Cooperative Spectrum Sensing in Cognitive Radio Networks

Cooperative Spectrum Sensing in Cognitive Radio Networks

Gourav Bansal

School of Electronics Engineering VIT University, Vellore-632014 India

Akshay Agnihotri School of Electronics Engineering VIT University, Vellore-632014

India

Abstract The entire operation of cognitive radio depends on the spectrum sensing technology. The main function of cognitive radio is to detect unused spectrum and sharing it to other user without causing harmful interference to the primary user induced by reporting phase. Basically it requires two phases: detection phase and reporting phase. In detection phase cognitive users detects the presence of primary users (Licensed user). In reporting phase cognitive user forward their detection report to fusion center. In this, we analyze the effect of ROC (Receiver Operating Characteristics) with and without dedicated reporting channel by jointly considering the signal detection and reporting phases

Keywords Cognitive radio, cooperative spectrum sensing, receiver operating characteristics, fusion center

1. INTRODUCTION

   Cognitive radio detects the unused spectrum and shares it with the other users. The use of cognitive radio improves the efficiency of wireless spectrum resources [1], [2]. Energy detection [3] , matched filter [4] detection and feature detection [5]: these three are the main categories of signal processing. In order to reduce the fading effect in wireless system, a cooperative spectrum sensing technique is used. In this technique the detection results from various cognitive users are obtained and then combined it at the fusion center together by using various logic rule such as AND fusion rule and OR fusion rule. The cooperative spectrum sensing process needs two phase: detection phase and reporting phase. For the spectrum sensing process one cannot be designed and optimized these two phases in isolation as they are not independent to each other. In detection phase cognitive users detects the presence of primary users (Licensed user) and cognitive user forward their detection report to the fusion center in reporting phase. At the fusion center the results are combined by the logic rule [6]. But there is a need to take care of time duration of both the phases as both the phases could affect each other. If the time duration of any phase is more then it will degrade the performance of overall spectrum sensing at the fusion cente

2. PROPOSED COOPERATIVE SPECTRUM SENSING IN COGNITIVE RADIO NETWORKS

   1. System Description

      Time duration for detection phase and reporting phases are and 1- fractions respectively of one time slots. It is to be assumed that is same for all cognitive users. CUs forwards their detection report to fusion center (FC) over the orthogonal sub-channel. Sub-channels are equally divided in reporting phase, resulting in multiple time slots. These all CUs will interfere primary user (PU) potentially in the reporting phase so in order to reduce this interference as much as possible we use a concept of selective relay based cooperative sensing scheme where all cognitive users sends their detection report to the fusion center in a selective fashion depending on the presence or absence of primary user. If CU detects that PU is absent in that case it will transmit an indicator signal with encoded cyclic redundancy code (CRC) to the FC else no signal is transmitted. At the fusion center the signal is decoded and if it is successfully decoded then it means CU detection report says that PU is absent else primary user is present. So the possibility of causing interference is reduced and controlled also as CU will interfere the PU only when it fails to detect the presence of PU

      Figure 1. System model of cooperative spectrum sensing

   2. Signal Model

   In this model we use a Rayleigh fading and it is constant during one whole time slot. N0 is the power spectral density of the additive white Gaussian noise (AWGN) that is same for the entire receiver. Assume that Pp and Ps are the transmit power of PU and CU. Let Hp=H1 denotes the presence of

   primary user and Hp=H0 represents absence of it. In the

   detection phase, the signal received at CUi for the k time slots, can be written as,

   checking fails means no signal is transmitted and PU is present otherwise PU is absent, and finally result is stored in

   (2). And the (2) can be written as,

   H i

   i

   yi (1) = Pp h pi (1) n (1), i = 1,2,3…..,M (i)

   H i

   H1 ,

   ic (2)1

   Where index(1) shows the 1st phase of k time slots, and M is number of CUs. hpi is fading coefficient from PU to CUi.

   H i (2)

   H0 ,

   ic (2)0

   (viii)

   0 ,

   p

   (1)

   x (1),

   H p H0 H p H1

   (ii)

   Where ic(2) is an outage event that occur when a channel capacity is below a required data rate. Therefore ic(2) is defined as,

   Where xp(1) is a complex symmetric Gaussian distribution

   (2) 1: (1-) log (1

   hic

   2 s 2

   ) 1

   (ix)

   and this is the transmit signal of PU in the 1st phase of time

   ic M

   2 2 2 BT

   i

   slots k. And yi(n)(1)2 is the n-th sample energy of signal

   hpc

   p 1

   received at i-th CU. Therefore the output statistic of energy detector of CUi is given by the formula,

   s Ps / No, p Pp / No

   1 N

   1. 2

      Where B is the frequency bandwidth.Over each time slot

      T yi (1) N

      y

      n1 i

      1. (iii)

   spectrum sensing is performed which gives data rate of initial detection results as 1/(BT). However it is completed during

   Where N=Tfs is number of samples where fs is sampling frequency and T is time slot length.

   Using an energy detection approach, the detection results

   the whole reporting phase time i.e (1-), so the reporting phase capacity is scaled by (1-). An outage event occur under two condition one is =0 when =H and other one is small

   i H i (1) 1

   H i (1) is given by,

   H0, T[ yi (1)] i

   value of hic2 which gives channel capacity below a required data rate 1/(BT). Now by using fusion rule FC combine all

   H i

   (2).So the final result obtained By using AND fusion rule,

   H i (1) H T[ y (1)]

   (iv)

   M

   1, i i

   H c H i (2) (x)

   i1

   Each CU sends a i to fusion center over orthogonal sub-

   By using OR fusion rule,

   channel. xi is an indicator signal with encoded CRC. i and M

   (2) is defined as,

   H c H i (2) (xi)

   i1

   i

   x ,

   i

   H i (1)H0

   (v)

3. PERFORMANCE ANALYSIS OF SPECTRUM SENSING SCHEME

   In this section we have to analyze ROC by the traditional

   0,

   H i (1)H1

   method as well as proposed scheme over Rayleigh fading channels.

   0 ,

   H pH0

   a). ROC Analysis:

   (2) x

   (2),

   H H

   (vi)

   We check the performance of Receiver operating

   p p 1

   Where xp(2) is a complex symmetric Gaussian distribution and this is the transmit signal of PU in the 2nd phase of time Slot k.

   Signal received at fusion center is given as,

   characteristic (ROC) with and without dedicated channel

   b). Traditional cooperative sensing with a dedicated channel: In this the results of initial detection of CU that is encoded with CRC are forwarded to FC trough a dedicated channel. At FC signal will be decoded and successfully decoded outcomes only will be combined. These successfully decoded

   outcomes after combined constitute a set C. Sample space of all possible set such that {C 0 U Cm} where m =

   yi (2) P h P h (2) n (2) (vii)

   1,2,3..,2M-1. Cm is sub collection of non empty subset of M

   c s ic i

   p pc c

   CUs.

   Where index 2 denotes the second phase.Now fusion center will decode the i and do the CRC operation and if CRC

   Case C=0: decoded operation fails at FC therefore

   log2(1 hic

   2 T ) 1

   s

   B T

   , i 1, 2,……., M

   (xii)

   M

   2 1

   OR m i,1

   Pd traditional Pr{C } Pr{C C }[1 (1 Pd

   )] (xxi)

   d d m1

   iCm

   Where BdTd is bandwidth time product of dedicated channel. Therefore at C=0 no fusion is done at FC and degrade the performance of spectrum sensing.

   Pf is probability of overall false alarm is ,

   2M 1

   OR m i,1

   Pf traditional Pr{C } Pr{C C }[1 (1 Pf )] (xxii)

   m1 iCm

   H c C H1 (xiii)

   Therefore the term Pr(=) are calculated as,

   Case C=Cm decoded operation successfully happens so fusion is happened at FC either by AND fusion rule or OR fusion rule,

   M

   Pr{C }

   i1

   [1 exp( )] (xxiii)

   2

   ic

   log (1 | h |2 T 1/ B T ,

   i C

   (xiv)

   1/(BdTd )

   2 ic s

   )

   2

   d d m

   1

   Where, [2

   1] / T

   log2 (1 h

   T

   s

   jc s

   , j Cm

   B T

   (xv)

   Similarly Pr(=) is given by,

   d d

   where Cm R C

   Pr{C Cm}

   iC

   exp( )

   2

   jCm

   [1 exp(

   )] (xxiv)

   2

   m

   AND Fusion Rule H c (C Cm ) H i (1) (xvi)

   iCm

   m ic jc

   c) By Proposed scheme:

   The probability of overall detection at the FC of presence of PU is given by,

   Pd proposed

   OR Fusion Rule

   H c (C C

   ) H i (1) (xvii)

   AND

   Pr{H c H1 H p H1} (xxv)

   Pd traditional

   m iCm

   M

   can be referred as probability of overall detection

   AND

   Pd

   of PU presence at FC for the AND based rule,

   Pr{ Hi (2) H1 H p H1}

   M

   i 1

   Pd

   traditional AND

   Pr{H c

   H1 H p

   H1}

   proposed AND

   Pdc,i i1

   (xxvi)

   Pr{H c H1 H p H1,C }Pr{C H p H1}

   Pdc,i Pr{H i (2) H1 H p H1} (xxvii)

   2M 1

   • Pr{Hc H1 H p H1,C Cm} pr{C Cm H p H1}

     m1

     2M 1

     Pfc,i

     Pr{H i (2) H1 H p

     H0} (xxviii)

     AND m i,1

     Pd traditional Pr{C } Pr{C C

     m1

     } Pd

     iCm

     (xviii)

     Where Pdc,i and Pfc,i is the probability of individual cognitive detection and individual false alarm.

     Pf traditional Pr{H c H H

     H ,C }Pr{C H

     H }

     Now the probability of overall false alarm for the presence of

     AND

     2M 1

     1 p 0

     p 0

     PU is given as,

     Pf

   • Pr{Hc H1 H p H0 ,C Cm} pr{C Cm H p H0}

   m1

   2M 1

   proposed AND

   Pr{H c H1 H p H1}

   AND m f ,1

   Pf tradional Pr{C } Pr{C C

   } Pd

   (xix) M

   m1

   iCm

   Pr{ H i,2(2) H1 H p H0}

   The probability of individual false alarm is given by,

   i1

   M

   proposed

   Pfi,1

   Pd,1 ,

   Pdi,1Q( N )

   (xx)

   Pf AND

   Pfc,i

   i1

   (xxix)

   2

   Pdi,1 Q(Q1(Pdi,1) 1 ) exp(i ) ,otherwise

   And now if we consider an OR logic rule, then

   M

   k

   pi i

   Q1 (Pd ) 1

   proposed OR

   1

   M

   i1

   (1 Pdc,i ) (xxx)

   Pd

   i p i,1

   k

   Q1 (Pd )

   N ,

   i,1 ,

   2 k

   2 4 k 2

   pi i pi i

   PfOR

   1 (1 Pfc,i ) (xxxi)

   p i

   proposed

   Pd is probability of overall detection by OR based rule,

   i1

   From the signal model Pdc,i can be written as,

   Pdc,i 1 Pr{H i (2) H0 H p H1} (xxxii)

   By using indicator signal and outage event eq. it can be further written as,

   AND and OR fusion rule, In this figure also we are getting the

   (1)

   h 2 1

   better result for AND based method.

   Pd 1 (1 Pd

   ) Pr{ log (1 ic s ) } (xxxiii)

   2

   c,i i,1

   M 2 h 1 BT

   pc p

   On solving, it can be rewritten as,

   Pd 1

   2 (1 Pd

   ) exp(

   ) (xxxiv)

   c,i

   ic i,1

   2 2 2

   where

   pc p ic ic

   s

   [2M /[(1 ) BT ] 1] / .

   Now as Pdc,i The Pfc,i can also be written as follows,

   Pfc,i 1 Pr{H i (2) H0 H p H0}

   Pfc,i 1 Pr{i (2) 0 H p H0}

   1 (1 Pf

   ) Pr{(1 ) log (1 h 2 ) 1 }

   i,1

   M 2 ic s BT

   Pfc,i 1 (1 Pfi,1

   ) exp( ) (xxxv)

   2

   ic

   Figure 3. ROC by proposed method

   Where Pfi,1 is probability of individual false alarm at CUi of presence of PU.

4. RESULT AND ANALYSIS

   In this section, we are showing the results of receiver

   And finally the figure 4 illustrates the combined result of ROC obtained by both traditional method as well as proposed method. In this paper the parameter that we used to get the

   results are:

   =5dB, T =10 dB, 5 dB, =0.2, 2 2 =0.2,

   p s s pc id

   operating characteristic (ROC) by using the traditional method and proposed scheme. The figure 2 explains the overall detection probability versus the overall false alarm probability for the traditional method, where the ROC obtained by AND has better result than OR based rule.

   Figure 2 ROC by traditional method

   In figure 3 we show the overall probability detection versus overall false alarm probability by the proposed scheme by the

   2 2 2 =1, M=2, Rp = 1bits/Hz, T=25ms, B=50 kHz, fs= 100kHz, BdTd=1000, Poutthr = 0.01.

   pi pd ic

   Figure 4 ROC by traditional method as well as proposed method

5. CONCLUSION

In this we used a concept of proposed selective relay based cooperative sensing scheme in cognitive radio network without dedicated reporting channel. We obtained the ROC from proposed method as well as from traditional method and on comparing we found that with this proposed scheme we can save the dedicated channel resources without reducing the ROC performance. But in Fig 4 we can see that in low probability detection region the overall false alarm probability obtained from proposed method are more than that of traditional method by any fusion rule that is the disadvantage of the proposed method and in higher detection region we can see that ROC obtained from both the scheme by any fusion rule is almost identical specially in AND fusion rule. So this proposed method is good when the overall probability detection region is more than 0.9.

REFERENCES

1. S.Haykin, Cognitive radio: brain-empowered wireless communications,"

   IEEE J. Sel. Areas Communication., vol. 23, no. 2, pp. 201-220, 2005

2. J. Mitola and G.Q. Maguire,Cognitive radio: making software radios more personal, IEEE Personal communication, vol. 6, pp. 13-18, 1999

3. V. I. Kostylev, Energy detection of a signal with random amplitude, ".in Proc. IEEE Int. Conf. Communication, 2002.

4. A. Sahai, N. Hoven, and R. Tandra, Some fundamental limits in cognitive rdio," inProc. Allerton Conf. Communication., Control Comput., 2004.

5. P.D. Sutton, K.E. Nolan, and L.E. Doyle,cyclostationary signatures in practical cognitive radio applications, IEEE J. Sel. Areas. Communi- cation, vol. 26, no. 1 ,pp. 13-24, 2008

6. A. Ghasemi and E. S. Sousa, Collaborative spectrum sensing for opportunistic access in fading environment," inProc. IEEE DySPAN 2005, pp. 131-136

1. Y.-C. Liang, Y. Zeng, E. Peh, and A. T. Hoang, Sensing-throughput tradeoff for cognitive radio networks," IEEE Trans. Wireless Commun., vol. 7, no. 4, pp. 1326-1337, Apr. 2008.

2. J. Ma, G. Zhao, and Y. Li, Soft combination and detection for cooperative spectrum sensing in cognitive radio networks,"IEEE Trans. Wireless Commun., vol. 7, no. 11, pp. 4502-4507, Nov. 2008.