A Novel On-Demand Cognitive Pilot Channel

enabling Dynamic Spectrum Allocation

Jordi Pérez-Romero, Oriol Sallent, Ramón Agustí, Lorenza Giupponi
Dept. Signal Theory and Communications
Universitat Politècnica de Catalunya (UPC)
Barcelona, Spain
E-mail: [jorperez, sallent, ramon, lorenza.giupponi]@tsc.upc.edu
Abstract—This paper addresses the implementation of the solving the spectrum allocation problem based on linear
Cognitive Pilot Channel (CPC), which has been recently programming are presented. In [6] a spectrum etiquette
proposed as a solution to assist the mobile reconfigurable and protocol for efficient coordination of radio communication
cognitive terminals in heterogeneous wireless scenarios with devices in unlicensed frequency bands using different radio
different access networks available and varying spectrum technologies is proposed. It uses a common protocol for the
allocations. The paper describes the operation of the CPC and
announcement of radio and service parameters, called the
the different approaches existing in the literature depending on
how it is mapped onto specific radio resources. Then, it focuses Common Spectrum Coordination Channel (CSCC). In [4] an
on the implementation of the CPC information delivery and overview of different spectrum sharing models is provided and
proposes the use of an on-demand CPC, which requires a a case study on how spectrum sharing can be beneficial in a
significantly lower bit rate than the broadcast approach to hierarchical cell structure is presented.
achieve similar performance.
Following a similar approach as the Spectrum Information
Keywords-Cognitive Pilot Channel; RAT selection; Dynamic Channel in [2] and the CSCC in [6], the Cognitive Pilot
Spectrum Allocation; Reconfigurability Channel (CPC) concept was recently conceived as a solution
I. INTRODUCTION to assist the mobile reconfigurable and cognitive terminal in
procedures like the Radio Access Technology (RAT) selection
The regulatory perspective on how the spectrum should be in heterogeneous networks with different access networks
allocated and utilized in future wireless scenarios is evolving available and varying spectrum allocations [7]-[9]. The CPC
towards a cautious introduction of more flexibility in the basically consists in a channel that carries the information
spectrum management together with economic considerations corresponding to the operators, RATs and frequencies
on spectrum trading. This new spectrum management allocated in a given area, so that cognitive terminals do not
paradigm is driven by the growing competition for spectrum require scanning the entire spectrum in order to find out the
and the requirement that the spectrum is used more efficiently available systems. Furthermore, terminals can make use of the
[1]. Then, instead of the classical fixed spectrum allocation to information transmitted in CPC in order to carry out several
licensed systems and services, which may become too rigid procedures, like decentralised RAT selection, optional
and inefficient, it is being recently considered the possibility software modules download for reconfigurability purposes or
to use flexible spectrum management strategies that detection of temporary unused frequency bands to enable a
dynamically allocate spectrum bands in accordance with the secondary usage of the spectrum for different applications
specific traffic needs in each area [2]-[4]. As a result of that, (e.g. establishment of an ad-hoc network, communication of
future wireless terminals will have to be reconfigurable in devices in personal area networks, etc.).
nature and will have to face the challenge of having to identify
which is the spectrum band that can be used for each specific Under this framework, the CPC channel can be regarded as
service. In turn, networks will have to adapt themselves to the a radio enabler of reconfiguration management in cognitive
varying conditions by being able to change the spectrum networks, and it is expected that it can provide benefits for the
bands that they are operating with. Consequently, future different players involved in the wireless communications
wireless systems will have to be designed under the principles arena, as summarised in the following:
of cognition, reconfigurability and adaptability [5]. a) The CPC helps the mobile terminal to select the
Several works in the literature have recently dealt with proper network depending on the specific conditions (e.g.
flexible spectrum management strategies. In [2] the desired services, RAT availability, interference conditions,
DIMSUMNet architecture is presented for coordinated, real- etc.). This provides support to Joint Radio Resource
time dynamic spectrum access based on a centralized entity Management (JRRM), enabling a more efficient use of the
called Spectrum Broker as opposite to other opportunistic, radio resources.
uncoordinated methods. In particular, this architecture b) It provides support to Reconfigurability by allowing
introduces the concepts of coordinated access band and the terminal to identify the most convenient RAT to operate
statistically multiplexed access to spectrum. Further work on with and to download in case the necessary software modules
this topic is presented in [3], where different formulations for to reconfigure the terminal capabilities.

1-4244-0663-3/07/$20.00 ©2007 IEEE

 c) It provides support to Context Awareness by helping existing RATs, operators and frequencies available in
the terminal in identifying the specific frequencies, operators each region would last a long time and therefore it
and access technologies in a given region without the need to should be avoided.
perform long time and battery consuming spectrum scanning • Terminals have a means to know the geographical
procedures. coordinates of their location thanks to some
d) It helps the network provider to facilitate dynamic positioning system.
changes in the network deployment by informing the terminals
of the availability of new RATs/frequencies, thus providing III. CPC OPERATION
support to Dynamic Network Planning (DNP) and Advanced Taking into account the above assumptions, the CPC
Spectrum Management (ASM) strategies. basically consists in a channel that carries the information
e) It helps the spectrum regulator to improve the about the operators, RATs and frequencies allocated in a given
spectrum utilisation thanks to enabling a secondary use of the area so that the terminals do not require scanning the whole
temporary unused frequency bands in a specific region. frequency band to retrieve this fundamental information.
Under this framework this paper describes the CPC A. CPC Operation Procedure
operation and focuses on the way how the CPC information The CPC operates in a certain geographical area
should be delivered to the terminals. Particularly, it proposes subdivided into meshes. A mesh is defined as an area where
the use of an on-demand CPC in which the information is only certain radio electrical commonalities can be identified (e.g. a
transmitted after a request from a mobile terminal. An certain frequency that is detected with a power above a certain
analytical evaluation reveals that the proposed strategy is an level in all the points of the mesh, etc.). The mesh is
efficient delivery strategy in comparison with other univocally defined by its geographic coordinates, and its
approaches based on broadcasting the CPC information. adequate size would depend on the minimum spatial
The rest of the paper is organised as follows. Section II resolution where the above commonalities can be identified.
describes the reference framework with the considered For simplicity, square meshes of identical dimension will be
assumptions for CPC operation. Section III focuses on the considered here. Nevertheless, other approaches could exist
CPC operation procedure, including a detailed quantification based on e.g. dynamic definition of meshes with an adaptive
of the amount of information to be transmitted. Section IV size.
analyses the CPC implementations existing in the literature The steps of the overall CPC operation procedure are
classified according to the mapping onto physical resources. described in Figure 1. After switching on, the terminal
Section V focuses on the options for the delivery of CPC determines its geographical information making use of some
information and proposes the on-demand CPC. Section VI positioning system and afterwards it detects the CPC. The
describes the evaluation model for the considered delivery CPC detection will depend on the specific CPC
options and Section VII provides some illustrative results. implementation in terms of the physical resources being used,
Conclusions are finally summarised in Section VIII. as it will be described in the next section. After detecting and
II. REFERENCE FRAMEWORK synchronizing with the CPC, the terminal retrieves the CPC
information corresponding to the mesh where it is located,
The starting assumptions that define the considered which completes the procedure. Notice that the steps 2 to 4
scenarios for CPC operation are detailed in the following can also be performed periodically to detect changes in the
points: environment due to either variations in the mobile position or
• A reconfigurable scenario with Dynamic Spectrum network reconfigurations.
Allocation (DSA) is considered. Then, the frequencies 1.- Switch-on
allocated to the different access technologies in the
different geographical regions can change during the
time. The range of spectrum may go from Fmin to
Fmax with a frequency raster of ∆f (e.g. a range of 10
GHz with ∆f=200 kHz). The specific allocation of 2.- Location determination
frequencies to RATs in each specific region is decided
by a network reconfiguration manager and supported
by different procedures, like e.g. JRRM, DNP and
ASM (see e.g. [10][11] and references therein as 3.- Detection of CPC
examples of these procedures).
• Wireless terminals with reconfigurable capabilities and
supporting different access technologies exist in the
scenario. Since DSA is applied, the terminals do not 4.- Extraction of the information corresponding
know the specific frequencies of each access to the mesh where the mobile is located
technology. In such a scenario, a scanning procedure in
the whole band from Fmin to Fmax to identify the Figure 1. CPC operation procedure
 The information retrieved from the CPC can then be used number of bits for one set of coordinates (latitude and
for different purposes, like performing RAT and operator longitude) is 20+21=41 bits.
selection in a heterogeneous reconfigurable environment or Assuming that the meshes are square, the mesh
enabling the wireless access to a secondary market, which can position will be completely defined by a set of
make use of temporary unused frequency bands for different coordinates and the dimension of one side, which can
types of applications (e.g. communication of devices in a be encoded in minutes and seconds (i.e. 12 bits),
personal area network, establishment of an ad-hoc network, assuming that the mesh size will be always much
etc.). smaller than 1º (corresponding to around 100 km). The
total number of bits for the geographical information is
B. Information transmitted in CPC
then BGEO=41+12=53 bits.
The information conveyed in the CPC for a given mesh is
illustrated in Figure 2. In particular, for each mesh, location • Operator information: For each operator, the identifier
information indicating the geographic coordinates is can be the Mobile Country Code and the Mobile
transmitted. Similarly, the CPC indicates the list of operators Network Code, consisting in 3 and 2 digits,
available in the mesh, including, for each operator, the respectively [12]. Then, the number of bits for the
available RATs and the corresponding frequency ranges per operator is BOP=20 bits.
RAT. In case that a secondary use of the spectrum is • RAT information: Let assume that a total of BRAT=4
envisaged, the corresponding information including the bits are used, so that 15 different RATs and one
frequencies available for this use can be transmitted for each reserved value for the secondary use (e.g. 1111) can be
operator using a specific field named “Secondary Use” in encoded.
Figure 2 as if it was an additional RAT. In addition to this
• Frequency information: Let assume a common
basic information, other optional terminal-dependent aspects
numbering for all frequencies in the range from Fmin
such as maximum transmitted power levels allowed depending (assumed here equal to 0 for simplicity) to Fmax=10
on e.g. whether the terminal is indoor or outdoor could also be GHz with a raster of ∆f=200 kHz, thus having a total
included in the CPC. of 50000 frequencies. This requires a total of BFREQ=16
bits per frequency.
Mesh #i: Location information
Operator #1
Then, assuming a total of NOP operators, with NRAT RATs
RAT #1
Frequency Range #1
operator and NFREQ frequency ranges per RAT, the total
Mesh #i Geographic area number of bits to be transmitted per mesh is:

(
I m = BGEO + N OP BOP + N RAT ( BRAT + N FREQ BFREQ ) )
Frequency Range #n
(1)
RAT # j

Frequency Range #n Table I provides some examples of different values of the

Secondary Use
number of bits per mesh Im for different combinations of
operators, RATs and frequencies.
Operator #m
It is worth mentioning that the number of bits Im computed
Figure 2. Information to be sent per mesh here constitute only useful information, but in order to protect
it from e.g. errors in the transmission, some redundancy
According to the basic CPC information shown in Figure should be added in the channel coding process. Similarly,
2, in the following, a computation of the number of bits to be other bits for physical procedures like synchronisation,
transmitted per mesh for the CPC is provided, depending on channel estimation, etc., should also be included, thus
how the different fields are encoded: increasing the total number of bits to be transmitted in the air
interface. Notice that these additional requirements in the
• Geographic information: This is given by latitude and
number of bits would depend on how the CPC is mapped onto
longitude coordinates. The latitude is coded as XXº
YY’ ZZ’’ plus an indicator of North/South. The range the specific physical resources.
of the degrees XXº is from 0 to 90º, thus being
TABLE I. AMOUNT OF BITS TO BE TRANSMITTED PER
encoded with 7 bits. Similarly, the range of minutes MESH FOR D IFFERENT CONDITIONS
YY’ and seconds ZZ’’ is from 0 to 59’, thus requiring
6 bits each. The North/South indicator can be simply 1 NOP NRAT NFREQ Im (bits)
bit. Therefore, the total number of bits for the latitude 3 2 10 1097
is 20 bits. Notice that the precision is 1’’, 3 5 10 2573
corresponding approximately to 30 meters. 5 5 10 4253
5 10 10 8353
Concerning the longitude, the coordinates are basically 10 10 20 32653
the same but the range of degrees is from 0 to 180º,
thus requiring one bit more. Consequently, a total of 21
bits are required for the longitude. As a result, the total
 IV. CPC MAPPING ONTO PHYSICAL RESOURCES C. Combined CPC
The different possibilities for implementation of CPC can This approach combines the in-band and the out-band
be classified, on the one hand, depending on the specific CPC. Particularly, as depicted in Figure 5, an out-band CPC
physical resources being used, and on the other hand, transmitting a minimum of information is transmitted over a
depending on the delivery mode, which specifies how the certain area. The out-band CPC basically contains a list of
information is conveyed through the CPC. Concerning the operators with only one RAT and frequency per operator
first classification, the following possibilities have been where the in-band CPC can be located. Then, the detailed
identified. information with the full operator/RAT/frequency list is
A. Out-band CPC transmitted for each mesh through the in-band CPC. As a
result, once the terminal has detected the out-band CPC, it
In this architecture the CPC is transmitted using a new retrieves the information regarding the location of the in-band
(ideally universal) frequency different from the frequencies of CPC which includes the specific information of the mesh
the existing RATs in a certain region [7]-[9], as illustrated in where it is located.
Figure 3. The main advantage of this approach is that any
CPC-compliant terminal can retrieve the information of the Notice that this approach ensures that the terminals do not
CPC no matter its supported technologies and the country have to scan the whole spectrum after switching on and that
where it is located. However, this solution suffers from two all CPC-compliant terminals can have access to the
main drawbacks. On the one hand the specific information regardless their supported RATs. However, the
frequency/frequencies to be used by CPC must be worldwide need for a common worldwide harmonised frequency for the
harmonised. On the other hand, this approach requires new out-band CPC is still present, or at least a set of CPC
infrastructure to be deployed in order to transmit the CPC frequencies should be identified for the different countries.
channel. out-band CPC

CPC Out-Band CPC

CPC in-band CPC
Out-band CPC CPC
Manager RAT 1
Manager
in-band CPC
RAT 1
RAT 2

in-band CPC
RAT 2
RAT 3

RAT 3 in-band CPC

RAT n

RAT n Figure 5. Combined CPC implementation

Figure 3. Out-band CPC implementation V. CPC DELIVERY MODE IMPLEMENTATIONS

B. In-band CPC In this section, two different CPC implementations
depending on how the CPC information is delivered to the
In this architecture the CPC is transmitted using specific
terminals are analysed. The first one is the broadcast approach
channels of existing access technologies, as illustrated in
considered in other works of the literature [7]-[9], while the
Figure 4. As a result, it does not require a new frequency to be
second one is the on-demand CPC proposed in this paper. It is
agreed and it uses existing infrastructure. Nevertheless, the
worth mentioning that these two implementations can be
main drawback from this solution is that the terminals still
applied with either the in-band, out-band or combined CPC
need to scan the spectrum to find the RAT where the CPC is
approaches described in Section IV.
located. In that sense, some possibilities would be to use some
default worldwide de-facto standard RATs supported by the A. Broadcast CPC
majority of terminals like e.g. GSM to convey CPC in fixed This strategy, illustrated in Figure 6, only uses a downlink
frequencies, but this would limit the flexibility of DSA. Notice Broadcast CPC (DBCPC) channel where the information of all
also that, with this approach, some terminals not supporting the Nm meshes in the region is broadcast periodically and
the specific RATs where the CPC is transmitted (e.g. wireless continuously. Once a terminal detects the CPC, it simply has
devices for secondary use) may not be able to find the CPC. to wait until the information corresponding to the mesh where
CPC
CPC it is located is transmitted. The total time to transmit the
RAT 1
Manager information of a mesh depends on the bit rate of the DBCPC
CPC
RAT 2 and is denoted in Figure 6 as Tm,B. It can be organised in
CPC
different time slots of duration Ts in order to simplify the
RAT 3 synchronisation of the terminals with the overall sequence of
CPC
information transmitted in the channel.
RAT n

Figure 4. In-Band CPC implementation

 BCPC (DL) Tm,OD=Ns·Ts being Ns an integer number of slots depending on
Tm,B
the bit rate of the downlink channel. It is assumed that the
CPC CPC CPC CPC CPC CPC CPC delay in getting the information from the CPC manager is
Mesh Mesh Mesh Mesh Mesh Mesh Mesh
#i-1 #i #i #i #i #i+1 #i+1 negligible. Similarly, the mobile terminal MT2 sends the
TS t request in slot #2 and receives the corresponding AI in the
downlink of slot 3. However, since the DODCPC in this slot is
Figure 6. Operation of the broadcast CPC transmitting the information of MT1, MT2 should wait until
slot #k in the figure to start receiving the information of its
B. On-demand CPC mesh. In slot #3 a collision occurs between MT3 and MT4,
The rationale of the proposed on-demand CPC approach is and therefore the AI in the subsequent slots indicate a Null
that, if all the CPC information should be continuously value, reflecting that no request has been received. Then, the
broadcast for all the meshes, this would require either a long terminals will wait a random retransmission time. In the
time or a wideband channel, particularly, if the mesh size is example, MT3 retransmits the request in slot #k+1.
small. However, if the number of terminals that are located in DODCPC+AICPC (DL)

each mesh is small, this information will be in practice most of T m,OD

the time unused. Consequently, it may become more efficient

CPC CPC CPC CPC CPC CPC CPC
from both power and bandwidth consumption point of view, to

AI Null

AI Null

AI Null

AI Null
AI 1

AI 2

AI 3
Mesh Mesh Mesh Mesh Mesh Mesh Mesh
MT 1 MT 1 MT 1 MT 1 MT 2 MT 2 MT 2
transmit the information only when needed and requested by a
t
terminal. RACPC (UL)
TS
Collision
TS
The on-demand CPC makes use of both the uplink and MT4

downlink components and it consists in the following logical MT1 MT2 MT3 MT3

channels: #1 #2 #3 #4 #k-1 #k #k+1 #k+2 t

• Random Access CPC (RACPC): It consists of an Figure 7. Operation of the on-demand CPC
uplink slotted channel where the mobiles operating
with CPC send requests to retrieve the CPC The proposed implementation of the on-demand CPC
information corresponding to their meshes. Each including both an uplink and a downlink channel enables a
request basically contains an indicator of the wider range of CPC-based applications in addition to the
geographical coordinates of the mobile terminal. An retrieval of the information about operators, RATs and
operation according to a simple access protocol such as frequency lists. For example, the CPC could eventually be
S-ALOHA can be envisaged for this channel. used by the terminals to retrieve other terminal-dependent
• Acquisition Indicator CPC (AICPC): This downlink information, such as software downloads to enhance the
channel follows the same slotted structure of the uplink reconfigurable terminal capabilities. Furthermore, the uplink
RACPC and is devoted to indicate that a request has channel could also be used to ensure that the information has
been successfully received. The channel consists in been delivered correctly thus improving the integrity and the
Acquisition Indicators (AI) each one indicating the security in the transmitted information.
identifier of the terminal whose request has been
On the other hand, notice that the interaction with the
received or the value Null if no request has been
received. mobiles in the on-demand CPC allows the network operator
and the spectrum regulator having eventually a higher control
• Downlink On-Demand CPC (DODCPC): This of the terminals accessing CPC than if the broadcast approach
downlink logical channel is used to transmit the CPC was used. In that sense, it is easier to fit the CPC operation
information corresponding to the mesh of each within specific business models and exploitation plans e.g. for
received request from a Mobile Terminal (MT). a controlled secondary use of the spectrum.
The operation of these channels is illustrated in Figure 7. VI. EVALUATION MODEL
The uplink and the downlink channels are organised in slots of
duration TS. The AICPC and the DODCPC are multiplexed on In the following, a model to evaluate the performance of
the same time slots by making use of different fields of a the two approaches for CPC delivery implementation
certain burst structure. In Figure 7, the terminal MT1 sends a described in Section V is provided. Let assume a scenario in
request in slot #1. This request simply contains the which each CPC transmitter sends the information
geographical coordinates of the terminal and a short random corresponding to Nm meshes. A number of wireless devices or
identifier. Since there is no collision in the transmission, the terminals require getting the CPC information corresponding
slot #2 in the AICPC indicates that MT1 request has been to the mesh where they are located. Then, let λ (requests/s) be
successfully received by means of the Acquisition Indicator the arrival ratio of requests corresponding to these devices in
(AI) including the random identifier sent by MT1. Then, the the whole area of one CPC transmitter, assumed to follow a
transmission of the CPC information corresponding to the Poisson distribution. Terminals are uniformly distributed in
mesh of MT1 starts in the DODCPC during a total of the area and meshes are of the same size, so the arrival rate is
the same in all the meshes. Let Im be the total number of part can be computed as [13]:
information bits to be transmitted for a single mesh. For
comparison purposes, it is assumed that the slot duration TS 1+ q 1 1 
and the downlink net bit rate of the CPC channel Rb (i.e. DRA = TS  + −  (4)
including only information bits, without considering other bits  qps 2 q 
for channel coding, synchronisation, etc.) is the same in both where ps is the success access probability (i.e. the probability
the broadcast and the on-demand CPC. that there is no collision in one slot) and is given by:
A. Broadcast CPC
In this case, since the downlink channel is slotted, the
ps = e − λT TS (5)
number of slots to transmit the information of one mesh is
where λT is the total access rate including retransmissions and
given by:
can be related with the arrival rate λ as:
 I 
N S ,B =  m  (2) λ = λT e − λ T
T S
(6)
 RbTS 
From (6), it is possible to derive the value of λT as:
Where x denotes the lowest integer high than or equal to
x. So the total transmission time of one mesh in the broadcast W ( λTS )
CPC is Tm,B=NS,B·TS. λT = − (7)
TS
Then, assuming that one CPC transmitter sends the
information of Nm meshes, the total CPC broadcast period is where W(x) is the Lambert W-function, which can be
T=Tm,B·Nm. Consequently, assuming that users are always computed from its series expansion as [14]:
synchronised with the CPC channel and that a user requires
( −1) n n− 2 x n
n −1
∞
getting the CPC information randomly with equal probability
W ( x) = ∑ (8)
in any time between 0 and T and the arrival can correspond to
any mesh with equal probability, the average delay in getting
n =1 ( n − 1)!
the information will be: After the random access phase, the system can be
modelled as a M/D/1 queue with arrival rate λ and service
N  time Tm,OD, corresponding to the transmission time of the
DB = Tm, B  m + 1 (3)
 2  information of one mesh. Assuming that the bit rate of the
downlink channel is Rb, that it is organised in slots of duration
Notice that the first term in (3) accounts for the average TS and that in each slot a number of IAI bits should be
time until the start of the desired mesh (i.e. T/2) while the transmitted for the AICPC, the total number of slots for
second term accounts for the time to transmit the information transmitting the information of one mesh is given by:
of this mesh.
B. On-demand CPC  Im 
N S ,OD =   (9)
In this case, the total delay in getting the information of the  RbTS − I AI 
desired mesh can be computed as the sum of two delays, as
depicted in Figure 8, corresponding to the uplink access phase Then, the service time to transmit the information of one
to send the request and the time to send the information of the mesh is Tm,OD=NS,OD·TS and the total delay including service
required mesh in the DODCPC. and queuing time is given by [15]:

λ (reqs/s)
λ (reqs/s) Rb (b/s)
λTm2,OD
DS = + Tm ,OD (10)
2 (1 − λTm ,OD )
Random Access
DODCPC: M/D/1 system Finally, the total delay corresponding to the on-demand
CPC is computed as:
Figure 8. Delay model of the on-demand CPC
DOD = DRA + DS (11)
For the random access phase a simple S-ALOHA model is
considered, with slots of duration TS. It is assumed that VII. RESULTS
terminals know at the beginning of one slot if a collision has
occurred in the previous slot. Then, after a collision, a This section presents some illustrative results for the CPC
geometric back-off is used for retransmissions, meaning that operation. For illustration purposes, the amount of information
retransmissions are carried out in the next slot with probability corresponding to a mesh is Im=4253 bits, corresponding to one
q. Taking this into account, the delay of the random access of the combinations in Table I. The rest of parameters are
detailed in the following:
 • Time slot duration: TS=10 ms. This would be the case load also increases with the range. Consequently, the figure
of e.g. one UMTS frame or approximately two GSM shows that both the delay of the on-demand CPC and the
frames. broadcast CPC increase with the range, but the delay
experienced in the broadcast case is in general much higher
• Bit rate of the downlink CPC: Rb=10 kb/s. Notice that,
in case of the broadcast CPC, the 10 kb/s are devoted than that of the on-demand CPC. Assuming a maximum delay
to CPC information while in the case of the on-demand of e.g. 5 s in getting the CPC information, it can be observed
CPC the 10 kb/s are shared between AICPC and that, for these conditions, the maximum range for CPC
DODCPC. In any case, Rb corresponds to the net bit operation with the on-demand approach is around 1.05 km. On
rate of information bits, without including redundancy the contrary, for the broadcast CPC the maximum range for a
bits for channel encoding, synchronisation bits, etc. delay below 5 s would be only about 300m. This observation
reflects that, for a given range, the broadcast CPC would
• Concerning the uplink requests in the RACPC the require a higher bit rate than the on-demand CPC in order to
mobiles simply have to transmit a random identifier achieve comparable performances. Notice that the difference
(e.g. 5 bits like in the GSM random access procedure) could even be higher if a lower maximum delay bound below
and a set of coordinates, thus having a total of
5 s was set.
Ir=5+41=46 bits.
It is worth mentioning that, for a given mesh size, it would
• The number of bits for the acquisition indication is
be preferred to operate the CPC with high transmitter ranges,
simply the random identifier of the mobile whose
request has been received, thus requiring IAI=5 bits. because this reduces the infrastructure cost required for the
deployment, consequently, the likely operating region of the
• A scenario with a homogeneous user density of η CPC would be the right side of the horizontal axis, where the
users/km2 is considered. Each user or wireless device differences between on-demand CPC and broadcast CPC are
generates a total of λu=0.0003 requests/s to access the higher.
CPC (i.e. around 1 request per hour).
2
• The area covered by a CPC transmitter is assumed to
1.8
be circular with a radius R km. Then, the total amount
1.6
of requests per second in this area is given by: Nm=1, OnDemand
1.4
Nm=1, Broadcast

λ = ηλuπ R 2
(12) 1.2
Delay (s)

Figure 9 and Figure 10 plot the performance in terms of 0.8

delay for the broadcast CPC and the on-demand CPC in the 0.6

case of a CPC transmitter with R=1km as a function of the 0.4

total user density η for a fixed number of meshes, Nm=1 (i.e. 0.2

0
mesh area of around 1700m×1700m) and Nm=30 (i.e. mesh 0 500 1000 1500 2000
area around 320m×320m), respectively. It can be observed User density (users/km2)

how the on-demand CPC outperforms the broadcast CPC

particularly when the number of meshes increases, because in Figure 9. Total delay as a function of the user density for the case Nm=1
this case the broadcast period becomes longer and therefore
the users have to wait a long time until they retrieve the 8

information corresponding to their mesh. Notice also that the 7

on-demand CPC performance is sensitive to the user density,
6
because when the number of requests increases the queuing Nm=30, OnDemand
waiting time also becomes longer. In that sense, there is a 5 Nm=30, Broadcast
Delay (s)

fixed limit given by the M/D/1 performance because it should 4

be fulfilled the condition λ ·Tm,OD<1. Nevertheless, the 3

performance in this case could even be improved either by 2
using multi-cast mechanisms (e.g. grouping user requests
1
corresponding to the same mesh in a single downlink CPC
transmission in DODCPC) or by adding an additional time 0
0 500 1000 1500 2000
slot for downlink CPC. In any case, these issues are left for User density (users/km2)
future work.
Figure 10. Total delay as a function of the user density for the case Nm =30
Figure 11 plots the performance as a function of the CPC
transmitter range R. In this case, the mesh size is assumed to
The behaviour of the CPC depending on the value of Rb is
be fixed to 100m×100m, so the larger the CPC transmitter further analysed in Figure 12, which plots the required
range the larger the number of meshes. Similarly, the user downlink CPC net bit rate Rb for the two approaches as a
density is equal to η=2000 users/km2, and therefore the total function of the CPC transmitter range if a maximum delay of
5s in retrieving the CPC information was set. The same required. In that sense, the value of 100m×100m selected as a
conditions in terms of user density and mesh size as in Figure reference in this study appears to be an appropriate order of
11 are considered. It can be observed how the on-demand magnitude of the mesh size to proof the concept of the on-
CPC requires around one order of magnitude less bit rate than demand CPC. A more detailed analysis to determine the
the broadcast CPC for comparable delay performance (e.g. the appropriate mesh size depending on the interference
required bit rate of the on-demand CPC is around tenths of characteristics of each region would require the definition of
kb/s while that of the broadcast CPC is around hundreds of specific scenarios in terms of access technologies available,
kb/s). Consequently, a more efficient CPC implementation deployment, propagation conditions, etc. and is therefore left
with the on-demand CPC follows. Notice that the lower bit for future work. In any case, it is worth mentioning that the
rate would make easier the implementation of the proposed lower the mesh size, the higher the difference between the on-
approach either with an out-band CPC (i.e. it can be easier to demand CPC and the broadcast CPC in Figure 12, because the
harmonise a frequency for a narrowband channel) or with an required bit rate in the on-demand CPC is not sensitive to the
in-band CPC (i.e. the on-demand CPC can be easily mapped mesh size, while in the broadcast CPC the number of meshes
onto existing low bit rate RATs, like e.g. GSM, while the increases the broadcast period and thus the bit rate
broadcast CPC would require broadband RATs). requirement to ensure a certain delay bound.
100 180
On Demand R=500 m
90
Broadcast 160 R=1 km
80
R=1.5 km
140
70 R=2 km
120
60

Bit rate (kb/s)

Delay (s)

50 100

40 80

30 60
20
40
10
20
0
0 0.2 0.4 0.6 0.8 1 1.2 0
CPC range R (km) 0 2500 5000 7500 10000 12500 15000 17500 20000
I m (bits)

Figure 11. Total delay as a function of the CPC range for a mesh size of
100m×100m Figure 13. Required net bit rate for the on-demand CPC to ensure a
maximum delay of 5s with a mesh size of 100m×100m as a function of the
1000
number of bits to be transmitted per mesh Im for different transmitter ranges R.
On Demand
Broadcast 18

16
100
Bit rate (kb/s)

14

12
RbBroad /RbOD

10 R=500 m
10
R=1 km
8
R=1.5 km
6 R=2 km

1 4
0 0.5 1 1.5 2
2
CPC range R (km)
0
0 2500 5000 7500 10000 12500 15000 17500 20000
Figure 12. Required CPC net bit rate to ensure a maximum delay of 5s as a I m (bits)
function of the CPC range for a mesh size of 100m×100m.

On the other hand, with respect to the mesh size, if small Figure 14. Ratio between the required net bit rate for broadcast CPC and for
meshes are considered a better spectrum usage can be on-demand CPC to ensure a maximum delay of 5s with a mesh size of
100m×100m as a function of the number of bits to be transmitted per mesh I m
achieved because in this case there is more granularity when for different transmitter ranges R.
identifying the different interference conditions in each region.
Notice that the inclusion of low range technologies (e.g. Finally, in order to analyse the sensitivity with respect to
deployment of WLAN hotspots) will tend to introduce more the amount of bits to be transmitted per mesh Im, Figure 13
variability in the way how the different frequencies are used in plots the required downlink net bit rate with the on-demand
a given region and therefore smaller mesh sizes could be CPC to ensure a maximum delay of 5s as a function of the
number of bits Im for different CPC transmitter ranges with a facilities. In that sense, it may be easier to fit the CPC
mesh size of 100m×100m, and Figure 14 plots the operation within specific business models and exploitation
corresponding gain of the on-demand CPC with respect to the plans e.g. for a controlled secondary use of the spectrum.
broadcast CPC (measured as the ratio of bit rate requirements
for the two approaches to ensure the maximum delay of 5s). ACKNOWLEDGMENTS
From Figure 13, it can observed that the bit rate requirements This work was performed in project E2R II which has
increase linearly with the number of bits per mesh for the on- received research funding from the Community's Sixth
demand CPC. This is also valid for the broadcast CPC case, Framework program. This paper reflects only the authors'
although in this case the bit rate requirements are between 12 views and the Community is not liable for any use that may be
and 16 times (i.e. more than one order of magnitude) higher made of the information contained therein. The contributions
than those of the on-demand CPC, as shown in Figure 14. As a of colleagues from E2R II consortium are hereby
result of this behaviour, Figure 14 reveals as a general trend an acknowledged. The work is also partially funded by the
asymptotically increase in the gain achieved by the on-demand Spanish Research Council under COSMOS (Ref. TEC2004-
CPC with the number of bits Im, being the gain approximately 00518) grant.
constant for values of Im higher than around 4000-5000 bits
for the considered transmitter ranges in the figure. On the REFERENCES
contrary, for small values of Im the gain is smaller because of [1] J. A. Hoffmeyer, “Regulatory and Standardization Aspects of DSA
the fixed overhead of IAI bits per slot required for the AICPC Technologies – Global Requirements and Perspectives”, 1st IEEE
international Symposium on New Directions in Dynamic Spectrum
with the on-demand approach, which becomes more Access Networks (DySPAN), Baltimore, MD, November, 2005.
significant for low values of Im. [2] M. Buddhikot, P. Kolodzy, S. Miller, K. Ryan, J. Evans “ DIMSUMNet:
New Directions in Wireless Networking Using Coordinated Dynamic
VIII. CONCLUSIONS Spectrum Access”, IEEE International Symposium on a World of
Wireless, Mobile and Multimedia Networks (IEEE WoWMoM),
This paper has analysed different implementations of the Taormina/Giardini Naxos, Italy, June, 2005.
Cognitive Pilot Channel, recently proposed in the literature as [3] M. Buddhikot, K. Ryan, “Spectrum Management in Coordinated
an enabler for supporting the operation of reconfigurable Dynamic Spectrum Access Based Cellular Networks”, 1st IEEE
terminals in scenarios with DSA in which the allocation of international Symposium on New Directions in Dynamic Spectrum
Access Networks (DySPAN), Baltimore, MD, November, 2005.
frequencies to RATs is not fixed and known a priori. Then, the
[4] H. R. Karimi, T.W.Ho, H. Claussen, L. G. Samuel, “Evolution towards
use of a CPC channel avoids the scanning time for detecting dynamic spectrum sharing in mobile communications”, IEEE PIMRC
available RATs and frequencies at the mobile terminal. Conference, Helsinki, September, 2006.
Furthermore, thanks to identifying the used frequency bands [5] “End to End Reconfigurability Phase II (E2R2)”, IST Project,
in a given region, the CPC also allows a secondary use of the http://e2r2.motlabs.com
temporary unused bands for different applications like e.g. [6] D. Raychaudhuri, X. Jing “A Spectrum Etiquette Protocol for Efficient
Coordination of Radio Devices in Unlicensed Bands”, 14th IEEE 2003
establishment of ad-hoc networks, communication of devices International Symposium on Persona1,lndoor and Mobile Radio
in personal area networks, etc. Communication Proceedings (PIMRC), Beijing, September, 2003.
Focusing on the implementation of the delivery of CPC [7] P. Houzé, S. Ben Jemaa, P. Cordier “Common Pilot Channel for Network
Selection”, IEEE VTC in Spring Conference, Melbourne, May, 2006.
information, the paper has proposed an on-demand CPC as [8] E. Mohyeldin, J. Luo, J. Pan, P. Slanina, “Common Pilot Method
opposite to the broadcast mechanism. Both approaches have Enabling Network Assisted Fast Scanning for Reconfigurable
been studied by means of an analytical model and it has been Terminals”, 4th Karlsruhe Workshop on Software Radios, Karlsruhe,
obtained that the proposed approach is able to achieve the March, 2006.
same performance in terms of delay to retrieve the information [9] P. Cordier et al., “E2R Cognitive Pilot Channel concept”, IST Summit,
Mykonos, June, 2006.
requiring a significantly lower net bit rate of the CPC channel.
[10] L. Giupponi, R. Agustí, J. Pérez-Romero, O. Sallent “Joint Radio
Specifically, differences in around one order of magnitude Resource Management algorithm for multi-RAT networks”, IEEE
have been observed in the required bit rate for the two Globecom, Saint Louis, USA, November, 2005.
approaches. [11] J. Nasreddine, O. Sallent, J. Pérez-Romero, R. Agusti “Dynamic
Spectrum Management Methodology for WCDMA Systems Based on
The proposed on-demand CPC can be implemented either Inter-Cell Interaction Approach”, IEEE WPMC Conference, San Diego,
using the out-band CPC or the in-band CPC architectures. USA, September, 2006
Furthermore, it enables a wider range of CPC-based [12] 3GPP TS 23.003 “Numbering, addressing and identification”
applications. For example, the CPC could eventually be used [13] Y. Yang, T-S.P.Ym, “Delay Distributions of Slotted ALOHA and
by the terminals to retrieve terminal-dependent information, CSMA”, IEEE Transactions on Communications, Vol. 51, No. 11,
November, 2003, pp. 1846-1857.
such as software downloads to enhance the reconfigurable
[14] E. W. Weisstein,. “Lambert W-Function.” From MathWorld--A Wolfram
terminal capabilities and to ensure that the information has Web Resource. http://mathworld.wolfram.com/LambertW-Function.html
been delivered correctly thus improving the integrity and the [15] L. Kleinrock, Queueing Systems. Volume I: Theory, Ed. John Wiley &
security in the transmitted information. Similarly, thanks to Sons, 1975.
the interaction with the mobiles existing in this approach, it is
possible for the network operator and the spectrum regulator
to have a higher control of the terminals using the CPC