground communication and satellite communication at Ka band.

With the knowledge that the satellite visibility is the predominant channel parameter and

very little to no multipath propagation can be observed the huge effort for channel mea-
surement campaigns cannot be justified. Future work should concentrate on analytical

characterisation of the satellite channel.

CHAPTER 6. Summary 205

Today the service model parameters used within this thesis need to be revised in general

and especially for the aeronautical market. Voice traffic bandwidth and QoS require-
ments are still unchanged, but the traffic mix has dramatically changed towards data

traffic significantly influenced by the technology change of mobile end user terminals

and applications. Moreover, the user experience of fixed and mobile data bandwidth
provided by terrestrial systems demands for revised parameters of web browsing and

email data models.

Today still no usage parameters of aeronautical GSM and Internet service are available

to public. Thus, a validation of the traffic models is not possible. But, the introduction

of small aeronautical GSM solutions may indicate that the total number and the number
of simultaneous calls was overestimated within the early studies.

Of increasing importance for a sustained business model for aeronautical communica-

tion the extension of cabin communication into the cockpit has to be considered. Cock-
pit services silently use already the higher cabin communication bandwidths. Research

work has to establish data models with a changed service mix of cabin and cockpit
applications, different data priorities and delay requirements.
Appendix A

Acronyms

2G 2nd generation

3D three dimensional

3G 3rd generation

3GPP 3rd Generation Partnership Project

4G 4rd generation

A/C aircraft

AAC airline administrative communications

ABATE “ACTS Broadband Aeronautical Terminal Experiment”

ACK acknowledge

ACeS ASIA Cellular Satellite

ACTS Advanced Communications Technology Satellite

AES aeronautical earth station

AHRS attitude and heading reference system

206
APPENDIX A. Acronyms 207

AirCom aeronautical communication

AOC airline operational control

APC airline passenger correspondence

ARINC Aeronautical Radio Incorporated

AtG direct Air to Ground

ATS air traffic services

BGAN Broadband Global Area Network

BGP border gateway protocol

CBB Connexion by Boeing

CCDF complementary cumulative distribution function

CDF cumulative distribution function

CDM code division multiplex

CDMA code division multiple access

CERS combined empirical roadside shadowing

CEPT Conférence Européenne des Administrations des Postes et des

Télécommunications

CONUS continental or contiguous US

CW continuous wave

D-GPS differential global positioning system

DAQ data acquisition

DAT digital audio tape

APPENDIX A. Acronyms 208

DLR German Aerospace Center – Deutsches Zentrum für Luft- und Raumfahrt

DoF degree of freedom

DSL digital subscriber line

DSP digital signal processing

DVB digital video broadcast

DVB-S digital video broadcast - satellite

DVB-S2 digital video broadcast - satellite version 2

DVB-RCS digital video broadcast - return channel via satellite

EIRP effective isotropically radiated power

EERS extended empirical roadside shadowing

EN European standard

ECEF earth centered, earth fixed

ENU east north up

EOC edge of coverage

ERS empirical roadside shadowing

ETSI European Telecommunications Standards Institute

ESA European Space Agency

EST earth stations on trains

ESV earth station on vessel

FDM frequency division multiplex

FDMA frequency division multiple access

APPENDIX A. Acronyms 209

FIFTH “Fast Internet for Fast Train Hosts”

FSS fixed satellite services

GEO geostationary earth orbit

GSM global system for mobile communications

GSO geosynchronous earth orbit

GPRS general packet radio service

GPS global positioning system

GW gateway

HEO highly elliptical orbit

HO handover

HW hardware

HSDPA High-Speed Downlink Packet Access

HSUPA High-Speed Uplink Packet Access

ICO intermediate circular orbit

IEEE Institute of Electrical and Electronics Engineers

IMAP internet message access protocol

IMU inertial measurement unit

INS inertial navigation system

IP internet protocol

ISL intersatellite links

ITU International Telecommunication Union

APPENDIX A. Acronyms 210

ITU International Telecommunication Union - Telecommunication Standardisation

Sector

JB jumbo A/C

JPL Jet Propulsion Laboratory

kbps kilobit per second

LEO low earth orbit

LH long haul flight

LHCP left hand circularly polarised

LoS line of sight

LMS land mobile satellite

LMS-PRO “LMS Propagation Model For Non-Urban Areas”

LTE “long term evolution”

Mbps megabit per second

MEO medium earth orbit

MEMS micro-machined electromechanical system

MERS modified empirical roadside shadowing

MMS multimedia messaging service

MOBILITY “Mobile Real Time TV via Satellite Systems”

MSL mean sea level

MSS maximum segment size

MSS mobile satellite service

APPENDIX A. Acronyms 211

MTU maximum transmission unit

NCF network control facility

NED north east down

NOC network operation centre

OFDM orthogonal frequency division multiplexing

PAT pointing, acquisition, and tracking

PDA personal digital assistant

PFH passenger flight hours

PID proportional plus integral plus derivative

POP3 post office protocol - version 3

PLL phase locked loop

PSTN public switched telephone network

QoS quality of service

RCS return channel via satellite

RHCP right hand circularly polarised

RTT round trip time

SA single aisle A/C

SBB Swift Broadband

SCC satellite control centre

SDMA space division multiple access

SH short and medium haul flight

APPENDIX A. Acronyms 212

SMS short messaging service

SMTP simple mail transport protocol

SNIR signal to noise and interference ratio

SNR signal to noise ratio

S-PCN satellite personal communication network

SECOMS “Satellite EHF Communications for Mobile Multimedia Services”

SUITED “Multi-Segment System for Broadband Ubiquitous Access to Internet

Services and Demonstrator”

TCP transmission control protocol

TDM time division multiplex

TDMA time division multiple access

TTC telemetry, tracking, and command

UMTS Universal Mobile Telecommunications System

USAT ultra-small aperture terminal

UT user terminal

UTC universal time clock

VPN virtual private network

VSAT very small aperture terminal

WB widebody A/C

WirelessCabin “Development and Demonstrator of Wireless Access for Multimedia

Services in Aircraft Cabins”
APPENDIX A. Acronyms 213

W-LAN Wireless local area network

WRC world radio conference

WSSUS wide-sense stationary uncorrelated scattering

Appendix B

Notation and Symbols

Terminology Equivalents

satellite spot beam ≡ terrestrial cell

satellite ≡ basestation

satellite constellation ≡ terrestrial network consisting of several basestations

Constants

Constant Value Meaning

c0 299792458 m/s Speed of light in vacuum

3
γ0 6.672 ·10−11 kgms2 Gravitational constant

k 1.38 ·10−23 WKs Boltzmann constant

ME 5.974 ·1024 kg Earth mass

RE 6378.144 km Mean equatorial earth radius

rn 6.62 GEO radius normalised by RE

TR 86164.091 s (23 h, 56 m, 4.091 s) Sidereal day

214
APPENDIX B. Notation and Symbols 215

Notation and Operators

x Variable

x Mean value
x() Function
x Vector

xT Vector transpose
X Matrix

XT Matrix transpose

Symbols

α Yaw angle of a mobile platform

α(p) Parameter of the EERS model

A Shadowing factor of the Lutz model

Ae Aperture area
A(p, ε, f ) Attenuation of the EERS model
β Roll angle of a mobile platform

β(p) Parameter of the EERS model

b Number of beams in a pattern

B Bandwidth

γ Pitch angle of a mobile platform

γ(p) Parameter of the EERS model

c Rice factor
d Slant range, Distance

D Diameter of an circular reflector antenna, Reuse distance in cell or spot

beam patterns
δS Latitude of satellite

δT Latitude of (mobile) terminal

APPENDIX B. Notation and Symbols 216

∆ Round trip time

∆δ Latitude difference between mobile and satellite

∆pE
TS Pointing error

ε, ε(t) Elevation angle

εmin Minimal elevation angle
f Frequency
fc Carrier frequency

fD Doppler frequency

fL Frequency at L band

hS Orbit height
i Inclination

η Antenna efficiency

ηi Orthogonality factor of interferer i

ηmodcod Modulation and coding efficiency

G, G(ϕ, ε) Antenna gain

Gmax Maximum antenna gain

K Cluster size
LP , LP Pagelength, Average pagelength

λ Wavelength, Call arrival time

∆λ Longitude difference between mobile and satellite

λS Longitude of satellite

λT Longitude of (mobile) terminal

µ Mean value of distribution, Call holding time

µi Activity factor of the interferer i

ν Mean number of echoes

N0 Noise power spectral density

NO Number of orbit planes of a constellation

NS Number of satellites of a constellation
APPENDIX B. Notation and Symbols 217

ψ Earth central angle

ψmax Max. earth central angle

p Percentage

pij Probability
pT S Terminal to satellite pointing vector, Line of sight vector
P Markov transition matrix
ϑ Nadir angle, Antenna off boresight angle

ϑmax Max. nadir angle

R Length of an edge of a hexagon for cell/spot beam pattern modelling

RE Mean earth radius

rS Distance Earth centre to satellite

rT Distance Earth centre to (mobile) terminal

S Transition probability matrix

b Number of beams in a pattern

nsb number of spot beams

σ Standard deviation

ς Polarisation angle
TE Earth rotation period

Ta Time for take off and climb

Td Time for decent and landing

T Orbit period, System temperature

Tvis,max Maximum visibility time of a satellite

v Velocity
ϕ, ϕ(t) Azimuth angle

χ Outer angle of spherical triangles

ω Angle between ascending node and perigee

ω(t) Angle between ascending node and satellite

Appendix C

Transformations

C.1 Earth - Mobile - IMU - Antenna Coordinate Trans-

formation

xL zL
2 4 zL
xL
yL zA
E
zI
p
TS

zE pET xA x I
yE 3
1 p SE
yL yI
xE
yA

Mobile Orientation of Orientation of

terminal terminal w.r.t. L antenna w.r.t. I
movements is unknown is unknown
E L I A
IMU, north
GPS Calibration
seeking sensor

Figure C.1: Transformation of earth to mobile to IMU to antenna coordinates.

218
APPENDIX C. Transformations 219

C.1.1 Earth Fixed Coordinates E

The pointing vector in earth fixed coordinates pE

T S can be calculated with user position

information and the satellite position only:

pE E E
T S = pS − pT , (C.1)

with

   
 cλT cδT   cλS cδS 
   
pE 
T = rT  sλT cδT
 and pE = r  s c  , (C.2)
 S S  λS δS 
   
sδT sδS

where cx and sx denote the cosine and sine, respectively, of the associated angle λ or δ.

C.1.2 Local Coordinates L

The pointing vector in local coordinates is given by

¡ ¢T E
pLTS = TE
L pT S , (C.3)

where TE
L is the basis of the local coordinate system L, while the column vectors of this

basis are defined in the earth coordinate system E. TE

L is defined such that

1. x-axis points toward north,

2. y-axis points toward west,

3. z-axis is normal vector of local tangential plane, hence x-y-plane is the local tan-

gential plane.
APPENDIX C. Transformations 220

 
 −cλT sδT sλT cλT cδT 
 
 −s s −cλT sλT cδT 
 λT δ T 
TL = 
E

 .
 (C.4)
 cδT 0 sδT 
 | {z } | {z } | {z } 
 
xL yL zL

Hence, information as provided by a GPS receiver is sufficient to get TE

L . However,

still the attitude with respect to the local coordinate system is unknown, but required for
pointing.

C.1.3 Inertial Measurement Unit Coordinates I

The pointing vector in IMU coordinates is given by

¡ ¢T
pIT S = TLI pLTS , (C.5)

with

TLI = YPR =
     
 cγ −sγ 0   cβ 0 sβ   1 0 0 
     
     
 s      (C.6)
 γ cγ 0   0 1 0   0 cα −sα 
      .
     
0 0 1 −sβ 0 cβ 0 sα cα
| {z } | {z } | {z }
Y: yaw P: pitch R: roll

An IMU provides information about its attitude (α, β, γ) relative to a known (with

respect to the local coordinates) initial "null-orientation". Definitely not required, but
for convenience, this "null-orientation" can be defined such that for α = β = γ = 0 the
APPENDIX C. Transformations 221

IMU’s coordinate system is aligned with the local coordinate system. Matching of the

z-axes (α = β =0) is provided by measurement of the local gravity gradient g, because

g
= −zL = −zI . (C.7)
|g|

The gravity gradient g is measured inside the IMU with some inertia sensors.

To match the x-axes (and the y-axes accordingly) additional information about the true

north direction is required by means of e.g. compass or a dual antenna GPS receiver,

because xL is equivalent to the north vector. Then, γ is the angle between the vectors

xI (α = β = 0) and xL .

Any misalignment between the north-seeking sensor and the IMU x-axis xI results in a
(possibly unknown) yaw-offset ² such that Y must be replaced by Y’, with

   
 c² −s² 0   cγ −sγ 0 
   
Y’ =    
 s² c² 0   sγ cγ 0  =
   
0 0 1 0 0 1
  (C.8)
 cγ+² −sγ+² 0 
 
= 
 sγ+² cγ+² 0  .
 
0 0 1

¡ ¢T
Hence, the calculation pIT S = TLI pLTS , with (α, β, γ) provided by the IMU, but
without taking care of ², is resulting in a pointing error.

C.1.4 Antenna Coordinates A

To control the antenna such that pointing towards the satellite is maintained, finally the

pointing vector in the antenna coordinate system is derived by

¡ A ¢T I
pA
T S = TI pT S . (C.9)
APPENDIX C. Transformations 222

Where TIA takes care of offset angles (∆α, ∆β, ∆γ) between the IMU and the antenna,

such that

TA
I =
     
 c∆γ −s∆γ 0   c∆β 0 s∆β   0 
     1 0 
     
 s   1 0    (C.10)
 ∆γ c∆γ 0   0   0 c∆α −s∆α 
      .
     
0 0 1 −s∆β 0 c∆β 0 s∆α c∆α
| {z } | {z } | {z }
yaw offset pitch offset roll offset

In general, (∆α, ∆β, ∆γ) will be unknown at first and must be found by some way.
However, it is not necessary to really find (∆α, ∆β, ∆γ) rather it will be sufficient to
calculate TIA without splitting it up down to the three offset matrices. Finally, it can
written

¡ I ¢T ¡ ¢T ¡ ¢T
pA
T S = TA TLI pLTS = TLI TIA pLTS =
     T
 c² −s² 0   cγ −sγ 0   cβ 0 sβ  1 0 0  
(C.11)
     
 s 0      0 c −s  TIA  pL .
 ² c ²   s γ cγ 0   0 1 0  α α   TS
     
0 0 1 0 0 1 −sβ 0 cβ 0 sα cα

Once pA
T S is known, the antenna just needs to be steered toward this direction to achieve

pointing.

C.2 Polarisation Transformation

Transformation of the equatorial and polar polarisation planes to the local horizontal of

an terminal antenna as used in section is detailed in the following.

APPENDIX C. Transformations 223

The two linear polarisations are polar and equatorial polarisation and are defined by

two orthogonal vectors. The polar polarisation vector is perpendicular to the satellite
signal propagation vector and lies in the plane defined by the satellite antenna beam

axis and Earth’s polar axis (or a line parallel to this axis). The equatorial polarisation

vector is, again, perpendicular to the satellite signal propagation vector and parallel to
the equatorial plane (note that the normal vector of the equatorial plane is parallel to

the polar axis). Because the polar axis vector is perpendicular to the equatorial plane,
the equatorial polarisation vector must be perpendicular to the polar axis vector. This is

illustrated in Figure C.2.

Figure C.2: Definition of polar and equatorial polarisation vectors.

The orientation of the polarisation vectors with respect to a local coordinate system

which is defined by the local tangential plane and true-north vector of a mobile satel-
lite receiver will vary, depending on the longitude and latitude of the receiver location.

Usually, this variation is measured as the angle between the polarisation vector and the
line of intersection between the local horizontal plane and a plane perpendicular to the

signal propagation vector, i.e. the plane containing the polarisation vectors of the signal.

This is clarified in Fig. C.3, where ςP,E and ςP,P denote the polarisation angle w.r.t. to
the equatorial and polar polarisation vector, respectively.
APPENDIX C. Transformations 224

Figure C.3: Definition of polarisation angles ςP,E and ςP,P .

Appendix D

Traffic Models

D.1 Web Browsing

D.1.1 Packet Level Model

Internet traffic must be distinguished between forward link and return link, where in the
return link only the 40 Byte acknowledgement TCP packets are considered that are sent
in reply to each TCP packet received.

Forward link

HTTP traffic is modelled according to the UMTS WWW model by ETSI [ETS98],

which is a hierarchical source traffic model with three levels: session, page (or packet

call), and packet.

In the following this model is shortly summarised and numerical parameter values which

adapt the quite outdated ETSI values (original ETSI values given in brackets where
adapted) to current WWW reality are introduced. More details of this adaptation of the

traffic parameters are given in our study [BUWH04].

A HTTP session consists of a sequence of page requests (packet calls), each of which
consists of a sequence of packets. The complete model comprises the following com-

ponents:

225
APPENDIX D. Traffic Models 226

• A Poisson session arrival process where the value of the arrival rate is left as free

parameter to adjust the target average data rate;

• the number of page requests per session, which is a geometrically distributed

random variable with a mean value of 10 (5);

• the reading time between page requests, which is a geometrically distributed ran-

dom variable with a mean value of 206 seconds (412 sseconds);

• the number of packets within a page request, which is a geometrically distributed

random variable with a mean value of 250 (25);

• the inter-arrival time between two consecutive packets inside a page request,

which is a geometrically distributed random variable with a mean value which

is directly linked with the average bit rate at source level;

• the packet size which is Pareto-distributed with cutoff to limit the maximum
packet size to 66666 bytes, where the mean value is 480 bytes and the minimum
is 81.5 bytes. With these parameters the mean page size is 120 kbytes (12 kbytes).

Note that the session length is modelled implicitly by the number of events during the

session, and is 2060 seconds on average for the assumed parameters. These components

fully describe the dominating forward link traffic from the Web server to the user.

Return link

For the sake of tractability in the simulation, the return link traffic is modelled as fol-

lows according to [RGCD99]: all return link packets are simply considered as 40-byte
responses to downlink TCP packets.

Although a TCP packet can be as large as approx. 64 kByte, it is often encountered

in TCP/IP implementations that the TCP Maximum Segment Size (MSS) is determined
with the IP Maximum Transfer Unit (MTU) by the relationship MSS = MTU - 40 Byte.
APPENDIX D. Traffic Models 227

To determine the number of ACK packets sent per page download, each packet gener-

ated by the ETSI/UMTS model is divided by the MSS.

This approach is pretty much in line with observations from real WWW traffic traces
and approximately yields the wanted asymmetry of 1:10 between return and forward

link traffic. Finally, it is assumed that the rate with that the ACK packets are sent is
variable and only determined by the total volume of ACK packets per page download

and the duration of the respective page download. Within one page download the rate is
constant.

D.1.2 Page Level Model

D.1.2.1 Forward Link

Deviating from the ETSI/UMTS model and according to the MMPP model of web traf-
fic of [MMM+ 04]), it is assumed that

• the session arrival process follows a Poisson process (ETSI/UMTS: same);

• page download times within an active session are generated according to a Poisson
process (ETSI/UMTS: geometrically distributed);

• packet generation times within a page download follow a Poisson process

(ETSI/UMTS: geometrically distributed inter-arrival times),

• the number of page downloads within a session is geometrically distributed

(ETSI/UMTS: same);

• the number of packets in a page download is geometrically distributed

(ETSI/UMTS: same).

Obviously, the MMPP model deviates from the ETSI/UMTS model in some points,
but both suggest the hierarchical model of the packet generation process. Note the well-

known similarity of the continuous exponential and discrete geometric distribution (e.g.,
APPENDIX D. Traffic Models 228

for small success probability the geometric random process approaches the exponential

random process).

In this study the capacity requirement analysis is based on page downloads only, but not
on generated packets. It is assumed that the page download duration is given by

LP
TP (LP ) = , (D.1)
RU

where LP is the page size and RU is the data rate per user. Accordingly, the average

page download duration TP is given with the average page size LP according to

LP
TP = , (D.2)
RU

Actually, the page size and thus the page download duration depends on the number of
packets per page (geometrically distributed) and on the packet size (the ETSI/UMTS
assumes a Pareto distribution). However, for simplification it is assumed that the page
size is exponentially distributed, consequently also the page download duration is ex-

ponentially distributed. Indeed, a comparison of the page size distribution according

to ETSI/UMTS model and an exponential distribution with the same mean page size

shows that this assumption is quite accurate.

The maximum page size is quasi limited to approx. 2 MByte, as the page size distribu-

tion quickly decays approx. exponentially. It is assumed that a throughput of 24 kbps

per page download is a good compromise between acceptable performance for “norma”

web browsing and bandwidth demand.

Return link

Similar to the web traffic modelling, the data rate required in the return link per page

download is variable, and given by the accumulated ACK volume per page, and the page
download duration. Again, according to the approx. 1:10 of the transmitted volumes in

the return and forward link, a 1:10 factor also in the data rate requirements is expected.
APPENDIX D. Traffic Models 229

D.2 Email

Paxson [Pax94] found the distribution of simple mail transport protocol (SMTP) orig-
inator bytes to be bimodal log2normal distributed, essentially reflecting the two cases
of emails with (larger) file attachment and such without. However, other publications

suggest a number of other approximations (trimodal lognormal, Pareto, Cauchy, just to

name a few). Here it is assumed as a simple model for email size distribution only one

log2normal distribution. Paxson assumes as the mean message size 1360 Byte, where
the minimum message size is 300 Byte. This reflects not anymore the message size

observed in more recent traffic traces where the mean email size lies rather between 20
and 60 kByte (cf., e.g., [Cha02]). Each email upload/download relies on TCP and thus
requires transmission of ACK packets from the receivers side. However, the volume of

the ACKs is very small compared the other traffic sources, and will thus be neglected in
the following (this will be justified by simulations).

D.2.1 Packet Level Model

The packet level model is based on the model described in the next section. The packets

are generated by assuming an MTU (typically 1500 Byte) and calculating the number

of packets simply from dividing the email size by the MTU. Again, as it is the case for
the web browsing model, the MSS is given by MTU minus 40 Byte, and a 40 Byte ACK

packet it sent for each received TCP packet.

D.2.2 Email Download/Upload Level Model

The mean and standard deviation of the log2normal distribution are

APPENDIX D. Traffic Models 230

σln2
E{x} = 2µ 2 2 + 0.3 = 60 kByte ,
(D.3)
√
Std{x} = 22(µ+σ2 ln2) − 22µ+σ2 ln2 = 500 kByte .

where the corresponding mean and standard deviation of the underlying normal distri-

bution are µ = 12.26 ; σ = 3.1 [Byte].

However, with the above assumptions, very large emails occur with a non-negligible

probability. Therefore, an email size limitation of 2 MByte is assumed. Then, the

average email size amounts to approx. 45 kByte.

Of course, the duration of an email upload/download depends on the achieved through-

put. It is assumed that a throughput of 8 kbps is sufficient to provide a tolerable up-

load/download duration for the average email size (45 kByte*8/8 kbps = 45 ssecons).
Of course, the times for text-only emails will be significantly shorter.

D.2.2.1 Return Link

The message generation rate for SMTP messages is modelled as a Poisson process with

1 email/hour.

D.2.2.2 Forward Link

Finally, due to the proven overall asymmetry of received vs. sent emails (emails are

in the average sent to more than one address), which ranges from 3:1 to 6:1, a best
guess average asymmetry of 4:1 of received vs. sent emails [IBI01] is assumed, hence

the generation rate for received emails – according to post office protocol - version
3 (POP3)/internet message access protocol (IMAP) – is modelled as Poisson process

with 4 emails per hour.

APPENDIX D. Traffic Models 231

Although the increased arrival rate may appropriately describe the rate at which emails

arrive at the email server, it does not necessarily mean that the user downloads emails
with the same increased rate from the email server. Then modelling the email traffic

with unchanged arrival rate, but with 4 emails being downloaded during each email

server access, would appear as a more appropriate model.

D.3 Voice

Voice service is assumed to require a symmetric data rate of 9.6 kbps. Call inter-arrival
times and call holding times are exponentially distributed, where the mean call inter-
arrival time is 5400 seconds, and the mean holding time is 180 seconds. Since voice
call service is symmetric, no differentiation of forward and return link is required.

D.3.1 Packet Level Model

Encoding of the voice traffic can be done in several ways. Assuming encoding after
G723.1 (5.3 kbps codec), two frames (each 20 Byte long for talk spurts, and 4 Byte for

silence periods) of 30 ms frame duration are assembled in one IP packet. This results

in a bit rate of approx. 10 kbps for talk spurts. Depending on the voice coder, silence

periods may be coded with a lower bit rate, however, here the higher bit rate for talk

spurts is only assumed to obtain worst-case estimates. Thus, it is assumed that on the

packet level, a voice call generates 80 Bytes every 60 ms.

D.3.2 Call Level Model

Classical telephone service is modelled on call level with Poisson arrivals (parameter

λ=1/5400 seconds) and negative exponentially distributed call holding times (parameter

µ=1/180 sseconds). The traffic generated by an user accepting the service is A= λ/µ =
APPENDIX D. Traffic Models 232

180/5400 = 0.033 Erlang. To determine the data rate requirements, it is assumed that no

call blocking occurs, such that the number of parallel voice calls is given by the Poisson
distribution.

Based on the assumed statistical properties of the service usage (Poisson), the data rate
can be derived, which is required to serve a certain number of concurrent voice calls,

occurring at a certain probability. As expected, the multiplexing gain is significant. The

accumulated data rate which is not exceeded with a probability of 99.9% is 96 kbps for

100 users, respectively 297.6 kbps (500 users), 508.8 kbps (1000 users), 1996.8 kbps

(5000 users), 3753.6 kbps (10000 users).

Bibliography

[ACe11] ACeS, “Aces International Website.” http://www.acesinternational.com/,

last visited 2011.

[AP95] B. Abbe and D. Pinck, “Amt experiment results,” in International Mobile

Satellite Conference 1993, pp. 147–157, 1995.

[APG+ 01] J. Alonso, J. Perez, L. Gonzalez, V. Schena, F. Ceprani, and M. Holzbock,

“SUITED vehicular Ka band satellite terminal,” in Proceedings of IST Mo-
bile Communications Summit 2001, pp. 120–125, 2001.

[ARI04] ARINC, “Arinc specification 429 part 1-17,” 2004.

[AVA+ 00] C. O. Adler, D. C. Vacanti, D. J. V. Alen, G. W. Fitzsimmons, W. D.

Walker, T. M. White, R. T. Ott, D. A. Pietila, and J. L. Freeman, “A phased

array antenna for the airborne reception of data and direct broadcast TV,”

in Proceedings Millennium Conference on Antennas & Propagation (AP

2000), Davos, Switzerland, April 2000. ESA.

[BB88] J. Berner and D. Bell, “Te jpl mechanically steered antenna,” in Interna-
tional Mobile Satellite Conference, 1988, pp. 241–247, 1988.

[Bel63] P. A. Bello, “Characterization of randomly time-variant linear channels,”

IEEE Transactions on Communications Systems, pp. 360–393, December,

1963.

233
BIBLIOGRAPHY 234

[BER92] G. Butt, B. G. Evans, and M. Richharia, “Narrowband channel statistics

from multiband propagation measurements applicable to high elevation

angle land-mobile satellite systems,” vol. 10, pp. 1219–1226, 1992.

[BH00] H. Brandt and K. Hochkirch, “Entwicklung eines Segeldynamometers zur

Erfassung der aero- und hydrodynamischen Kräfte.” Abschlußbericht zum

BMBF Projekt: 18S0164, 2000.

[BJL+ 93] A. Böttcher, A. Jahn, E. Lutz, B. Schmidt, and M. Werner, “Networking

requirements for user oriented leo satellite systems,” in ESA Study No.

9732/91/NL/RE, Final Report, ESA/ESTEC, 1993.

[BN86] D. Bell and F. Naderi, “Vehicle antenna pointing techniques for mobile
satellite communications,” in AIAA Space System Technology Conference,

pp. 22–30, 1986.

[Bru96] P. Brunt, “Iridium: Overview and status,” Space Communications, vol. 14,
no. 2, pp. 61–68, 1996.

[BUWH04] L. Battaglia, P. Unger, M. Werner, and M. Holzbock, “Satellite capacity

dimensioning for in-flight internet services in the north atlantic region,”

in Proceedings of 22nd AIAA International Communications Satellite Sys-
tems Conference & Exhibit 2004 (ICSSC 2004), Monterey, USA, pp. Paper

Nr. AIAA 2004–3189, May 2004.

[BWH03] L. Battaglia, M. Werner, and M. Holzbock, “Revenues and performance

of global satellite aeronautical communications,” in Proceedings of 21st

AIAA International Communications Satellite Systems Conference (IC-

SSC ’03), Yokohama, Japan, pp. Paper No. AIAA–2003–2318, Apr. 2003.

[CCI] CCIR - Consultative Committee for International Radio, “CCIR recom-

mendation 791.”
BIBLIOGRAPHY 235

[CEP10] CEPT, “FM4810-012 Prioritisation of band under discussion.”

http://forum.ero.dk, November 2010.

[Cha02] J. Charzinski, “Observations in e-mail performance,” in Proceedings ITC

Specialist Seminar, Würzburg, Germany, pp. 133–142, July 2002.

[Cla45] A. C. Clarke, “Extra-terrestrial relays,” Wireless World, vol. 51, no. 303,

pp. 305–307, 1945.

[CMSS03] P. Conforto, R. Mura, L. Secondiani, and S. Scalise, “Multimedia service

provision in mobility: the fifth solutions for the railroad environment,”

in Proceedings of 9th Ka and Broadband Communications Conference,
Ischia, Italy, November 2003.

[Cox89] D. C. Cox, “Portable digital radio communications - an approach to teth-

erless access,” IEEE Communications Magazine, vol. 27, pp. 30 – 40, July

1989.

[Dav93] F. Davarian, “The acts propagation campaign,” in Eighth International

Conference on Antennas and Propagation, pp. 159–166, 1993.

[DEJ+ 98] M. Döttling, H. Ernst, A. Jahn, J. Kunisch, J. Pamp, and E. Zollinger,

“Land mobile satellite propagation model for non-urban areas.” Final Re-
port, ESA Study 12044/96/NL/NB(SC), 1998.

[Den93] A. Densmore, “A satellite-tracking k- and ka-band mobile vehicle antenna

system,” in IEEE Transactions on Vehicular Technology, Vol. 42, Iss. 4,

pp. 502–513, 1993.

[DFS06] DFS - Deutsche Flugsicherung, “Flight data base,” 2006.

[DG96] A. Densmore and M. Guler, “An aeronautical-mobile 20/30 ghz satellite-

tracking antenna for high data rate satcom,” in Anntennas and Propagation

Society International Symposium, 1996, pp. 1108–1111, 1996.

BIBLIOGRAPHY 236

[DJKB98] M. Döttling, A. Jahn, J. Kunisch, and S. Buonomo, “Versatile channel sim-

ulator for land mobile satellite applications,” in Proceedings IEEE Vehicu-

lar Technology Conference (VTC’98), Ottawa, Ontario, Canada, pp. 213–

217, 1998.

[DJWW93] A. Densmore, V. Jamnejad, T. K. Wu, and K. Woo, “K and Ka band

mobile-vehicular satellite-tracking reflector antenna system for the NASA

ACTS mobile terminal,” in Proceedings of the Third International Mobile

Satellite Conference (IMSC 93), pp. 563–568, 1993.

[ESK+ 02] H. Ernst, S. Scalise, J. Kunisch, J. Siemons, G. Harles, and J. Hörle, “Mea-
surement campaign for the land mobile satellite channel in Ku band,” in

European Personal and Mobile Satellite Conference (EMPS), Bologna,

Italy, September 2002.

[ETS98] ETSI, “Universal mobile telecommunications system (UMTS): Selec-

tion procedures for the choice of radio transmission technologies of the
UMTS..” Technical Report TR 101 112 v3.2.0, Aug. 1998.

[ETS04] ETSI, “EN 302 186, Harmonized EN for satellite mobile aircraft earth
stations AESs operating in the 11/12/14 GHz frequency bands covering

essential requirements under article 3.2 of the R&TTE directive.” Candi-

date Harmonized European Standard (Telecommunications series), 2004.

[ETS05] ETSI, “EN 302 340; Harmonized EN for satellite earth stations on board

vessels ESVs operating in the 11/12/14 GHz frequency bands allocated to

the fixed satellite service FSS covering essential requirements under article
3.2 of the R&TTE directive.” Candidate Harmonized European Standard

(Telecommunications series), 2005.

[ETS06a] ETSI, “EN 302 448; Harmonized EN for satellite earth stations on trains

ESTs operating in the 11/12/14 GHz frequency bands allocated to the

BIBLIOGRAPHY 237

fixed satellite service FSS covering essential requirements under article

3.2 of the R&TTE directive.” Candidate Harmonized European Standard

(Telecommunications series), 2006.

[ETS06b] ETSI, “Terms and definitions database interactive website.”

http://webapp.etsi.org/Teddi/, last visited 2006.

[Eur05] Eurocontrol, “Departures, arrivals, internals and overflights (DAIO) web-

site.” http://www.eurocontrol.int/statfor/public/standard_page/daio.html,

2005.

[GHLW03] D. Giggenbach, M. Holzbock, O. Lücke, and M. Werner, “Optical-SDMA

for broadband aeronautical communication,” in Proceedings of 21st AIAA
International Communications Satellite Systems Conference (ICSSC ’03),

Yokohama, Japan, pp. Paper No. AIAA–2003–2320, Apr. 2003.

[GK76] R. Gagliardi and S. Karp, Optical Communications. New York: John

Wiley & Sons, 1976.

[Glo10] Globalstar, “Globalstar Announces Successful Launch of Six New

Second-Generation Satellites.” Press Release 20. October 2010, October

2010.

[GV88] J. Goldhirsh and W. J. Vogel, “Propagation effects by roadside trees mea-

sured at UHF and L band for land-mobile satellite systems,” in Mobile

Satellite Conference, JPL Pub. 88-9, pp. 87–94, 1988.

[GV95] J. Goldhirsh and W. J. Vogel, “An extended empirical roadside shadowing

model for estimating fade distributions from UHF to K band for mobile

satellite communications,” in Space Communications (Special Issue on

Satellite Mobile and Personal Communications), vol. 13, pp. 87–94, 1995.
BIBLIOGRAPHY 238

[HB99] N. Haala and C. Brenner, “Extraction of buildings and trees in urban en-

vironments,” in International Archives of Photogrammetry and Remote

Sensing 54, pp. 130–137, 1999.

[HBWJ03] M. Holzbock, L. Battaglia, M. Werner, and A. Jahn, “Cost and perfor-

mance of HEO and GEO European satellite radio systems,” in Proceedings

of 21st AIAA International Communications Satellite Systems Conference

(ICSSC ’03), Yokohama, Japan, pp. Paper No. AIAA–2003–2349, Apr.

2003.

[HHE89] G. Hilton, G. Hawkins, and D. Edwards, “Novel antenna tracking mecha-

nism for land mobile satellite terminals,” in Fifth International Conference

on Mobile Radio and Personal Communications, pp. 182–186, 1989.

[HHG10] G. Hoffmann, A. Helberg, and M. Grobler, “A brief survey of channel

models for land mobile satellite communications,” in Southern Africa
Telecommunication Network and Applications Conference, 2010.

[HHJW04] M. Holzbock, F. Hu, A. Jahn, and M. Werner, “Advances of aeronautical

communications in the EU framework,” International Journal of Satellite

Communications and Networking, vol. 22, pp. 113–137, January/February

2004.

[HJ98a] M. Holzbock and A. Jahn, “Broadband end-to-end satellite service demon-

strator in the frame of ABATE,” in 3rd ACTS Mobile Communication Sum-

mit, Rhodos, 1998.

[HJ98b] M. Holzbock and A. Jahn, “Combined PAT algorithms for mobile multi-

media satellite services,” in Proceedings 48th IEEE Vehicular Technology

Conference (VTC ’98), pp. 1553–1557, 1998.

[HJG+ 00] M. Holzbock, A. Jahn, F. Grassl, J. van Noten, M. Pugliese, V. Schena,

and S. Dragas, “Ka band satellite network technology: Demonstration of
BIBLIOGRAPHY 239

mobile services in multi-segment IP networks,” in Proceedings IST Mobile

Summit 2000, pp. 663–668, 2000.

[HJL99] M. Holzbock, A. Jahn, and E. Lutz, “Aeronautical channel measurement

trials at K band,” in Proceedings of 6th International Mobile Satellite Con-

ference (IMSC99), pp. 144–149, 1999.

[HLCL99] M. Holzbock, E. Lutz, M. Connally, and G. Losquadro, “Aeronautical

multimedia service demonstration at K/Ka band,” in Proceedings of 6th

International Mobile Satellite Conference (IMSC99), pp. 5–11, 1999.

[HLL99] M. Holzbock, E. Lutz, and G. Losquadro, “Aeronautical channel measure-

ments and multimedia service demonstration at K/Ka band,” in Proceed-
ings of 4th ACTS Mobile Communications Summit 1999, pp. 841–847,

1999.

[HLO00] M. Holzbock, O. Lücke, and B. Oeste, “Mobility requirements of land-

mobile antennas for broadband satellite communication,” in Mobile and
Personal Satellite Communications 4, Proceedings 4th European Work-

shop on Mobile/Personal Satcoms (EMPS 2000), J. G. Gardiner and I. A.

Glover (eds.), London, UK, pp. 34–43, Sept. 2000. University of Bath,

Bath, UK.

[Hoc00] K. Hochkirch, Entwicklung einer Meßyacht zur Analyse der Segelleistung

im Originalmaßstab. Berlin: Mensch & Buch Verlag, 2000.

[HS99] M. Holzbock and C. Senninger, “An aeronautical multimedia service

demonstration at high frequencies,” IEEE Multimedia, vol. 6, no. 4,

pp. 20–29, 1999.

[IBI01] IBIS, “IBIS service and system requirements.” EC project IST-2000-

25091, Deliverable D20, 2001.

BIBLIOGRAPHY 240

[Inm10] Inmarsat, “Inmarsat to invest US$1.2bn in Ka-band network.” Press Re-

lease 06. August 2010, November 2010.

[Iri10] Iridium, “Iridium Announces Comprehensive Plan for Next Generation

Constellation.” Press Release 02. June 2010, June 2010.

[ITU92] ITU, “Recommendation P.682-1, Propagation data required for the design

of earth-space aeronautical mobile telecommunication systems,” 1992.

[ITU95a] ITU, “Recommendation P.341-4, The concept of transmission loss for ra-

dio links,” 1995.

[ITU95b] ITU, “Recommendation P.531-5, Ionospheric propagation data and pre-

diction methods required for the design of satellite services and systems,”
1995.

[ITU99a] ITU, “Recommendation P.839-2, Rain height model for prediction meth-
ods,” 1999.

[ITU99b] ITU, “Recommendation P.840-1, Attenuation due to clouds and fog,”

1999.

[ITU03a] ITU, “Recommendation P.618-8, Propagation data and prediction meth-

ods required for the design of earth-space telecommunication systems,”

2003.

[ITU03b] ITU, “Recommendation P.837-4, Characteristics of precipitation for prop-

agation modelling; model to derive the rainfall rate exceeded for a given

probability of the average year and a given location,” 2003.

[ITU03c] ITU, “Resolution 902 (WRC-03), Provisions relating to earth stations lo-

cated on board vessels which operate in fixed-satellite service networks in

the uplink bands 5925-6425 MHz and 14-14.5 GHz,” 2003.
BIBLIOGRAPHY 241

[ITU05] ITU, “Software concerning tropospheric propagation website.”

http://www.itu.int/ITU-R/software/study-groups/rsg3/databanks/tropo-
sph, last visited 2005.

[Jah99] A. Jahn, “Propagation characteristics for land mobile satellite systems

from L band to EHF band,” in Joint International Workshop Cost 252,

Cost 253 Cost 255, pp. 89–102, 1999.

[Jak75] W. C. Jakes, Microwave Mobile Communications. New York: John Wiley

& Sons, February 1975.

[JH98] A. Jahn and M. Holzbock, “EHF band channel characterisation for mo-
bile multimedia satellite services,” in Proceedings 48th IEEE Vehicular
Technology Conference (VTC ’98), pp. 209–212, 1998.

[JHE+ 99] A. Jahn, M. Holzbock, H. Ernst, O. Grémillet, and B. Oeste, “Supply of

propagation data by field trial.” ICO Contract ISL/98-1418EC„ 1999.

[KC96] A. G. Kanatas and P. Constantinou, “Narrowband characterisation of the

land mobile satellite channel: a comparison of empirical models,” in Eu-

ropean Transactions on Telecommunications, 1996.

[Kep09] J. Kepler, Astronomia nova. 1609.

[Kep19] J. Kepler, Harmonice Mundi. 1619.

[KVH06] KVH, “KVH TracVision A5 Website.” http://www.kvh.com/, last visited

2006.

[KYE88] A. Kuramoto, T. Yamane, and N. Endo, “Mechanically steered track-

ing antenna for land mobile satellite communications,” in Anntennas

and Propagation Society International Symposium, 1988, pp. 1314–1317,

1988.
BIBLIOGRAPHY 242

[LCD+ 91] E. Lutz, D. Cygan, M. Dippold, F. Dolainsky, and W. Papke, “The land

mobile satellite communication channel – recording, statistics, and chan-

nel model,” IEEE Transactions on Vehicular Technology, vol. 40, pp. 375–

386, 1991.

[LH03] O. Lücke and M. Holzbock, “Antenna agility requirements and pointing,

acquisition and tracking for broadband mobile satellite terminals,” in Pro-

ceedings of 21st AIAA International Communications Satellite Systems

Conference (ICSSC ’03), Yokohama, Japan, pp. on CD–ROM, Apr. 2003.

[Lig11] LightSquared, “LightSquared Website.” http://www.lightsquared.com/,

last visited 2011.

[Liv11] LiveTV, “LiveTV Website.” http://www.livetv.net/, last visited 2011.

[Loo85] C. Loo, “A statistical model for a land mobile satellite link,” in IEEE

Transactions on Vehicular Technology, p. 122Ű127, 1985.

[LPG+ 03] O. Lücke, J. Pérez, J. A. Guerra, A. Rodríguez, and V. Gennatos, “The

MOBILITY project: Providing DVB-S services on the move,” in Pro-

ceedings of 21st AIAA International Communications Satellite Systems

Conference (ICSSC ’03), Yokohama, Japan, April 2003. AIAA.

[LR03] E. Laase and W. R. Richards, “Deploying Connexion by Boeing satellite

services,” in Proceedings 21st International Communication Satellite Sys-

tems Conference & Exhibit (ICSSC), Yokohama, Japan, April 2003. AIAA
2003-2315.

[Luf06] Lufthansa, “Lufthansa Technik Website.” http://www.lufthansa.com/, last

visited 2006.

[Luf10] Lufthansa, “Lufthansa FlyNet service now available on board.” Press Re-

lease 30. November 2010, November 2010.

BIBLIOGRAPHY 243

[Lut88] E. Lutz, “Land mobile channel-recording and modelling,” in Forth Inter-

national Conference on Satellite Systems for Mobile Communications and

Navigation, pp. 15–19, 1988.

[Lut96] E. Lutz, “A Markov model for correlated land mobile satellite channels,”
International Journal of Satellite Communications, vol. 14, pp. 333–339,

July/Aug. 1996.

[LWJ00] E. Lutz, M. Werner, and A. Jahn, Satellite Systems for Personal and

Broadband Communications. Berlin Heidelberg: Springer-Verlag, 2000.

[MB02] G. Maral and M. Bousquet, Satellite Communications Systems. Chich-

ester: John Wiley & Sons, 2002.

[MMI+ 03] A. Monk, P. Martens, Y. Inasawa, N. Yoneda, I. Naito, M. Miyazaki,

Y. Shimawaki, Y. Konishi, A. Iida, and S. Makino, “Ultra-low profile

airborne reflector antenna subsystem for broadband satellite communica-

tions,” in Proceedings 21st International Communication Satellite Systems

Conference & Exhibit (ICSSC), Yokohama, Japan, April 2003.

[MMM+ 04] L. Muscariello, M. Mellia, M. Meo, M. A. Marsan, and R. L. Cigno, “A

MMPP-based hierarchical model of internet traffic,” in Proceedings IEEE

International Conference on Communications (ICC ’04), 2004.

[Mob06] Mobile Satellite Ventures, “Press release: Mobile satellite ven-

tures engages boeing to develop next generation satellites.”

http://www.msvlp.com/, January 2006.

[MP92] M. Mouly and M.-B. Pautet, The GSM System for Mobile Communica-

tions. Telecom Pub, June 1992.

[Nak09] T. Nakamura, “Proposal for Candidate Radio Interference Technologies

for IMT-Advanced Based on LTE Release 10 and Beyond.” 3GPP IMT-

Advanced Evaluation Workshop, Dec. 2009.

BIBLIOGRAPHY 244

[Neu89] A. Neul, Modulation und Codierung im aeronautischen Satellitenkanal.

Dissertation, 1989.

[New87] I. Newton, Philosophiae Naturalis Principia Mathematica. 1687.

[NSR10] Nothern Sky Research, “Mobile satellite services 6th edition,” August
2010.

[Par92] J. D. Parsons, The Mobile Radio Propagation Channel. London: Pentech

Press, 1992.

[Pax94] V. Paxson, “Empirically derived analytic models of wide-area TCP con-

nections,” IEEE/ACM Transactions on Networking, vol. 2, no. 4, pp. 316–

336, 1994.

[PEBB96] M. Parks, B. Evans, G. Butt, and S. Buonomo, “Simultaneous wideband

propagation measurements applicable to mobile satellite communication
systems at L and S band,” in International Conference on Satellite Systems
for Mobile Communications and Navigation, pp. 64–71, 1996.

[PMT01] R. J. Piechocki, J. P. McGeehan, and G. V. Tsoulos, “A new stochastic

spatio-temporal propagation model (SSTPM) for mobile communications

with antenna arrays,” May 2001.

[Pos96] P. Poskett, “Satellite systems architectures,” in Mobile and Personal Satel-

lite Communications 2 – Proceedings 2nd European Workshop on Mo-

bile/Personal Satcoms (EMPS ’96), pp. 485–500, London: Springer, 1996.

[Ray06] Raysat, “Raysat EagleRay Website.” http://www.raysat.com/, last visited

2006.

[RCD94] B. R. Rao, L. Callahan, and R. Davis, “Shf cassegrain antenna with elec-
tronic beam squint tracking for high data rate mobile satellite commu-
BIBLIOGRAPHY 245

nication systems,” in Anntennas and Propagation Society International

Symposium, 1994, pp. 1028–1031, 1994.

[RGCD99] A. Reyes, E. Gonzales, E. Casilari, and A. Diaz, “A page-oriented WWW

traffic model for wireless system simulation,” in Proceedings 16th Inter-

natinal Teletraffic Congress ITC, 1999.

[Ric48] S. O. Rice, “Statistical properties of a sine wave plus random noise,”

vol. 27, pp. 109–117, 1948.

[Ric86] M. Richaria, “Design considerations for an earth station step-tracking sys-

tem,” Space Communication and Broadcasting, vol. 4, 1986.

[RM96] J. Restrepo and G. Maral, “Providing appropriate service quality to fixed

and mobile users in a non-geo satellite-fixed cell system,” in Mobile and
Personal Satellite Communications 2 – Proceedings 2nd European Work-

shop on Mobile/Personal Satcoms (EMPS ’96), London, pp. 79–96, 1996.

Springer.

[Row11] Row44, “Row44 Website.” http://www.row44.com/, last visited 2011.

[SBB93] M. Sforza, S. Buonomo, and J. P. V. P. Baptista, “Global coverage mobile

satellite systems: System availability versus channel propagation impair-
ments,” in Proceedings IMCS Š93, pp. 361–372, 1993.

[Sch95] J. Schindall, “Concept and implementation of the Globalstar mobile satel-

lite system,” in Proceedings 4th International Mobile Satellite Conference

(IMSC ’95), pp. A11–A16, 1995.

[SH01] R. E. Sheriff and Y. F. Hu, Mobile Satellite Communication Networks.

Chichester: John Wiley & Sons, 2001.

[SLH+ 00] V. Schena, G. Losquadro, M. Holzbock, A. Jahn, J. Alonso, and S. Dra-

gas, “SUITED demonstrator proposal for a new technology in the Ka band

BIBLIOGRAPHY 246

radio frequency front-end for wideband satellite mobile terminals,” in

Proceedings of IST Mobile Communications Summit 2000, pp. 747–752,

2000.

[Suz77] H. Suzuki, “A statistical model for urban radio propagation,” in IEEE

Transactions on Communications, p. 673Ű680, 1977.

[TA96] S. C. Taylor and A. R. Adiwoso, “The Asia Cellular Satellite System,” in

16th AIAA International Communications Satellite Systems Conference,

Washington, DC, pp. 1239–1249, 1996.

[Thu11] Thuraya, “Thuraya Website.” http://www.thuraya.com/, last visited 2011.

[TW05] D. H. Titterton and J. L. Weston, Strapdown Inertial Navigation Technol-

ogy. Institution of Engineering and Technology, 2005.

[Vat91] G. Vatt, “The Iridium system,” in Virginia Tech. 1st Symp. on Wireless
Personal Communications, pp. 4.1–4.9, 1991.

[Ver10] Verizon Wireless, “Verizon Wireless 4G LTE: More than just speed.” Press
Release 01 December 2010, Dezember 2010.

[Via11] ViaSat, “ViaSat Website.” http://www.viasat.com/, last visited 2011.

[Vio95] M. Violet, “The development and application of a cost per minute metric

for the evaluation of mobile satellite systems in a limited-growth voice

communications market,” in MasterŠs thesis, Massachusetts Institute of

Technology, 1995.

[Vod10] Vodafone, “Vodafone continues to pave the way for a nationwide LTE

network.” Press Release 21 August 2010, August 2010.

[VTL95] W. J. Vogel, G. W. Torrence, and H. P. Lin, “Simultaneous measure-

ments of L and S band tree shadowing for space-earth communications,”

BIBLIOGRAPHY 247

in Proceedings 4th International Mobile Satellite Conference (IMSC ’95),

pp. 105–108, 1995.

[WHS97] M. Werner, Y. Hu, and R. Sheriff, “Satellite EHF communications for

mobile multimedia service SECOMS - network dimensioning report.” EC

project AC004, Deliverable D022/b1, 1997.

[Woo88] K. Woo, “Vehicle antenna development for mobile satellite applications,”

in Fourth International Conference on Satellite Systems for Mobile Com-

munications and Navigation, pp. 23–27, 1988.

[WV93] R. A. Wiedemann and A. J. Viterbi, “The Globalstar mobile satellite sys-

tem for worldwide personal communications,” in Proceedings of 3rd In-
ternational Mobile Satellite Conference (IMSC Ś93), Pasadena, CA, June

16-18 1993.