A Novel Mesh Division Scheme using Cognitive Pilot

Channel in Cognitive Radio Environment

Qixun ZHANG1, Zhiyong FENG2, Fang TIAN1 and Zheng HU2
Key Laboratory of Universal Wireless Communications, Ministry of Education
Wireless Technology Innovation Institute (WTI), Beijing University of Posts and Telecommunications
Beijing, P.R.China, 100876
e-mail address:1{zqx830311, tianfang.tf.mail}@gmail.com, 2{fengzy, huzheng}@bupt.edu.cn

Abstract—In the context of B3G heterogeneous environment, the potential capability to complete the reconfiguration
convergence and cooperation of different wireless network autonomously by learning from the changing environment [5].
technologies are inevitable. The concept of Cognitive Pilot
Channel (CPC), which is one of the candidate transmission Taking into account both the reconfigurability concept and
solutions for the dynamic spectrum sharing of network the CR concept, the traditional fixed spectrum assignment
information in the Cognitive Radio (CR) environment, is which grants exclusive use to licensees is unsuitable anymore
proposed to provide the user equipments (UEs) with the in the rapid changing heterogeneous radio environment.
necessary network information for both the switch-on and the Therefore, flexible spectrum management technology should
on-going phases. Optimal mesh division problem appears based be used and new dynamic spectrum allocation (DSA) schemes
on the basic assumptions in CPC concept that the geographical are proposed accordingly [6]-[7], in order to make efficient
region is organized in meshes and the coverage of different Radio spectrum utilization in terms of the traffic condition and
Access Technologies (RATs) overlaps with each other. In this deployment of different RATs in the specific area.
paper, a novel optimal mesh division scheme is designed, taking
into account both the Global Position System (GPS) localization With these dynamic spectrum allocation schemes, user
shift scenario and multi-RATs scenario. Both the error equipments (UEs) should be intelligent enough to have the
probability and the information loss ratio are investigated and capability of spectrum awareness. Different technologies have
their impacts to the optimal mesh division scheme are studied by been proposed to provide the UEs with the ability of spectrum
using the Analytic Hierarchy Process (AHP) and Grey Relational awareness necessarily. In the literatures, some sensing
Analysis (GRA) algorithms. Optimal mesh division scheme is mechanisms are used to identify the network information, such
verified by numerous simulation results. as the RATs and unoccupied frequency, by UEs independently
or cooperatively with other terminals [7]-[8]. Shortcomings of
Keywords-B3G; cogntive pilot channel; cogntive radio; global these mechanisms are long measurement time, high system
positioning system; reconfigurability complexity and high power consumption in terminals.

I. INTRODUCTION Other alternatives are proposed by using the signaling

channels to transmit the network information, such as the
Recent years have witnessed the evolution of many new spectrum vacancy, RAT occupancy and available networks in
concepts and new technologies towards a heterogeneous and the specific area. The concept of Cognitive Pilot Channel
convergence wireless network which is beneficial to users, (CPC), which was proposed within the E2R project [9] and
operators and regulators. continued to be one of the highlights within the E3 project [14],
Many discrete technologies are migrating towards a global is one of the solutions to broadcast the network information to
technology, often called Beyond 3rd generation (B3G). As a the UEs through a specific signaling channel either by
major contributor to this global B3G technology, the concept of broadcast mode or on-demand mode, enabling dynamic
reconfigurability is proposed by [1]. Reconfigurability is an spectrum sharing in CR environment. The CPC concept is
evolution of the “software-defined radio” [2], which implies proposed and implemented through a public signaling channel,
the dynamic adaptation ability of terminals to the set of Radio either by broadcast mode or on-demand mode [15]. In the
Access Technologies (RATs) with appropriate conditions for former research, the out-band CPC scheme is brought forward
specific services. Reconfigurability includes the changing of to provide the vital network information to UEs in the switch-
RATs, spectrum bands [3], reconfiguration of network on period by using a common world wide frequency [9] or a
parameters, reconfigurable protocols in both terminal side and collection of several available frequency bands [20].
network side, in order to obtain the optimal global utilization of Furthermore, the in-band CPC scheme is proposed to provide
radio resources. more detailed network information in the on-going period by
using the specific frequency band of existing RATs. Both
As first proposed by Mitola in 1991, the Software-defined schemes are implemented in a wide area divided by a great
Radio (SDR) is a multi-band radio, supporting multiple air number of meshes [9], enhancing the accuracy and efficiency
interfaces and protocols for reconfigurable purpose [4]. Also of network information delivery. As assumed in [15], the
proposed by Mitola in 1998, the concept of Cognitive Radio square mesh of identical dimension is considered, but there is
(CR), which is based on the software radio platform, is a no further definition or analysis on the appropriate mesh size
context-aware intelligent radio technology which has the and the factors affecting the mesh division scheme.

This work was sponsored by End-to-end Efficiency (E3) project (FP7-

ICT-2007-216248) within Community’s Seventh Framework program, Key
Project of National Natural Science Foundation of China (60832009) and
National Basic Research Program of China (973 Program) (2009CB320406).

978-1-4244-2515-0/09/$25.00 ©2009 IEEE

 Considering the UE in a specific mesh, it is a challenge for small square size meshes equally. Overlapped meshes are
the UE to obtain the network information in accordance to its shown in the general scenario in Fig. 1. Each RAT covers a
geographical location through the CPC channel. One of the specific area which is assumed as a circle. The coverage of
possible solutions to deal with this challenge is to use the each RAT is different and the number of meshes covered by
Global Positioning System (GPS) technology [9]. However, the each RAT will be variable, caused by different mesh division
positioning technique based on the GPS technology has the sizes. Each RAT and UE will be positioned by the GPS
problem of GPS localization shift which will lead to the technology with the latitude and the longitude as an index.
localization error to the UEs. Due to the parameters affecting
the GPS precision, such as the satellitic clock error, ionospheric
delay error [10]-[11] and multi-path error, the precision of GPS
localization system is around 10 to 100 meters in practical use
[12]. This effect will affect the optimal mesh division scheme
and will be studied thoroughly.
Furthermore, the network information transmitted through
the CPC channel should be accurate and efficient for the UEs
in the specific mesh region. However, problem appears with
the multi-RATs overlapped meshes scenario. Because the
coverage of each RAT is different from that of other RATs in a
specific mesh, some of the RATs would cover a large
percentage of the whole mesh and its network information
would be of great importance to the UEs. Furthermore, the
network information transmitted by CPC should be precisely
and efficiently, so some unimportant network information
would be omitted due to its small percentage of coverage in the
specific mesh. The concept of information filter is proposed
accordingly in the multi-RATs overlapped mesh scenario.
Under these above considerations, this paper presents a
novel research on the optimal mesh division scheme using the Figure 1. General mesh division scenario
CPC channel, taking into account both the GPS localization
shift scenario and the multi-RATs overlapped meshes scenario.
The purpose of this optimal mesh division scheme is to deliver III. ERROR PROBABILITY CAUSED BY GPS LOCALIZATION
the network information to the UE in the specific mesh region SHIFT
precisely and efficiently. Our work will focus on the impact of In this section, the error probability caused by GPS
parameters such as the GPS localization shift parameter and the localization shift with a precision shift around 10 to 100 meters
information filter parameter to the optimal mesh division is calculated and its effect to the mesh division size is also
scheme which is based on the Analytic Hierarchy Process analyzed.
(AHP) [16] and Grey Relational Analysis (GRA) [17]
algorithms [19]. To the author’s best knowledge, this is the first
time that the impacts of both GPS localization shift and the
multi-RATs overlapped meshes are taken into account in the
area of the optimal mesh division size using CPC channel in
the CR environment. It is worth noting that the research work
is a good exploration in this field and will pave the way for
further study.
The rest of this paper is organized as follows. In Section II,
the general scenario for mesh division in CPC is proposed.
Section III will focus on the localization error caused by the
GPS localization shift. Section IV will focus on the information
loss ratio caused by the information filter parameter in the
multi-RATs overlapped scenario. In Section V, optimal mesh
division scheme based on the AHP and GRA algorithms will
be investigated. Numerous simulation results will be shown in
Section VI. Finally, the last section concludes the paper.

II. GENERAL SCENARIO OF MESH DIVISION IN CPC

In this section, a generic scenario of mesh division in CPC Figure 2. Error probability calculation based on GPS localization shift
with GPS localization shift and multi-RATs overlapped meshes scenario
is depicted in Fig. 1. We assume that the region is divided into
 In the process of theoretical calculation of error probability 3 in this scenario. As shown in Fig. 3, mesh j is covered by
cause by the GPS shift, the rectangular coordinate system is three different RATs which are RAT1, RAT2 and RAT6. To
proposed and different zones are divided accordingly. We interpret the scenario more vividly, the overlapped area
define the parameters as follows: parameter Z represents the covered by different RATs is shown in Fig. 3 (b). The set of
GPS localization shift, parameter Y represents the mesh points covered by RAT1, RAT2 and RAT6 is represented by S1,
division size (as assumed the mesh is square) and UE1(x1,y1) S2 and S6 respectively. And the set of points covered by both
represents the localization of the UE. The UEs will follow a RAT1 and RAT2 is represented by S1,2. The S1,2,6 is used to
uniform distribution in each mesh covered by CPC. As shown denote the set of points covered by RAT1, RAT2 and RAT3
in Fig. 2 (a), the possible error zone of UE1 is shown in the together.
peripheral area of the mesh with a width of Z. The location
error zone of UE outside the square mesh area is the key area
that should be calculated carefully. In order to deal with this
challenge, different error probabilities caused by UEs in
different positions are investigated in detail as shown in Fig. 2
(b)-(d).
To be specific, the error probability density parameters of
three error zones are represented by perr_1, perr_2 and perr_3. And
the mathematic formulas are shown in (1)-(3), respectively.

⎡ ⎛x ⎞ ⎤
perr _1 = ⎢ Z 2 * arccos ⎜ 1 ⎟ − x1 * Z 2 − x12 ⎥ π Z 2 (1) Figure 3. Multi-RATs overlapped scenario
⎣ ⎝Z ⎠ ⎦

⎧π Z 2 Z 2 ⎡ ⎛x ⎞ ⎛ y ⎞⎤ x y ⎫ (2) Due to the limitation of the convergence of network

perr _ 2 = ⎨ + * ⎢arccos ⎜ 1 ⎟ + arccos ⎜ 1 ⎟⎥ − x1 * y1 − 1 * Z 2 − x12 − 1 * Z 2 − y12 ⎬ π Z 2
⎩ 4 2 ⎣ ⎝Z⎠ ⎝ Z ⎠⎦ 2 2 ⎭ information in one specific mesh, not all the information
should be transmitted to the UEs in the mesh. On this basis,
⎧ ⎡ ⎛x ⎞ ⎛ y ⎞⎤ ⎫ (3) only the most important and representative RAT’s information
perr _ 3 = ⎨ Z 2 * ⎢arccos ⎜ 1 ⎟ + arccos ⎜ 1 ⎟ ⎥ − x1 * Z 2 − x12 − y1 * Z 2 − y12 ⎬ π Z 2
⎩ ⎣ ⎝Z⎠ ⎝ Z ⎠⎦ ⎭ will be significant for transmission to the UEs, while other
Based on the error probability density results, the error unimportant information will be discarded. A novel standard to
probability of each zone is calculated by using the definite decide the importance of the network information in one
integral formula based on the specific area of each zone. Then, specific mesh has been brought forward in this paper, and the
the error probability for each mesh can be calculated by adding information filter F is defined in order to decide which
up the error probability in three different zones, where the Pr information should be transmitted to the UEs, where 0 < F ≤ 1 .
represents the error probability as shown in (4). Let us assume that if the coverage of one network in the mesh
is lower than a specific percentage as shown in Si / S < F , the
Z Y −Z Z Z 2 − x12 Z Z
∫ ∫
0 Z
perr _1dy1dx1 + ∫
0 ∫
0
perr _ 2 dy1dx1 + ∫
0 ∫ Z 2 − x12
perr _ 3 dy1dx1 information of network i ( 1< i < Nj ) should be omitted and will
Pr = 4 *
Y2 (4) not be transmitted to the UEs.
16YZ − 3Z 2
= Furthermore, the set of indexes of networks in mesh j
6π Y 2
Based on the mathematical analysis, the error probability Pr which are not filtered out by the parameter F is defined by the
is related to the GPS localization shift parameter Z and the formula B j = {i | S i / S ≥ F } . The area of unfiltered networks
mesh division size parameter Y. And its impact on the mesh is denoted by the parameter S U F , j , as shown in (5). Then, the
division size will be evaluated by numerous simulation results
in this paper. information precision ratio in terms of the information filter
parameter F is defined by PF , j , as shown in (6). Taking into
IV. INFORMATION LOSS RATIO IN MULTI-RATS SCENARIO account the whole area, the average information precision ratio
is depicted by PF , as shown in (7), where the N is the total
In this section, information loss ratio in multi-RATs
overlapped scenario is proposed with analysis as depicted in number of meshes in this region. And the average information
Fig. 1. Due to the variable coverage of each RAT and the loss ratio is defined by Fr as shown in (8).
changeable mesh division size, it is of great importance to
1
design a mesh division scheme based on this multi-RATs
scenario. For the theoretical calculation of information loss
SUF , j =
Nj
∑
i∈ B j
Si (5)

ratio, mesh j is selected as an example to illustrate the

calculation process as shown in Fig. 3. PF , j = SUF , j S (6)

In mesh j, there are three RATs overlapped with each other.

⎛ N ⎞
Let us assume that the area covered by each RAT is PF = ⎜ ∑ PF , j ⎟ N (7)
represented as Si, where the parameter i depicts the index of ⎝ j =1 ⎠
different RATs. Moreover, we also assume that the area
Fr = 1 − PF (8)
covered by mesh j is the collection of S, where S=Y2. And the
number of RATs in mesh j is defined as the Nj, which equals to
 V. NOVEL MESH DIVISION SCHEME
In this section, the novel optimal mesh division scheme is
proposed to analyze the impact of GPS localization shift
scenario and multi-RATs overlapped scenario to the mesh
division scheme by using APH and GRA algorithms [16]-[19].
In virtue of the AHP algorithm, which is a powerful and
flexible mathematical decision making technique, taking into
account both qualitative and quantitative aspects of decisions.
The AHP algorithm helps decision makers find out the best
decision by reducing the complex decision into a series of Figure 5. Structure of AHP algorithm for optimal mesh division scheme
decision factors and synthesizing their importance to the object.
Besides, the GRA algorithm is good at finding the best option
through comparing each option to the ideal one by the Grey TABLE I. AHP MATRIX AND WEIGHTS OF OPTIMAL MESH DIVISION
relational coefficient (GRC) value. The more is the similarity SCHEME
between the chosen option and the ideal option, the more A Pr Fr W
preferable is the chosen option. On this basis, the optimal mesh
Pr 1 1 1/2
division scheme is proposed and illustrated in detail. The
optimal mesh division scheme flow is proposed which include Fr 1 1 1/2
the key steps of both the AHP algorithm and the GRA The weights of the factors are achieved through calculating
algorithm, as shown in Fig. 4. the eigenvector W of the matrix A by (9), where λmax is the
maximum eigenvalue of matrix A , n equals to 2, Wi represents
the weight of the ith factor. The consistency index (CI) of the
AHP matrix is calculated by (10)-(11). If CI value is equal to
zero, the matrix is perfectly consistent. Otherwise, the ratio of
CI to random index (RI) for the same dimension matrix which
is called the consistency ratio (CR) is taken into account as
shown in TABLE II [16]. Adjustment of the AHP matrix is
needed when CR > 0.1.

AW = λmaxW (9)

CI = (λmax − n) /(n − 1) (10)

1 n
λmax = ∑ ( AW )i / Wi
n i =1
(11)

TABLE II. CONSISTENCY INDEX

Figure 4. Flow of optimal mesh division scheme Dimension 1,2 3 4 5 6 7 8 9 …
RI 0 0.58 0.9 1.12 1.24 1.32 1.41 1.45 …
Given the layout of a network scenario, both parameters of
GPS localization shift error Pr and the information loss ratio Fr Based on the AHP matrix and weights of parameters on the
are calculated based on the formulas proposed in this paper. second level such as the GPS localization shift error Pr and the
Then, the pairwise comparison is performed to create ratio information loss ratio Fr in TABLE I, the optimal mesh
scales aiming at establishing relations between parameters and division is to choose the ideal scheme among all candidate
the AHP matrix is built by taking into account key parameters schemes with the smallest GPS shift error and information loss
in three levels which are Object level, Criteria level and ratio based on the GRA algorithm.
Alternative level as shown in Fig. 5. And numbers from 1 to 9 The performance of each candidate mesh division scheme
are used to present equally, weakly moderately, moderately, is evaluated by the difference from the ideal mesh division
moderately plus, strongly, strongly plus, very strongly, very scheme predefined as (E0). First, the parameters are normalized
very strongly, and extremely important to the objective, and the ideal mesh division scheme (E0) is defined. Second, the
respectively. As a result, the AHP matrix A is shown in GRC values of each candidate mesh division scheme are
TABLE I. At last, the weights of factors are calculated and the calculated according to the global weights and normalization
consistency check [16] of AHP matrix is carried out in order to value of parameters. Third, the GRC values of each candidate
avoid the subjective judgment errors. mesh division scheme are compared, and the one with the
largest GRC value is selected as the optimal mesh division
scheme. Considering both the GPS localization shift error Pr
and the information loss ratio Fr, which are smaller-the-better,
the normalization is performed as shown by (12). And the A. Case 1: Error Probability caused by GPS Localization
normalization value is shown in (13). Shift
Based on the theoretical calculation of the error probability
e*i ( j ) = ⎡⎣U j + L j − ei ( j ) ⎤⎦ (U j + L j ) (12) caused by the GPS localization shift, the impact of both the
mesh division size parameter and the GPS localization shift
E * = {ei* ( Pr ), ei* ( Fr ) | i = 1,..., N } (13) parameter to the error probability is investigated and numerous
Where j=1, 2,…,k, Uj=max{e1(j), e2(j),…,en(j)}, simulation results are shown in Fig. 6. Fig. 6 (a) depicts the
Lj=min{e1(j), e2(j),…,en(j)}. evident trend of the impact of the error probability Pr caused by
different GPS localization shift Z and mesh division size Y.
At last, the GRC values are calculated using (14)-(17), Considering the scenario with the same GPS localization shift
where the idle mesh division scheme is E0. The GRC values of Z, the larger the mesh size Y is, the lower the error probability
all mesh division schemes are shown by (18). Pr is. As shown in Fig. 6, different curves depict that the
smaller the GPS localization shift Z is, the smaller the error
Γ 0,i = ( Δ min + Δ max ) ( Δi' + Δ max ) (14) probability Pr is. The pentagram marker in each curve is the
minimum error probability Pr corresponding to different GPS
k localization shift Z.
Δ i' = ∑ w j e0 ( j ) − ei* ( j ) (15)
j =1

Δ max = max(| e0 ( j ) − ei* ( j ) |) (16)

(i , j )

Δ min = min(| e0 ( j ) − ei* ( j ) |) (17)

(i , j )

Γ = {Γ 0,1 ," , Γ 0, N } (18)

Where the parameter w j is the jth element’s weight and the
max/ min is the function of computing the maximum/minimum
(i, j ) (i , j )

value of a set of numbers varying with i and j, which are

independent.
In general, the candidate mesh division scheme with the
largest GRC value is chosen as the optimal mesh division
scheme.

VI. SIMULATION RESULTS

In this section, the proposed optimal mesh division scheme
is evaluated by numerous simulation results. Simulation Figure 6. Simulation results of optimal mesh division scheme (a) Error
scenarios are divided into three cases, including the error probability caused by GPS localization shift (b) Information precision ratio (c)
probability caused by GPS localization shift in case 1, the Information loss ratio (d) Optimal mesh division topology
information loss ratio caused by multi-RATs overlapped
meshes in case 2 and the optimal mesh division scheme in B. Case 2: Information Loss Ratio caused by Multi-RATs
case 3. Overlapped Meshes
Based on the simulation scenario as shown in Fig. 1, the In the multi-RATs scenario, parameters of information
simulation region of 3 000m × 3 000m is taken into account. precision ratio PF and information loss ratio Fr, which are
The parameter Z which represents the GPS localization shift is caused by the information filter parameter F, are evaluated
set to 10m, 30m, 50m and 80m respectively. And the mesh with simulation results as shown in Fig. 6. Considering the
division size parameter Y is ranged from 50m to 1 045m with information precision ratio as shown in Fig. 6 (b), the trend is
an interval of 5m to verify the impact to the error probability obvious that the smaller the information filter parameter F is,
the bigger the information precision ratio PF is. In terms of the
caused by different GPS localization shift parameter Z. And
impact of mesh division size Y to PF, the smaller the mesh size
the error probability parameter Pr is calculated in accordance Y is, the bigger the information precision ratio PF is. Fig. 6 (c)
with different mesh division size parameter Y and different denotes the trend that the smaller the information filter F is, the
GPS localization shift parameter Z. Besides, the information smaller the information loss ratio Fr is. With the information
filter F is set to 0.1, 0.4 and 0.8 respectively to verify its filter F fixed, the smaller the mesh size Y is, the smaller the
impact to the mesh division scheme. information loss ratio Fr is. As shown in each curve, the
pentagram marker is the minimum information loss ratio Fr
corresponding to different information filter F. Based on these
analysis, the candidate scheme with the smallest information
loss ratio Fr should be taken into account in the optimal mesh work will focus on these parameters, such as the transmission
division scheme. delay of mesh information and the overall procedure efficiency
of proposed scheme.
C. Case 3: Optimal Mesh Division Scheme based on AHP
and GRA Algorithms ACKNOWLEDGMENT
Based on the AHP and GRA algorithms, the optimal mesh The contributions of the colleagues from E3 consortium are
division scheme is achieved and verified by numerous hereby acknowledged. The author would also like to extend
simulation results. In order to decrease the error probability great gratitude to the colleagues from Wireless Technology
caused by the GPS localization shift Pr, the mesh division size Innovation Institute of Beijing University of Posts and
should be large. On the other hand, the mesh division size Telecommunications.
should be small in order to reduce the information loss ratio Fr
caused by the multi-RATs overlapped meshes scenario. As a
REFERENCES
result, by using the AHP and GRA algorithms together, we try
to find an equilibrium point of the mesh division size, taking [1] U. Varshney, “Recent advances in wireless networking,” IEEE
Computer, Vol.33, No. 6, June 2000.
into account both the Pr and the Fr parameters. Based on the
flow of optimal mesh division scheme as shown in Fig. 4, the [2] L. Kleinrock, “Nomadic computing: An opportunity,” Computer
Communications Review, January 1995.
calculation results of GRC values are shown in TABLE III.
[3] P. Houze, S. Ben Jemaa, P. Cordier, “Common Pilot Channel for
Y network selection,” IEEE VTC in Spring Conference, Melbourne, May,
Different mesh division schemes are represented as S F , 2006.
where the parameter F stands for the information filter and the [4] J. Mitola, “Software radios: wireless architecture for the 21st Century,”
parameter Y stands for the mesh division size. For example, New York: Wiley, 2000.
50
S0.1 represents the mesh division scheme with parameter F=0.1 [5] J. Mitola, “Cognitive radio,” Licentiate proposal, KTH, Stockholm,
Sweden.
and parameter Y=50. Then the optimal mesh division scheme is [6] Q. Zhao, B. M. Sadler, “A Survey of Dynamic Spectrum Access,”
chosen with the largest GRC value from TABLE III. Signal Processing Magazine, IEEE, vol.24, no.3, pp.79-89, May 2007.
[7] G. Ganesan, Y. Li,B. Bing, S. Li, “Spatiotemporal Sensing in Cognitive
TABLE III. GRC VALUES OF DIFFERENT MESH DIVISION SCHEMES Radio Networks,” IEEE JSAC, vol.26, no.1, pp.5-12, Jan. 2008.
[8] A. Ghasemi, E.S. Sousa, “Collaborative spectrum sensing for
Scheme 50
S0.1 55
S0.1 60
S0.1 ... 1045
S0.8 opportunistic access in fading environments,” DySPAN’05, 8-11 Nov.
Parameter
Pr Value 0.452 0.4156 0.3846 ... 0.0242
[9] P. Cordier et al., “E2R Cognitive Pilot Channel concept,” IST Summit,
Mykonos, June, 2006.
Normal- 0.0509 0.1272 0.1924 ... 0.9491
[10] R. Filjar, T. Kos, I. Markezic, “GPS Ionospheric Error Correction
ization
Fr Value 0.0466 0.049 0.0507 ... 0.9613
Models,” Multimedia Signal Processing and Communications, 48th
International Symposium ELMAR-2006, pp.215-217, June 2006.
Normal- 0.9566 0.9523 0.9507 ... 0.0647 [11] G. Bishop, A. Mazzella, E. Holland, S. Rao, “Algorithms that use the
ization ionosphere to control GPS errors,” IEEE Position Location and
GRC 0.6883 0.7058 0.722 ... 0.6898 Navigation Symposium, pp.145-152, Apr. 1996.
Finally, based on the simulation scenario as shown in Fig. 1, [12] X. Gao, Y. Dai, K. Wang, “A study on the self-difference GPS
the mesh division scheme with the largest GRC value of positioning by dynamic and fictitious datum station,” Proceedings of the
0.8633 is chosen, where the mesh division size Y is 200m with IEEE International Vehicle Electronics Conference, vol.1, pp.16-18,
Sep. 1999.
the information filter F=0.1 and the GPS localization shift
[13] IST-2003-507995 E2R (End-to-End Reconfigurability) Project,
Z=30m. And the layout of the simulation scenario with the http://e2r.motlabs.com/.
optimal mesh division scheme is shown in Fig. 6 (d).
[14] ICT-2007-216248 E3 (End-to-End Efficiency) Project, http://www.ict-
e3.eu/.
VII. CONCLUSION [15] J. Perez-Romero, O. Sallent, R. Agusti, L. Giupponi, “A Novel On-
Demand Cognitive Pilot Channel enabling Dynamic Spectrum
In the CR environment, the CPC concept is proposed as one Allocation,” 2nd IEEE International Symposium on New Frontiers in
of the candidate solutions to transmit the network information Dynamic Spectrum Access Networks, pp.46-54, April 2007.
to the UEs, enabling dynamic spectrum sharing. A novel [16] T. L. Saaty, “Fundamentals of Decision Making and Priority Theory
optimal mesh division scheme is proposed in order to make with the Analytic Hierarchy Process,” RWS Publications, U.S.A., 2000.
network information transmitted precisely and efficiently [17] J. L. Deng, “Introduction to grey system theory,” The Journal of Grey
through CPC channel. Taking into account both the error System, vol.1, no. 1, pp. 1-24, 1989.
probability caused by the GPS localization shift and the [18] L. F. Fan, H. B. Jiang, K. S. Chen, “The application of fuzzy analytic
information loss ratio caused by the multi-RATs overlapped hierarchy process in the location selection of distribution centers,”
Railway Freight Transport, vol. 20, no. 11, pp. 15-18, 2005.
meshes scenario, the novel optimal mesh division scheme is
[19] L. Liang, Z. Feng, P. Zhang, L. Wang, Q. Zhang, Y. Yu, Y. Bai, L.
designed based on the AHP and GRA algorithms. Through Chen, “A Mode and Channel Selection Scheme for Plug-and-Play Multi-
theoretical research and numeric simulation results, we arrive Mode Access Point,” 5th IEEE Consumer Communications and
at an equilibrium point of mesh division size with small error Networking Conference (CCNC), pp.538-542, Jan. 2008.
probability Pr and small information loss ratio Fr in this [20] Y. Zhang, Y. Jiang, Y. Ji, “Network Selection Process Based on In-band
scenario. However, other parameters that would impact the and Out-band Cognitive Pilot Channel in the end-to-end Reconfigurable
mesh division size should also keep in mind and our future Systems,” WWRF16, Apr. 2006.