New optimization scheme for cooperative spectrum sensing taking different snr in cognitive radio networks

IJRET: International Journal of Research in Engineering and Technology eISSN: 2319-1163 | pISSN: 2321-7308
__________________________________________________________________________________________
Volume: 03 Issue: 04 | Apr-2014, Available @ http://www.ijret.org 597
NEW OPTIMIZATION SCHEME FOR COOPERATIVE SPECTRUM
SENSING TAKING DIFFERENT SNR IN COGNITIVE RADIO
NETWORKS
Harsh N. Thakkar1
, Kiran R. Parmar2
1
M.E, Electronics & communication dept, L D College of engineering, India
2
Professor, Electronics & Communication dept, Govt. Engineering College, Gandhinagar, India
Abstract
This paper proposes new method to optimize the overall performance in hard cooperative spectrum sensing in cognitive radio.
Optimization strategy is proposed in order to optimize the overall performance by variation of SNR. Here given strategy contributes to
the methods in the literature by taking their performances to the peak point. Additionally, the effects of spectrum sensing technique
type that used locally at each CR, the local SNR, and the total number of cooperated CRs on the optimal fusion rule are found. The
energy detector (ED) spectrum sensing technique is examined as local spectrum sensing techniques. Here different error levels are
founded by variation of SNR. The optimal number of CRs form minimizing the error at SNR 5,10,13,17,18,20 are found to be 4 or 5,
5, 5 or 6, 6, 8, 9 respectively.
Keywords: cognitive radio; spectrum sensing; cooperative spectrum sensing; cooperative spectrum sensing optimization
----------------------------------------------------------------------***------------------------------------------------------------------------
1. INTRODUCTION
Providing frequencies for the new wireless technologies
increases the demand for spectrum, which is a scarce resource.
An ineffective use of the already licensed spectrum, meets that
high demand for the same [6].The technique of cognitive radio
(CR), has been proposed to deal with such problems [11]. In
Cognitive Radio system, the CR, which is called secondary
user, senses its surrounding radio frequency (RF) environment
to detect the vacant frequencies, which are being unused by
their licensed users. These users are called primary users.
Cognitive radio can use these vacant frequencies
opportunistically to transmit and receive its data by adapting its
transmission parameters like frequency. It enables secondary
user/network to utilize the spectrum. So, it is the strategy
proposed as a promising technology to improve spectrum
utilization efficiency.
2. EASE OF USE
As defining the vacant frequencies is the way to exploit these
unoccupied bands; the spectrum sensing is a key functional
factor in cognitive radio. Energy detector (ED) is a one of the
best spectrum sensing technique that does not require prior
information about the Primary signal. This technique is
simple, but that at the expense of its performance at low SNR.
Cooperative spectrum sensing technique is proposed to
eliminate the effects of shadowing and multipath fading on the
spectrum sensing of primary user, when only one CR module
is used [10]. In hard cooperation, each CR senses and decides
about the PR‟s signal in a specific frequency band, then a
binary information 1 or 0 is sent to the CR base station(CR-
BS) via dedicated control channel (CC), representing the
presence or absence of Primary signal. Then, the CR-Base
station decides on the all received digits using logical fusion
rule. Different strategies and factors have been investigated to
optimize the hard cooperative sensing performance by
minimizing the total error probability, or maximizing the
probability of detection [4]. It was achieved by optimizing the
number of cooperated Cognitive radios and the threshold. The
author has taken the global probability of detection in “OR‟
and “AND‟ fusion rules to peak by fixing the global false
alarm probability In [7]. In [3] Strategies to decrease the total
error probability under Neyman Pearson, and Bayesian
criterions have been studied. In this paper, we add our
contribution to the hard cooperative spectrum sensing
optimization area, by adding an important factor that can be
controlled in term to minimize the total error probability. Our
work here can be applied to all mentioned optimization
strategies to take them to the optimist point All optimization
published works, focused only on ED as a local spectrum
sensing. In this paper, the effects of using different numbers of
CRs, different SNR on the optimal fusion rule have been
investigated. The paper is organized as follows: Section III
defines the models for the local spectrum sensing techniques
when ED locally. Section IV presents the theoretical work of
the cooperative spectrum sensing, that includes the
optimization for the ED, and total number of CRs. Section V
concludes the paper.

__________________________________________________________________________________________
3. LOCAL SPECTRUM SENSING
We have a number of G (or r = 1, 2. . . G) CRs in the CR
network, where each CR performs spectrum sensing locally
using Energy Detection. Each CR transceiver is supported by
(N-IFFT/FFT) processers to perform both tasks of
communication and sensing the environment. The primary
transmitter with N subcarriers (N-IFFT/FFT) transmits
OFDM-QPSK signal with energy over each sub carrier, and
Ts which is symbol duration. So, each CR estimates the power
within each subcarrier in the frequency domain, with = 0,
1/N, 2/N. . . N-1/N are the bins of normalized frequency.
When we have fading environment where there are P
resolvable paths between the PR’s transmitter and CR’s
receiver, , p= 0, 1 … P-1 represents the discrete – time
channel impulse response between PR’s transmitter and CR’s
receiver. The hypothesis test which is binary for CR spectrum
sensing at the lth time is given by:
: (1)
(2)
Where l=0, 1 . . . L-1 is OFDM block’s index,
and denote the CR received, noise and PR
transmitted samples. Additive white Gaussian noise with zero
mean distorts the transmitted PR signal. The discrete
frequency response of the channel is obtained by taking the N
point FFT, with N > P as given below:
H( fi ) = (3)
Here H0represents the absence of PR’s signal and
H1represents its presence. Now to evaluate performance of the
local spectrum sensing using the rth
CR user, the probability of
detection Pd,r (fi), the probability of false alarm Pf,r (fi), and
the probability of missed detection Pm,r (fi) at each frequency
bin fi are considered based on the Neyman-Pearson (NP)
criterion. The probability that the rth
CR detector decides
correctly the presence of the PR’s signal is Pd,r (fi).The
probability that the rth
CR detector decides the PR’s signal is
present when it is absent is Pf,r (fi) . Lastly, is the probability
that the rth
CR fails to detect the PR’s signal when it is present
is Pm,r (fi) .
As following the same work in [4], we assume that all CRs are
much closed to each others in distances. Hence , wireless
environments here can be assumed as an identical and
independent in the CR‟s network, and SNR = for
each CR.
So, the Pd,r (fi), Pf,r (fi), and Pm,r (fi) will be replaced by
Pd(fi), Pf(fi), and Pm(fi) respectively in the remaining part of
this paper.
In General, the probabilities of detection Pd(fi), and false
alarm Pd(fi) can be defined for normally distributed statistic as
follow:
Pd(fi) = P( DEC (fi) >y/H1)
= Q( ) (4)
Pf(fi) = P( DEC (fi) > y/H0)
= Q( ) (5)
Finally, the probability of missed detection Pm(fi) can be
defined as:
Pm(fi) = P( DEC (fi) < y/H1)
= 1 - Q( ) (6)
Where DCE (fi) , is the decision statistic at. The symbol Q(x)
is the complementary cumulative distribution function, Q(x)
= ; it calculates the tail probability, and represents
the threshold that we choose. Here we note that, y can be
controlled based on L (threshold).Threshold’s values are
chosen based on .In this paper we examine the technique of
local spectrum sensing at each CR receiver; the Energy
Detector In the next two sub-sections a brief about technique
is provided.
3.1 Energy Detector
In this scheme, the received signal is sampled to generate a
finite discrete time samples series {xt ; t= 0, 1 . . . N-1}, where
t index of time. These samples are dot multiplied with
rectangular window. Hence, for each frequency bin ,fi the
decision statistic is computed by the summed energy over
samples as:
DECED(fi) = |^2 (7)
On the basis of central limit theorem, when L is large (e.g.
L>10),the decision statistic can be approximated to normal
distribution with the mean as given:

__________________________________________________________________________________________
E[DECED(fi)] = L for H0
= L(|H(fi)|^2 Es + forH1 (8)
and variance is defined as given:
E[DECED(fi)] = 2L for H0
= 2L (2|H(fi)|^2 Es + for H1 (9)
Here different local probabilities of ED-based spectrum
sensing can be evaluated by substituting (7) and (8), into (3)-
(5).
4. HARD COOPERATIVE SPECTRUM SENSING
The spectrum sensing technique used locally at each CR has
been described theoretically when ED is used. Generally, the
cooperation in spectrum sensing is achieved when a number of
CRs in the CR‟s network send their local decisions to the CR-
BS via Communication Channel. Then after, CR-BS combines
these decisions and decides finally about the presence of PR’s
signal in frequency bin fi. Here we note that, in our work, we
are interested to examine the performance when the power
spectrum is sensed at fi when the whole band under sensing is
occupied by PR’s signal in the case of H1. The hard
cooperative spectrum sensing starts from performing local
spectrum sensing using ED, The decision that rth
CR makes is
represented by binary digit br=’1’,or H0represented by binary
digit br=’0’, based on its own local decision statistics. At the
end, the CR-BS combines the received digits from different
CRs to declare the final decision about the presence of
primary signal.
All the received binary digits at the CR-BS from the different
CRs, in the CR network, are fused together to declare the final
decision using the logic rule as given below:
DECCOP (fi) = >= g for H1
= < g for H0 (10)
Where H1 represents that the final decision that has been
made by the CR-BS, stating that the PR’s signal is present in
fi, and H0 represents the PR signal’s absence. Number of CRs
g that decides the presence of PR’s signal at fi, determines the
type of fusion rule at CR Base Station. When g=1 out of total
G CRs, the fusion rule is “OR”, the fusion rule is AND if only
and if all g=G CRs decides H1 case. Lastly when 1<g<G the
“VOTING” fusion rule is applied.
In order to evaluate the cooperative spectrum sensing
performance, we define three joint probabilities; the joint
probability of detection, Qd(fi) the joint probability of false
alarm Qf (fi), and the joint probability of missed detection
Qm (fi). The joint probability of detection can be written as
given below:
Qd (fi)= P( DECCOP (fi) >= g/H1)
=
= (11)
And the joint probability of false alarm Qf(fi) can be written
as given below:
Qf(fi)= P( DECCOP (fi) >= g/H0)
=
= (12)
Here we note that DEC(fi) here means the decision statistic of
the used local sensing. So, finally the joint probability of
missed detection can be written as follows:
Qm (fi) = P( DECCOP (fi) < g/H1)
= 1 – Qd (fi) (13)
Now the total error probability of the cooperative CR
spectrum sensing is defined as given below [4]:
Qerror = Qm (fi) + Qf (fi) (14)
4.1 Local Spectrum Sensing Technique
To examine the performance optimization, of the hard
cooperative spectrum sensing, when local spectrum sensing
techniques are used; the total error probability Qerror are
evaluated at frequency bin fi, using Energy Detection. As we
mentioned earlier in this paper, the different probabilities will
be computed at a specific frequency bin fi. We have G = 10
CRs co-operate the spectrum sensing decisions, at a CR-BS, in
the CR’s network. The local spectrum sensing techniques is
ED. The local SNR = 10db, and L = 10 samples (i.e., OFDM
blocks) are used locally for sensing. Here fig. 1 shows the total
error probability ( Qerror) versus the chosen local threshold
for SNR = 10 db & g =5 by theoretical method using ED
technique.

__________________________________________________________________________________________
Fig. 2 shows the total error probability (Qerror) versus the
chosen local threshold for SNR = 10 db & g = 5 by monte-
carlo simulation method using ED technique.
Here fig. 3 shows the total error probability (Qerror) versus
the chosen local threshold for different number of g out G CRs
that controls the fusion rule in (13) using ED technique. If we
compare the different curves that represent the total error for
different numbers of g in Fig. 3, we observe, there are
noticeable difference in the performance through using g = 1
to 10 as a G=10 fusion rule. Here, g = 10 which represent
“AND” fusion rule, gives high total error compared to the
other curves; it is found that g = 5 gives the minimum total
error (min Qerror) at the Same values of SNR and thresold.
Hence, g = 5 is the optimal fusion rule here (i.e., goptimal =
2).
SNR = 10 db, L = 10
Fig. 1 Total error probability (Qerror ) for g = 5 CRs versus
local threshold when ED is used locally with SNR = 10 db and
L = 10 sensed samples used at each CR.(theoretical)
4.2 Different Number of G CRS
An interesting question now, is the goptimal that achieves min
Qerror same when the number of total CRs is different? Table
I shows the optimal fusion rule and min Q error when SNR is
varied and the ED is used locally, with same number of the
sensed samples (i.e. L = 10). The improvement in the
performance by increasing the total number G for different
SNR at CRs at fixed L, is noticeable. For example, min Qerror
= 0.2511 when SNR=5 db and CRs = 4 or 5, and min=0.00251
when SNR is increased to 10 db and CRS = 5. The increase in
SNR causes decrease in the min Qerrorwith variation in
number of CRs. Furthermore, for fixed SNR if the number of
the total co-operated CRs, G, is increased above optimal then
the Qerror is increased.
SNR = 10 db, L = 10
Fig. 2 Total error probability (Qerror ) for g = 5 CRs versus
local threshold when ED is used locally with SNR = 10 db and
L = 10 sensed samples used at each CR.(monte-carlo
simulation)
SNR = 10 db, L = 10
Fig. 3 Total error probability (Qerror ) for g out of G = 10 CRs
versus local threshold when ED is used locally with SNR= 10
db and L = 10 sensed samples used at each CR

__________________________________________________________________________________________
SNR = [5, 10, 13, 17, 18, 20] db, L = 10
Fig. 4 Total error probability (Qerror) for g out of G = 10
CRs versus local threshold when ED is used locally with
SNR= 5, 10, 13, 17, 18, 20 db and L = 10 sensed samples used
at each CR
In Fig. 3 we can see that for different SNRs we get different
results regarding minimum Qerror and goptimal as we change
the SNR error is decreased and also goptimal value is
changing.
Table 1: Optimal number of g CRs for different SNR and its
error level
SNR in db Error level Number of cognitive radio user
5 4 or 5
10 5
13 5 or 6
17 6
18 8
20 9
Table I shows tabular form of fig. 3. It shows the variation in
error level by changing SNR and respective Number of
cognitive radio user.
5. CONCLUSIONS
This paper contributes to the cooperative spectrum sensing
optimization by introducing an efficient optimization factor;
the number of cognitive radio user, G. As at each CR module
we may have different SNR, depending upon the value of
SNR we have derived the total error rate for different g (1 to
10). We have found the goptimal for each SNR. For SNR
value 5, 10, 17, 18, 20 we get goptimal as 4 or 5, 5, 5 or 6, 6,
8, 9.
REFERENCES
[1]. O. A. Alghamdi, M. A. Abu-Rgheff, and M. Z. Ahmed,
"MTM Parameters Optimization for 64-FFT Cognitive Radio
Spectrum Sensing using Monte Carlo Simulation," in
EMERGING 2010 : The Second International Conference on
Emerging Network Intelligence, Florence-Italy, 2010, pp. 107-
113.
[2]. O. A. Alghamdi, M. Z. Ahmed, and M. A. Abu-Rgheff,
"Probabilities of Detection and False Alarm in Multitaper
Based Spectrum Sensing for Cognitive Radio Systems in
AWGN," in The IEEE International Conference on
Communication Systems (IEEE ICCS 2010) Singapore: IEEE,
2010
[3]. J. Shen, S. Liu, L. Zeng, G. Xie, J. Gao, and Y. Liu,
"Optimisation of cooperative spectrum sensing in cognitive
radio network," Communications, IET, vol. 3, pp. 1170-1178,
2009.
[4]. T. Yucek and H. Arslan, "A survey of spectrum sensing
algorithms for cognitive radio applications," Communications
Surveys & Tutorials, IEEE, vol. 11, pp. 116-130, 2009.
[5]. Z. Wei, R. K. Mallik, and K. Ben Letaief, "Cooperative
Spectrum Sensing Optimization in Cognitive Radio
Networks," in Communications, 2008. ICC '08. IEEE
International Conference on, 2008, pp. 3411-3415.
[6]. QinetiQ, "Cognitive Radio Technology - A Study for
Ofcom – Summary Report," QINETIQ/06/00420, Issue 1.1,
February 2007.
[7]. E. Peh and L. Ying-Chang, "Optimization for Cooperative
Sensing in Cognitive Radio Networks," in Wireless
Communications and Networking Conference, 2007.WCNC
2007. IEEE, 2007, pp. 27-32.
[8]. A. Ghasemi and E. S. Sousa, "Collaborative spectrum
sensing for opportunistic access in fading environments,in
New Frontiers in Dynamic Spectrum Access Networks, 2005.
DySPAN 2005. 2005 First IEEE International Symposium on,
2005, pp. 131-136.
[9]. S. Haykin, "Cognitive radio: brain-empowered wireless
communications," IEEE journal on selected areas in
communications, vol. 23, pp. 201-220, 2005.
[10]. D. Cabric, S. M. Mishra, and R. W. Brodersen,
"Implementation issues in spectrum sensing for cognitive
radios," in Signals, Systems and Computers, 2004. Conference
Record of the Thirty-Eighth Asilomar Conference on, 2004,
pp. 772-776 Vol.1.
[11]. J. Mitola and G. Q. Maguire, "Cognitive radio: making
software radios more personal," IEEE personal
communications, vol. 6, pp. 13-18, 1999.

New optimization scheme for cooperative spectrum sensing taking different snr in cognitive radio networks

More Related Content

What's hot (20)

Viewers also liked (20)

Similar to New optimization scheme for cooperative spectrum sensing taking different snr in cognitive radio networks (20)

More from eSAT Publishing House (20)

Recently uploaded (20)

New optimization scheme for cooperative spectrum sensing taking different snr in cognitive radio networks