Counting Credibility based Cooperative Spectrum Sensing Algorithm

Lianlian Song, Li Wang and Shibing Zhang

School of Electronics and Information, Nantong University, Seyuan Road, Nantong, China

Keywords: Cooperative Spectrum Sensing, Sensing Node, Channel Overhead, Lifecycle.

Abstract: In the cooperative spectrum sensing, if too many nodes take part in the cooperative data fusion, it would

weigh the channel overhead and energy loss lot but improve the spectrum sensing performance little. This

paper focuses on the channel overhead of cooperative spectrum sensing and the lifecycle of cognitive

networks, and proposes a novel cooperative spectrum sensing algorithm. In the algorithm, all of the nodes

are sorted by means of counting reliability. Only a part of nodes participate in the cooperative data fusion in

the fusion centre. It cut down the number of nodes participating in the data fusion and save the average

energy of the sensing nodes. The simulation results show that the proposed algorithm can effectively reduce

channel overhead and prolong the lifecycle of cognitive network in the premise of ensuring the spectrum

detection performance.

1 INTRODUCTION

With the growth of the wireless data traffic, the

spectrum resources become more and more scarce

(Akyildiz, 2008). Cognitive radio (CR) is an

intelligent spectrum sharing technology and taken as

a promising way to solve the problem (Wang et al.,

2011). The main idea of CR is to access spectrum

dynamically (Qu and Wang, 2009), (Yang et al.,

2009), (Li et al., 2011). In the CR network, cognitive

users (secondary users) opportunistically access the

empty spectrum bands which has been assigned to

the primary user (PU) but unused at present. The key

to reuse the empty spectrum and to improve the

spectrum efficiency is to ensure the CR senses

spectrum accurately. However, due to the channel

fading and multipath, a single cognitive node is

often difficult to guarantee the validity of the

spectrum sensing. Therefore, cooperative spectrum

sensing is put forward to improve the performance

of the spectrum sensing (Bai et al., 2013), (Mai et al.,

2011), (Liu et al., 2012), (Bao et al., 2012).

The cooperative spectrum detection based on soft

decision fusion makes full use of the information of

sensing nodes to make accurate spectrum decision,

but it increases the system overhead and the energy

loss of sensing nodes (Zhang and Yang, 2003). It

should be considered in cooperative spectrum

sensing that how to reduce the overhead of the data

transmission and the energy loss of the sensing

nodes as far as possible in the premise of ensuring

the spectrum sensing performance. Some algorithms

were proposed to overcome these problems (Chair

and Varshney, 1986), (Chen et al., 2008), (He et al.,

2008). But they solve the problems only from the

view of energy loss or lifecycle. A cooperative

spectrum sensing algorithm based on node

recognition (NRCS) was proposed to improve the

spectrum sensing performance in the case of

malicious nodes and reduce the system overhead

simultaneously (Zhang et al., 2014). But the

overhead of the data transmission and the energy

loss of the sensing nodes are not lowest because all

reliable nodes participate in the data fusion.

In this paper, we propose a counting credibility

based cooperative spectrum sensing algorithm

(CCCS) to reduce the channel overhead and prolong

the lifecycles of cognitive networks. In the

algorithm, all of the nodes are sorted according to

their counting reliability. Only a part of nodes with

largest or next larger reliability weighted factors take

part in the cooperative data fusion in the fusion

centre.

The rest of this paper is organized as follows.

Section II presents the system model. Section III

describes the cooperative spectrum sensing

algorithm. Some simulation results are discussed in

section IV. Conclusions are stated in section V.

Song L., Wang L. and Zhang S..

Counting Credibility based Cooperative Spectrum Sensing Algorithm.

DOI: 10.5220/0005574700710075

In Proceedings of the 12th International Conference on Wireless Information Networks and Systems (WINSYS-2015), pages 71-75

ISBN: 978-989-758-119-9

 2015 SCITEPRESS (Science and Technology Publications, Lda.)

2 SYSTEM MODEL

Assume that there are one primary user and N

cognitive users in the cognitive network, as shown in

Figure 1. Two hypotheses, H

and H

, represent the

spectrum detected in the network is busy (the

primary user uses the spectrum at present) and is

free (the primary user does not use the spectrum at

present), respectively. The spectrum sensing of the

cognitive user (sensing node), i=1······N, can be

modelled as a binary hypothesis testing problem as

follows

   

 

ii i

xt ht st nt

Hxtnt





(1)

where x

(t) is the signal received in the i

sensing

node, s(t) is the signal transmitted by the primary

user, h

(t) is the channel gain of the i

sensing node,

(t) is the additive white Gaussian noise (AWGN)

in the signal received of the i

sensing node.

The cooperative spectrum sensing can be divided

into two steps, local detection and data fusion. In the

local detection, the i

sensing node makes

hypothesis testing after receiving the signal x

(t), and

obtains local detection result “1” or “0”. “1”

represents the hypothesis H

is supported, “0”

represents the hypothesis H

is supported. In the data

fusion, the fusion centre fuses the local detection

results from the sensing nodes, and makes final

decision according to the decision rule and decision

threshold.

Figure 1: System model.

3 COOPERATIVE SPECTRUM

SENSING ALGORITHM

It has been showed that the spectrum sensing

performance in the cooperative spectrum sensing is

dependant on the number sensing nodes which

participate in the data fusion, and their reliabilities

(Chair and Varshney, 1986). Note that the credibility

of the sensing node is a accumulative result of the

historical sensing information. That is to say, the

present credibility of the sensing node is related to

the sensing node’s historical sensing results.

Definition 1. The credibility of the i

sensing node

in the m

spectrum sensing is defined as

,1 ,1 ,1 ,1

im i im FCm

rrdd













  







  







(2)

where r

i,m-1

is the credibility of i

sensing node in

(m-1)

spectrum sensing , ρ is a attenuation factor

which represents the strength of association with

historical information, 0<ρ<1; d

i,m-1

is the local

detection result of i

sensing node in (m-1)

spectrum sensing, d

FC,m-1

is the global decision result

in (m-1)

spectrum sensing.

According to the local detection result and the

global decision result in last time, the fusion center

updates the credibility of the sensing node i

cumulatively. When the local detection result of i

sensing node, d

i,m-1

, is the same as the global

decision result of the fusion centre, d

FC,m-1

, in (m-1)

spectrum sensing, “1” is added to the historical

weighted credibility. And then, the credibility of the

sensing node in the m

spectrum sensing is

updated. When the local detection result of i

sensing node, d

i,m-1

, is different with the global

decision result , d

FC,m-1

, in (m-1)

spectrum sensing,

“1” is subtract from the historical weighted

credibility. And then, the credibility of the i

sensing

node in the m

spectrum sensing is replaced. If the

credibility replaced is smaller than 0, it will be

replaced by 0. Moreover, the later credibility of the

sensing node has larger weighted factor by means of

the attenuation factor ρ. Therefore, the impact of

accidental errors on spectrum detection caused by

local detection can be eliminated as much as

possible.

Definition 2. The reliability weighted factor of the

sensing node in the m

spectrum sensing is

defined as

1, 2,

wiN











(3)

When the fusion centre obtains the reliability

weighted factors of all of the sensing nodes, it sorts

them according to their reliability weighted factors

and chooses the sensing node with largest reliability

weighted factor, for example sensing node l,

WINSYS2015-InternationalConferenceonWirelessInformationNetworksandSystems

l∈{1······N }, to participate the data fusion. Then,

i,m-1

is sent to the fusion centre and the global

detection statistics is formed as follows

Clmlm

Twd

(4)

Next, the fusion center makes the global decision

according to the decision threshold as follows

:1,

:0,

FC FC

Hd T













(5)

where λ is the decision threshold (Zhang et al,

2014).

If the hypothesis H

is supported, the fusion

centre terminates the data fusion and achieves the

spectrum detection result H

in this time. Otherwise,

the fusion centre will select another with next larger

reliability weighted factor, for example sensing node

k, k

∈

{1······N }, to participate the fusion to form the

new global detection statistics based on the last

statistics as follows

CFCkmkm

TTwd 

(6)

The fusion center will make the global decision

again according to (5) until

is supported or all of

the sensing nodes have been selected to participate

in the data fusion.

The cooperative spectrum detection algorithm

based on the counting credibility above can be

summarized as in Algorithm 1.

Algorithm 1: Counting Credibility Based Cooperative

Spectrum Sensing Algorithm.

1: Calculate the credibility of all sensing nodes

according to (2);

2: Calculate the reliability weighted factor of the all

sensing nodes according to (3);

3: Sort all of the sensing nodes according to their

reliability weighted factors;

4: Choose the sensing node with largest reliability

weighted factor and form the global detection

statistics according to (4);

5: Makes the global decision according to (5);

6: If the hypothesis

is supported, the fusion

centre ends the data fusion and achieves the

spectrum detection result in this time. Otherwise,

the fusion centre will select another with next

larger reliability weighted factor to participate

the fusion. Then it forms the new global

detection statistics according to (6).

7: Go back to Step 5 until

is supported or all of

the sensing nodes have been selected to

participate in the data fusion.

8: End.

4 SIMULATION AND ANALYSIS

We simulate the CCCS algorithm proposed in this

paper in AWGN channel and compared it with

cooperative spectrum detection algorithm based on

node recognition (NRCS) (Zhang et al, 2014). In the

simulation, the primary signal is modelled as a phase

shift keying (PSK) signal with the 5000 Bauds and

10 MHz carrier frequency. The sampling frequency

is 100 MHz and the number of sampling is 512.

There are 8 sensing nodes in the CR network. The

attenuation factor

ρ of the node’s credibility is 0.5.

Figure 2 shows the percentages of the nodes

selected to participate in the cooperative data fusion.

We compare the percentages between the CCCS and

NRCS algorithms in two cases, there is one

malicious node (Num = 1) and two malicious nodes

(Num = 1), respectively. With the increase of SNR,

the number of the nodes selected to participate in the

cooperative data fusion in the CCCS algorithm

decreases, but the one in the NRCS algorithm is

relatively stable. In the case of one malicious node,

when SNR is equal to -13 dB, the percentage of the

CCCS algorithm is 0.5, while the one of the NRCS

algorithm is 0.87. In the case of two malicious

nodes, when SNR is equal to -13 dB, the percentage

of the CCCS algorithm is less than 0.4, while the

one of the NRCS algorithm is close to 0.75.

Compared with the NRCS algorithm, the CCCS

algorithm cuts down the number of the nodes to

participate in the cooperative data fusion and

reduces channel overhead effectively.

-16 -15 -14 -1

0.2

0.4

0.6

0.8

SNR

(

)

Cooperative Percentage (



)

CCCS-



(Num=1)

NRCS-



(Num=1)

CCCS-



(Num=2)

NRCS-



(Num=2)

Figure 2: Percentages of cooperative nodes in the CCCS

and NRCS algorithms.

Figure 3 and Figure 4 show the comparisons of the

detection probabilities and false alarm probabilities

of the CCCS and NRCS algorithms in the two cases

respectively. It is obvious that the spectrum sensing

CountingCredibilitybasedCooperativeSpectrumSensingAlgorithm

performance of the CCCS algorithm, no matter the

detection probability or the false alarm probability,

is almost the same as one of the NRCS algorithm.

That is to say the node selection algorithm based on

the counting reliability proposed does not decrease

the spectrum sensing performance of the cognitive

network.

-16 -15 -14 -1

0.5

0.6

0.7

0.8

0.9

SNR

(

)

Detection Probability ( p

)

CCCS-P

(Num=1)

NRCS-P

(Num=1)

CCCS-P

(Num=2)

NRCS-P

(Num=2)

Figure 3: Detection probabilities of the CCCS and NRCS

algorithms.

-16 -15 -14 -13

0.2

0.4

0.6

0.8

SNR(dB)

False Alarm Probability ( p

)

CCCS-P

(Num=1)

NRCS-P

(Num=1)

CCCS-P

(Num=2)

NRCS-P

(Num=2)

Figure 4: False alarm probabilities of the CCCS and

NRCS algorithms.

Figure 5 shows the lifecycles of cognitive

networks which adopt the CCCS and NRCS

algorithms in the two cases. When the NRCS

algorithm is used, all of the reliable nodes participate

in the cooperative data fusion, every node consumes

it’s energy in each data fusion. Consequently, the

lifecycle is shorted. When the CCCS algorithm is

used, only nodes selected participate in the

cooperative data fusion, the average frequencies of

the sensing nodes participating in the data fusion is

reduced as far as possible, the energy loss of each

sensing node is minimized. Therefore, the lifecycle

is prolonged.

-16 -15 -14 -13

400

500

600

700

800

SNR(dB)

Lifecycle ( Round )

CCCS-Round(Num=1)

NRCS-Round(Num=1)

CCCS-Round(Num=2)

NRCS-Round(Num=2)

Figure 5: Lifecycles of cognitive networks with the CCCS

and NRCS algorithms.

From Figure 2 to Figure 5, we see that the CCCS

algorithm would cut down the number of node

participating in the cooperative data fusion, save the

average energy of the sensing nodes, reduce the

channel overhead of the system, and prolong the

lifecycle of the cognitive network. But it does not

debase the spectrum sensing performance of the

cognitive network.

5 CONCLUSIONS

In order to reduce the channel overhead and prolong

the lifecycle of the cognitive network, a novel

cooperative spectrum sensing algorithm based on

counting credibility is proposed. In the algorithm, all

of the nodes are sorted according to their counting

reliability. Only a part of nodes with best or Sub-

best reliability take part in the cooperative data

fusion in the fusion centre. It decreases the number

of node participating in the data fusion and save the

average energy of the sensing nodes. The simulation

results show that the proposed algorithm can

effectively reduce channel overhead and prolong the

lifecycle of cognitive networks.

ACKNOWLEDGEMENTS

This study is supported by the National Science

Foundation of China under 6137111 and 6137112,

and the applied basic research project of the

WINSYS2015-InternationalConferenceonWirelessInformationNetworksandSystems

Ministry of Transport of China under grant

2014319813220.

REFERENCES

Akyildiz I. F., Lee W. Y., Vuran M. C. and Mohanty S.,

2008, A survey on spectrum management in cognitive

radio networks, IEEE Commun. Mag., vol. 46, no. 4,

pp. 40-48.

Wang B., Liu K. J. Ray, 2011. Advances in cognitive

radio networks: a survey, IEEE Journal on Selected

Topics in Signal Processing, vol. 5, no. 1, pp. 5-23.

Qu D. M., Wang Z. Q., 2009. Signal spectrum forming

method of OFDM opportunistic spectrum access,

Journal of Electronics and Information Technology,

vol. 31, no. 8, pp. 1965-1968.

Yang X. Y., Yang Z. and Liu S. B., 2009. Prediction

mechanism-based opportunistic spectrum access in

cognitive radio networks, Journal of Chongqing

University of Posts and Telecommunications, vol. 21,

no. 1, pp. 14-19.

Li Z., Zhao L. J. and Liu Q., 2011. SDM-based

opportunistic spectrum access in cognitive radio

networks, Journal of Electronics and Information

Technology, vol. 33, no. 5, pp. 1172-1177.

Bai Z. Q., Wang L. and Liang X. Y., 2013. Robust

cooperative spectrum sensing based on STBC scheme,

Transactions of Beijing Institute of Technology, vol.

33, no. 9, pp. 956-960.

Mai L. X., Qin X. W. and Dai X. C., 2011. Hidden

Markov model based spectrum sensing strategy with

cooperation, Journal of University of Science and

Technology of China, vol. 41, no. 4, pp. 283-292.

Liu J., Chen W., Cao Z. G. and Zhang Y. J., 2012.

Cooperative beamforming for cognitive radio

networks: a cross-layer design, IEEE Trans.

Commun., vol. 60, no. 5, pp. 1420-1431.

Bao Z. H., Zhang S. B. and Zhang X. G., 2012. Research

on cognitive user pairing and cooperative sensing,

Journal of Communication, vol.33, no. 1, pp. 128-135.

Zhang X., Yang D. C., 2003. A study on the reduction of

overhead of resource management scheduling

algorithm, Journal of Beijing University of Posts and

Telecommunications, vol. 26, no. 2, pp. 48-52.

Chair Z., Varshney P. K., 1986. Optimal data fusion in

multiple sensor detection systems, IEEE Transactions

on Aerospace and Electronic Systems, vol. 22, no. 1,

pp. 98-101.

Chen J. J., Fan X. P. and Qu Z. H., 2008. Subtractive

clustering based clustering routing algorithm for

wireless sensor networks, Information and Control,

vol. 37, no. 4, pp. 435-444.

He Z. Y., Long C. F. and Yin Q., 2008. A distributed

clustering algorithm for sensor networks based on

node density, Computer Applications and Software,

vol. 25, no. 12, pp. 19-20.

Zhang S. B., Song L. L. and Liu Y., 2014. Cooperative

spectrum detection algorithm based on node

recognition, Journal of Data Acquisition and

Processing, vol. 29, no. 9, pp. 688-693.

CountingCredibilitybasedCooperativeSpectrumSensingAlgorithm