IEEE COMMUNICATIONS LETTERS, VOL. 18, NO. 11, NOVEMBER 2014 1987
A Robust Cooperative Spectrum Sensing Method in Cognitive Radio Networks
RuiGao,ZanLi,Member, IEEE, Peihan Qi, and Husheng Li, Member, IEEE
Abstract—In real cognitive radio networks, the requirement for
spectrum sensing in the case of lacking prior information and
dynamically varying noise levels indeed poses a severe challenge.
Motivated by these, in this letter, a Cooperative Power spectral
density Split Cancelation (CPSC) method is proposed. The CPSC
makes full use of the asymptotic normality and independence of
a Fourier transform to get the stochastic properties of the power
spectral density. We derive an exact closed-form expression of a
decision threshold and provide an analysis about computational
complexity. Analytical and simulation results show that the CPSC
is robust under uncertain noise. Meanwhile, the CPSC benefits
from a low computational cost and existing mature intellectual
property cores; thus, it can be implemented efficiently.
Index Terms—Cognitive radio, cooperative spectrum sensing,
robust detection, noise uncertainty.
I. INTRODUCTION
T
O address the conflicts between a huge demand on the
deployment of wireless services and increasingly scarce
spectrum resources, cognitive radio (CR) [1] has emerged to
enable the access of the intermittent periods of unoccupied
frequency bands. The fundamental task of each secondary user
(SU) is to detect the primary users (PUs). This process is
called spectrum sensing which has twofold objectives: first,
SUs should not cause harmful interference to PUs by switching
to an available band; second, SUs should efficiently identify
and exploit the spectrum holes for required throughput. Thus,
the spectrum sensing performance is crucial to the performance
of CR networks.
Although there has been considerable work on spectrum
sensing, there still exist many severe challenges. First, given
the protection for PUs and the inaccuracy of estimating electro-
magnetic environment parameters in low SNR regime, a spec-
trum sensing method without any prior information is a severe
challenge [2]. Moreover, many factors in practice such as noise
uncertainty (including receiver device noise uncertainty and
environment noise uncertainty [3]) may seriously compromise
the detection performance. Finally, the sensing time must be
limited at an acceptable level while guarantee sufficiently low
detection error probability. Hence, the robustness and efficiency
Manuscript received April 11, 2014; revised August 21, 2014; accepted
September 14, 2014. Date of publication October 8, 2014; date of current
version November 7, 2014. This work was supported in part by the National
Natural Science Foundation of China under Grants 61301179 and 61401323,
by the Doctoral Programs Foundation of the Ministry of Education under Grant
20110203110011, and by the China Scholarship Council. The associate editor
coordinating the review of this paper and approving it for publication was
D. B. da Costa.
R. Gao, Z. Li, and P. Qi are with the State Key Laboratory of Integrated
Service Networks, Xidian University, Xi’an 710071, China (e-mail: rgao_1@
stu.xidian.edu.cn; zanli@xidian.edu.cn; phqi@stu.xidian.edu.cn).
H. Li is with the Department of Electrical Engineering and Computer Sci-
ence, College of Engineering, University of Tennessee Knoxville, Knoxville,
TN 37996 USA (e-mail: husheng@eecs.utk.edu).
Digital Object Identifier 10.1109/LCOMM.2014.2361851
are essential characters for the spectrum sensing method in
realistic scenarios.
To this end, several spectrum sensing methods have been
proposed and investigated which can be seen in [4]–[10] and
references therein. Without prior knowledge of PU, energy de-
tection [6] is optimal when signals are uncorrelated. However, it
has the issue of noise uncertainty, which may render the energy
detection useless [7]. Cyclostationary detection [8] is more
robust to noise uncertainty than energy detection by exploiting
the cyclostationary feature embedded in the received signal.
Nevertheless, it has the shortcomings of high computational
complexity and long sensing time. Eigenvalue-based detection
methods [3], [9], [10] can be used for various signal detection
and applications without knowledge of the signal, channel,
and noise. Nevertheless, it is very difficult (mathematically
intractable) to obtain a precisely closed form formula for the
threshold. Although [9], [10] have made progress in this issue,
it is not resolved absolutely yet. And this issue leads to some
troubles in application as optimal thresholds need the help of
empirical value.
Considering the challenges mentioned above, especially
most existing spectrum sensing methods are limited by dy-
namical noise, a Cooperative Power spectral density Split
Cancellation method (CPSC) was proposed in this letter. The
meanings of dynamical noise are twofold here: first, the noise
power is not certainty which is also called noise uncertainty
[7]; second, the noise levels of different SUs are not identical.
Analytical and simulation results show that the CPSC has the
inherent advantages of robustness under dynamical noise, low
computational complexity and efficient implementation.
II. S
PECTRUM SENSING PRELIMINARIES
In this section, we present the general model of cooperative
sensing. Spectrum sensing for primary signal detection can be
formulated as a binary hypothesis problem as follows
x
u
(t)=
ω
u
(t), H
0
h
u
(t)s(t)+ω
u
(t), H
1
, (1)
where u =1, 2,...,U. We make the following assumptions
• (AS1) x
u
(t) denotes the received signal at the uth SU,
s(t) is a real unknown deterministic PU signal, h
u
(t) is
the channel coefficient of the sensing channel;
• (AS2) ω
u
(t) is additive Gaussian noise which is the circu-
lar symmetric complex Gaussian (CSCG) independent and
identically distributed (i.i.d.) random process with mean
zero and variance σ
2
u
, namely ω
u
(t) ∼CN(0,σ
2
u
).
• (AS3) h
u
(t) is a complex random process with mean μ
u,h
and variance σ
2
u,h
, which is independent of ω
u
(t). Since
we only consider slow fading channel in this letter, h
u
(t)
is fixed in a symbol period.
1089-7798 © 2014
IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.