SuperCell: Optimizing coverage

and capacity for Rural Connectivity

Pratheep Bondalapati, Julius Kusuma, Mustafa Sahin, Abhishek Pathak

and Jin Bains

January 2020
  Rural cellular communication
– Challenges in 3G/4G for rural connectivity

 Solution: SuperCell
- Architecture & simulation results

Agenda  SuperCell channel models

- Challenges & novel solutions

 Antenna systems for SuperCell

- Requirements overview
- Candidate class of antennas & comparison

2
 Mission:
Rural Cellular Connectivity
Make rural LTE deployment attractive to Mobile Operators by
dramatically reducing the TCO by deploying tall towers with spectrally
efficient base-stations.

 Rural population is typically distributed as settlements scattered

over a very wide area
250m
 Low promising ROIs due to low demand and distributed
settlements
40m
 A possible solution: single BTS site mounted on a high-elevation
platform that can provide similar coverage as multiple traditional
macrocell sites

Approach:

Mount SuperCell LTE base-stations on tall towers. Each SuperCell site

replaces multiple conventional 40 m rural macrocell sites → lowers
TCO for mobile network operators and consumers (fewer sites, lower
cost overall).

3
Resulting Coverage & Link budget
Conventional three sector system
• Shown is an acceptable heatmap of Sample cell-edge link budget for SuperCell
the resulting SINR by employing HOS
over a coverage area of 80kmx80km. Parameter Value (downlink) Value (uplink)

• The corresponding throughput Tx power 46 dBm 23 dBm

coverage for an LTE base station is
also shown Tx ant gain 28 dBi 0 dBi

• For a given bandwidth and a transmit Pathloss 170 dB 170 dB

power, the conventional three sector
system can achieve an aggregate of Rx ant gain 0 dBi 28 dBi
approximately 200 Mbps
Rx power -103 dBm -119 dBm
• Whereas the proposed architecture Proposed higher order sectorization
achieves an aggregate throughput of Bandwidth 20 MHz 500 kHz
2 Gbps
Noise power -101 dBm -117 dBm

Noise figure 8 dB 4 dB

SNR -3 dB - 6 dB
Cellular BTS Architecture Comparison
Conventional Proposed: SuperCell

5
Channel models Propagation model?
Multipath profile?

for SuperCell
6
Site Planning Workflow
SINR heatmap example

Antenna pattern
intended for
deployment
Antenna radiation pattern
(measured in chamber)
Model
Pathloss model tuning parameters
with Ray tracing

Post processing
+ Cell planning Cell SINR Profile
Field
measurements Laplacian/Gaussian tool (eg.
curve fit ATOLL/Planet)
Model
Propagation Gaussian/Laplacian parameters
model model fit
+
Multipath Coverage
Cell Radii
profile (Throughput)

7
Challenges in site-survey for
long-range coverage

• The channel modeling studies have been trending towards the

advances in cellular industry from 2G through 5G.

• Lack of interest leads to lack of technology investment, both

on RAT and channel models.

• Key technical & logistical challenges with field measurements:

• Stringent link budget considerations for long range power

calculations
• Equipment dimension requirements
• Tower climbing & equipment installation
• Wind loading during measurement

8
High-resolution PAS Clutter interference definition Test methodology

measurement

• PAS: Power azimuth spectrum

• Motivation:
Higher-order sectoring is required to meet the
increased population being covered in the 50km Anticipated results
radius. However HOS beyond 6 sectors poses a risk of
inter sector interference. Power azimuth spectrum results

Sectorization gain
• What is causing intersector interference?
• Antenna side/back lobe’s contribution to a Location 1, location 2
neighboring sector
No. of sectors
• Scattering caused by the clutter in the environment

9
Multi-dimensional field measurements - I

1
0
PAS measurement results

The PAS is expected to depend on:

1. Elevation difference
2. Tx-Rx Distance
3. Pathloss
4. Delay spread

Principle component analysis, shows PAS dependence on Pathloss to

be the highest, as 36%

1
1
Multi-dimensional field measurements - II

1
2
 Learnings summary:
• Propagation hypothesis: Cellular propagation models predict ~ 6 dB
lower path loss upon each doubling of BTS height

• Angular spread hypothesis: As the BTS height increases above the

average clutter the angular spread of the incident signal reduces

Adaptive high order sectorization (AHOS) for greater capacity

• Frequency reuse offered by AHOS provides greater capacity

over an extended coverage region. Inter-sector interference
limits the capacity gain of AHOS

• Lower angular spread observed by AHOS antenna placed on tall

tower reduces inter-sector interference and enables the
capacity benefits of HOAS

• Narrow sector beamwidth result in additional antenna gain to

compensate for UE device EIRP limitations 13
Power Angular Spectrum w.r.t. height
Facebook Confidential &
Need for Ray-tracing • Effect of antenna pattern: Conduct Ray-tracing simulations to resolve the
dependency of measured pathloss values on the antenna patterns and orientations.
simulations
• PAS heatmap: Generate a heatmap of PAS over a specified region based on a few
field measurements

• Pathloss vs. PAS: Generate additional PAS data points to study of the effect of
pathloss on PAS

• Channel model vs. Clutter density: Analyze Pathloss and PAS as a function of clutter
density by artificially placing clutter in the Tx-Rx vicinity

• MIMO condition number: Assess the MIMO channel richness by artificially placing
buildings and evaluating the multipath inside the buildings.

• Reciprocal pathloss model: Provide simulation proof to enable the Pathloss

measurement without the use of an elevated platform i.e. with a dish antenna on
the ground

14
3D Ray-tracing simulation setup for Multipath

Calibration with
field measurements

1
5
Studying the effect of Clutter on Multipath

1
6
Angular spread 3D Simulation results

17
Antenna systems for
SuperCell
1
8
19
Candidate class of antennas: Pros & Cons
Total (per site) EPA/
Solution Bands Cost Modularity achievable capacity weight Pros Cons
(20 MHz BW) for 360
coverage
High, low and mid bands High 12/10 sectors 1620 Mbps Low - Lower side lobes - High cost
in the same product per panel - No NRE required - Low modularity
Lens - Low EPA - No steerability
(698-896 MHz) - High capacity - Cusping loss
(1710-2690 MHz) - Long range (High
gain)
- Positive IRR
Either high or mid band High 2/3 sectors 1620 Mbps High - Dynamic steering - High cost
per panel - High gain - NRE required
Active (1.7 – 2.6 GHz) + Dynamic - Good modularity - High EPA
FDD only steering - High capacity
- Long range
- Cusping loss can be
avoided
Different products for Low 1/2 sectors 270 Mbps (with 6 Med - Low cost - Narrow elevation
high, low and mid bands per panel sectors) - Long range (high beam
Passive gain) - Higher side lobes
(694 – 960 MHz) - No NRE required - Poor capacity (lower
(1710 – 2170 MHz) - No cusping loss no. of sectors)
(2300 – 2960 MHz)

20
Additional benefits: Cost-Effective Backhaul to Small Cells
and WiFi-Hot Spots

SuperCell

MNO Core

WiFi Hotpots

Small Cell

21
Additional benefits: Adaptive sectorization

Real Time State of Connectivity datasets: Network

Population density datasets Insights, Market Insights powered

Data fusion and computation

engine

Avionics sensing suite

on ground on elevated platform

+
CPRI
RRU
Real-time KPIs and resource usage information for interference coordination

Baseband
processing unit
(eNodeB)
+ RRU

+ RRU

Baseband
+ RRU

processing unit
(eNodeB)
+ RRU

+ CPRI
RRU
data format Antenna system
conversion

+ RRU
Baseband
processing unit
(eNodeB)
Waveform Agnostic Adaptive
Sectorization block + RRU

+ RRU

Complex number multiplier 22

+ Complex number adder
Uniform sector
allocation: Example 1

2
3
Uniform sector
allocation : Example 2

2
4
Non-uniform sectorization algorithm

2
5
Non-uniform sector
allocation: Example 1

2
6
Non-uniform sector
allocation: Example 2

2
7
Conclusion • Rural long-range cellular connectivity has its own challenges – both technical and
logistical

• We addressed some of these challenges jointly by working with site planning

experts in industry

• Any novel ideas on enabling Pathloss and PAS measurements that could
potentially overcome these issues can be a good scope for future work.

• We performed ray-tracing simulations to extrapolate the measurements to a

wider range of test points.

• Such a calibrated simulation model allows us to remove the dependency on field

measurements in future.

• Overall, the measurement procedure and data can be applied in diverse use
cases in sub-6GHz frequency bands, such as Internet-of-things (IoT), long range
massive MIMO, IoT for agriculture, drone-based communication systems

28
Questions?

2
9
Backup

3
0