Cognitive Radio Spectrum Sensing Techniques and Performance Analysis

This report provides a comprehensive overview of cognitive radio spectrum sensing techniques, addressing the growing need for enhanced wireless communication and the challenges posed by current spectrum policies. It explores the role of cognitive radio in managing dynamic spectrum access and highlights spectrum sensing as a crucial initial step. The report delves into the cognitive cycle, detailing the analysis of the radio scene, channel identification, and spectrum decision. It focuses on three key non-cooperative spectrum sensing techniques: matched filter detection, energy detection, and cyclostationary feature detection, providing detailed explanations, block diagrams, and comparisons of their advantages and disadvantages. Furthermore, the report presents simulation results, comparing the performance of these techniques based on detection probability and probability of false alarm using MATLAB, with DVB-T signal parameters. The analysis concludes that cyclostationary detection outperforms the other methods, especially at lower signal-to-noise ratios.

Cognitive Radio
Spectrum Sensing Techniques and Cognitive Radio
Tayyab Ali Babar
Macquarie University
Sydney, Australia
[e-mail]
Abstract— There currently exists a growing global need to
enhance flexible wireless communication. This is largely due to
the fact that development of wireless networks has portrayed a
trend which will in prospect bring together both the mobile
systems and the internet systems in order to provide many
different services to the prospective customers. Spectrum is
therefore a very vital resource for the development of these
services. Cognitive radio equally has the ability to solve the lack
of or inadequate access to dynamic spectrum. Current policies on
spectrum present a major challenge in the development of
wireless communication technology. These policies have
ultimately resulted to inadequate usage of radio spectrum. In a
bid to outdo this problem, studies have revealed that cognitive
radio is the most appropriate technology that can be used to
provide an answer. Spectrum sensing appears to be the most vital
and initial step in the development of this technology. It is useful
in identifying bands of radio spectrum that are unused also called
spectrum holes. Since the emergence of CR technology, different
research works have been conducted in order to understand how
this technology can be used in managing the radio spectrum [1].
This report therefore intends to carry out a detailed survey on
the vital developments that have been made in spectrum sensing
as well as looking into its development since its emergence to its
current state. Relevant highlights have also been put on the
effectiveness and the challenges of different spectrum sensing
techniques and the common challenges that are always faced
when implementing the technology [2]
I. INTRODUCTION
The past decades have witnessed a fast rise in the number
of wireless communication devices together with the radio
spectrum static management. This has greatly led to lack of
available radio spectrum. Studies have revealed that an
estimated 50 billion devices will have a connection in the 2
years. The connections to all these devices will highly likely
be in need of accessing the internet. It is evident that
frequency spectrum is often reduced. The rising demand for
the frequency spectrum due to the constantly emerging
technologies and different uses for mobile users has been the
major reason for the limited availability of the frequency
spectrum given that it is not utilized evenly [4]. The radio
spectrum static management has no longer been effective
enough to allow access to all connected devices. Considering
this allocation, some segments of the radio spectrum are
highly utilized whereas are never used or used sparingly. The
concept of not sharing the radio spectrum with the users can
lead to the development of unwanted denial of service events.
The shortage of this technology therefore presents one of the
most critical concerns that should be looked at in the
prospective research works on networks that have not been
discussed [3].
One of the major and most vital answers to these concerns
and many others is to adopt the use of radio technology. The
radio spectrum is a technology that has been exposed to
several detailed investigations and studies by different
research communities for several decades. This technology
enables wireless devices to sense the radio spectrum. It is then
able to make a choice on the condition of the frequency
channels [5]. It then reconfigures the communication
parameters in order to attain high standard requirement of
service as it also minimizes the consumption of energy. The
devices are then able to use both unlicensed and licensed
bands whenever their main licensed users are not active hence
avoiding severe interface.
II. COGNITIVE CYCLE
Within the past decade, several technologies of sensing
have been suggested and can be grouped into 2 major
categories namely the wideband and the narrowband [6].
Wideband sensing is able to analyze several frequencies all at
one time whereas the narrowband sensing is able to analyze
just 1 frequency at one time. Examples of narrowband sensing
include detection of energy, sensing based on machine
learning as well as detection that is based on covariance.
Examples of wideband sensing includes spectrum that is
divided in several sub bands that are then sensed either
simultaneously or sequentially by the use of the techniques of
narrowband sensing. Approaches to sequential sensing are
generally not effective due to the fact that they need increased
energy and longer time due to the use of Analog to Digital
Converters of high rate (ADC). The ADCs are usually costly
and basically not practical for timely communications [7]. The
concept of simultaneous schemes of testing often need joint
synchronized functions and high amounts of sensors thus
raising the complexity of any given implementation.

The traditional schemes of licensing controls the results
and usage of frequency spectrum in instances where there is
reduced usage of the frequency spectrum. As a result of the
licensing scheme, there is emergence of spectrum holes.
Spectrum holes can be described as frequency bands that are
allocated to some locations but are mostly not exploited by the
licensed users. This therefore provides a chance to secondary
unlicensed users to utilize it [8]
The following cognitive cycle describes how a cognitive
radio functions.
Figure 1 Cognitive Cycle [9]
The major tasks that are performed from the cognitive
cycle include the following:-
 Analysis of radio scene. This handles the interference
temperature estimation as well as detecting the spectrum
holes. Is also referred to as sensing of spectrum [10].
 “Identification of channel. It handles the channel state
information (CSI) estimation. Also referred to as analysis of
spectrum”
 Power control transmit and management of dynamic
spectrum. Also known as spectrum decision.
“This research work mainly addresses the analysis of
performance for the 3 major techniques of spectrum sensing
under the category of non-cooperative [9]. The techniques
include;
 Matched filter detection
 Energy detection
 Cyclostationary feature detection”
III. SPECTRUM SENSING TECHNIQUES
This is among the most vital procedures of the cognitive
cycle. The purpose of spectrum sensing is to discover the
availability of transmissions from the basic users. There exists
3 major spectrum sensing types namely; “cooperative sensing,
non-cooperative sensing as well as interfaced based type of
sensing” [11]
Figure 2 Techniques of Spectrum Sensing [11]
A. Matched Filter Detection
“A matched filter represents a linear filter that makes best
use of the output signal to noise ratio for each input signal.
This technique is used mainly when the primary user priori
knowledge is known. During this procedure, the unknown
signal is often convolved with the response from the filter
impulse. This represents the mirror as well as the version of
the shifted time for the primary user signal. The mathematical
expression of the matched filter detection is as shown below”
𝑌ሺ𝑛ሻ= ෍ ℎሺ𝑛 − 𝑘ሻ𝑥(𝑘)
∞
𝑘=−∞
The following represents the block diagram for the
matched filter detection
Figure 3: “Block Diagram for the Matched Filter Operation” [11]
B. Energy Detection
“Detection of energy is basically the simplest technique of
sensing mainly because it doesn’t need any earlier information
or knowledge on the signal of the primary user. During this
operation, the detection of the primary user is on the basis of
the sensed energy. During this operation, the signal that is
received goes through the band pass filter [12]. The limited
signal of the band is then integrated within a given interval of
time. What follows is the comparison of the integrated time
signal with the earlier defined threshold.” This is carried out in
order to find out the availability of the primary signal. The
expression for the hypothesis testing for the identification of
the signal is as shown below”
Where
“n – The transmitted signal by the primary users
x (n) – received signal by the secondary user
w (n) – additive white Gaussian noise”

The hypothesis H0 indicates the unavailability of the
primary user while the hypothesis H1 indicates the availability
of the primary user [13]. The energy calculation can be
achieved by the mathematical formula shown below
෍ | 𝑥ሺ𝑛ሻ| 2
𝑁
𝑛=0
The energy is subsequently likened to the threshold in
order to find out the true hypothesis. The following
expressions will assist in the comparison.
The block diagram for the procedure is as shown below
Figure 4: Energy Detection Block Diagram [14]
C. Cyclostationary Feature Detection
The signal that is transmitted from the primary user always
portrays a pattern that is periodic. The pattern is described as
cyclostationary and can be used to determine the availability
of a licensed user. In case the signal autocorrelation is a
periodic function, then the signal will be described as a
cyclostationary signal. “The expansion of the autocorrelation
function in regards to the Fourier series coefficient, it results
to a function that depends only on frequency. The spectral
features for this procedure do not have any correlation to each
other. The expansion of the Fourier series is often described as
the cyclic auto-correlation function (CAF) whereas the related
frequency is known as the cyclic frequency.” “The calculation
of the cyclic spectral density (CSD) can be achieved by using
the Fourier transform of the CAF. This is a representation of
the correlation density between 2 spectral components. They
have a separation quantity that is equal to the cyclic frequency
[14]. The pattern is very vital when identifying the primary
user from the noise.
The signal that is received, x (t) is the cyclostationary
whenever the auto correlation and the mean portrays a
periodicity
Where; the auto correlation and mean period is represented
by T0. Whenever they and t are replaced with t-τ/2 and t+τ/2
in the equation, then
Where; Rax indicates the cyclic autocorrelation function
(CAF. a represents the cyclic frequency. The cyclic frequency
is often taken as a known parameter.” The calculation of the
CAF is as expressed below
The transmitted signals from the primary users are mainly
coupled by spreading codes, cyclic prefix etc. that leads to the
periodicity of the statistics like the auto correlation and mean.
When the CFD calculations for these signals are attained, they
are useful in highlighting the stated periodicities. The Fourier
transform results for the correlated signals in the frequencies
peak that are specific to the signal are important when
determining the availability of the primary users while noise
generally has a random nature. It also does not portray the
stated periodicities thus is not often highlighted whenever
correlation is carried out [15]. The block diagram for the
cyclostationary process is as shown below
Figure 5: Cyclostationary Feature Detection Block Diagram [16]
To Sum these three Techniques I have recorded the
advantages and disadvantages in the table below
Table 1: Pros and Cons of Sensing Techniques
IV. RESULTS AND ANALYSIS
All the simulations for performing an analysis for these
sensing techniques were conducted through MATLAB. The
major reason for performing the simulations was to carry out a

comparison of the operation performance ability for the 3
three sensing techniques [17]. The metrics of performance that
are considered in this study involves the detection probability
and the probability of false alarm. The parameters of
simulation have been shown in the table below. The primary
signal that is preferred for the simulation of this study is the
“digital video broadcast- terrestrial (DVB-T) signal at
90MHZ.” The DVB-T parameters have also been illustrated
as shown below
Table 2 Simulation Parameters [17]
Table 3 The DVB-T Parameters [17]
The results for simulation for the metrics have been shown
in the figures below. It can be observed that the detector of
cyclostationary is able to attain a 100% detection at close to -
8db SNR while the in a bid to attain the energy of detection,
the detector requires close to 8db of SNR as well as matched
filter detector that can attain similar performance of detection
at close to 15db SNR. “It can therefore be concluded that the
cyclostationary detector that is suggested here performs better
than the other two main detectors by a wider margin since it
does best at reduced SNR which is the most vital DVB-T
aspect basic signal as per the standards of IEEE 802.22 of
cognitive radio” [18]
Figure 6 Detection Probability [18]
The figure below represents the “probability of false
detection as a function of SNR. From the graph, it is evident
that false detection probability for the energy detectors rises
with a fall in the SNR. That implies that the SNR becomes
negative as the probability of false detection rises for the
energy detector.” The matched filter similarly follows that
path through the probability of detection for all the approaches
for the detectors to the minimum at 14db and 8db SNR. After
observation of the cyclostationary detector, it is evident that
the false detection probability gets close to zero, implying that
it is negligible when compared to other 2 detectors. It can be
concluded that the cyclostationary detector parameter has a
higher performance than the remaining 2 detectors [19].
Figure 7 False Alarm Probability [18]
V. SHORTCOMINGS OF COGNITIVE RADIO NETWORK
An effective cognitive radio should have the ability to
make use of the existing freedom frequency degree, space and
time. Considering the cognitive radio aspect, in case each node
is having the cognitive features, then it is possible to improve
the efficiency of transmission to a higher extent. There
however still exist several issues when developing an effective
and robust model of the cognitive radio for efficient sharing of
spectrum. The major issues have been described below
A. Real oppurtunity and economical aspect

There is need to determine the engineering and economic
benefits of adopting the use of CRN-based systems over the
traditional systems of wireless communications. Besides that,
the underlying issues on models of network economy should
be designed such that they the commercial community is able
to get comfortable with the CRNs. Several measurements of
the spectrum are needed in order to bring an understanding of
the number of spectrum holes that are considered viable
commercially. The reduced utilization does not imply that the
SUs can make use of the opportunity in a way that is sensible
economically [20].
B. Combined spectrum sensing and access
The process of developing sensing of spectrum and access
always occurs individually due to the fact that sensing of
spectrum is able to achieve specific performances of detection.
The spectrum access equally aims at improving the system
capacity on the basis that is recognized by the spectrum hole.
Theses 2 aspects are however coupled inevitably. For
instance, various power level transmissions for the users of
cognitive radio may need various decisions models in the
sensing of spectrum. Another challenging concern may also be
the joint development of distributed random access and
multichannel sensing.
C. Common control channel
There currently exists an issue on whether to develop a
common channel of control for the operations of the cognitive
radio. A common channel of control will open the way for the
path to a simpler way of facilitating exchange of information
during the activities of spectrum sensing as well as access in
the cognitive radio networks. However, unlike in the case of
conventional networks, a common channel of control could
not be available in the previous phase when the spectrum
holes are not effectively recognized. A channel that is not
identified could also be preoccupied by the Pus no matter the
time. This could possibly interfere with the messages of
coordination when it is used as a channel for common control.
The concept of setting up and maintaining a common control
channel is basically vital for proper CRNs operations.
D. Replacing spectrum sensing in cognitive radio networks
There basically exist 2 major problems from all the above
discussed issues. The 2 basic challenges that are realized are
hidden terminal problems as well as spectrum sensing. “The
service providers for the SUs are needed in order to place the
devices of sensing within the networks of primary users or the
licensed users. The devices of sensing are able to spy on the
activities of the licensed users. The sensing devices could also
be used to make a decision on whether the transmission of the
secondary users can be admitted. For this reason, a new
cognitive cycle has also been proposed appropriately. A
technique of reduced temperature handshake has also been
proposed for the handshakes that exist between the sensing
devices and the secondary users for the instances where there
is no separate control channel” [20]
VI. CONCLUSION
This research work has looked into the 4 basic functions of
a cognitive radio. “The functions include spectrum
management, spectrum sensing, spectrum mobility as well as
spectrum sharing.” A detailed analysis of all the spectrum
techniques has also been discussed. The findings from this
research work can be vital for the researchers who would like
to carry studies on the wireless networks field. This is because
it has presented a detailed account of the several techniques if
spectrum sensing that currently exists in this industry.
Different stakeholders within the industry can also adopt the
findings from this study in order to make improvements that
will improve the efficiency of the techniques of wireless
networks.
“Several frequency spectrum bands have a license to
various services in order to prevent interference that could
occur between different networks. Spectrum occupancy
measurements however portray that just a few segments of the
bands are used effectively. Cognitive radio therefore a
technology that is conscious of its internal state, its
environment and possess the ability to change the parameter of
transmission very effectively. This technology enables the
unlicensed users to make use of the underutilized spectrum
without any possible interference or harm to the licensed
users. The main domains of cognitive radio include
adaptation, cognition and sensing. The concept of spectrum
sensing has been a major challenging concern in the systems
of cognitive radio that detect the frequency bands that are
available. This research work has therefore adopted different
techniques of detecting transmission. These techniques include
matched filter, energy detection as well as cyclostationary
detection feature. MATLAB software was used to perform the
analysis.”
REFERENCES
[1] Akan, O.B., Karli, O.B. and Ergul, O., 2018. Cognitive radio sensor
networks. IEEE network, 23(4), pp.34-40.
[2] Akyildiz, I.F., Lee, W.Y., Vuran, M.C. and Mohanty, S., 2016. NeXt
generation/dynamic spectrum access/cognitive radio wireless networks:
A survey. Computer networks, 50(13), pp.2127-2159.
[3] Akyildiz, I.F., Lee, W.Y., Vuran, M.C. and Mohanty, S., 2018. A survey
on spectrum management in cognitive radio networks.
[4] Ariananda, D.D., Lakshmanan, M.K. and Nikookar, H., 2015, May. A
survey on spectrum sensing techniques for cognitive radio. In 2015
Second International Workshop on Cognitive Radio and Advanced
Spectrum Management (pp. 74-79). IEEE.
[5] Axell, E., Leus, G., Larsson, E.G. and Poor, H.V., 2012. Spectrum
sensing for cognitive radio: State-of-the-art and recent advances. IEEE
Signal Process. Mag., 29(3), pp.101-116.
[6] Cabric, D., Mishra, S.M. and Brodersen, R.W., 2014, November.
Implementation issues in spectrum sensing for cognitive radios. In
Conference Record of the Thirty-Eighth Asilomar Conference on
Signals, Systems and Computers, 2004. (Vol. 1, pp. 772-776). Ieee.
[7] Chen, R. and Park, J.M., 2016, September. Ensuring trustworthy
spectrum sensing in cognitive radio networks. In 2016 1st IEEE
Workshop on Networking Technologies for Software Defined Radio
Networks (pp. 110-119). IEEE.
[8] Chen, R., Park, J.M. and Bian, K., 2018, April. Robust distributed
spectrum sensing in cognitive radio networks. In IEEE INFOCOM

2008-The 27th Conference on Computer Communications (pp. 1876-
1884). IEEE.
[9] Farhang-Boroujeny, B. and Kempter, R., 2018. Multicarrier
communication techniques for spectrum sensing and communication in
cognitive radios. IEEE Communications Magazine, 46(4), pp.80-85.
[10] Ghasemi, A. and Sousa, E.S., 2018. Spectrum sensing in cognitive radio
networks: requirements, challenges and design trade-offs. IEEE
Communications magazine, 46(4), pp.32-39.
[11] Hur, Y., Park, J., Woo, W., Lim, K., Lee, C.H., Kim, H.S. and Laskar,
J., 2016, May. A wideband analog multi-resolution spectrum sensing
(MRSS) technique for cognitive radio (CR) systems. In 2016 IEEE
International Symposium on Circuits and Systems (pp. 4-pp). IEEE.
[12] Kaligineedi, P., Khabbazian, M. and Bhargava, V.K., 2018, May. Secure
cooperative sensing techniques for cognitive radio systems. In 2008
IEEE International Conference on Communications (pp. 3406-3410).
IEEE.
[13] Quan, Z., Cui, S., Sayed, A.H. and Poor, H.V., 2018. Optimal multiband
joint detection for spectrum sensing in cognitive radio networks. IEEE
transactions on signal processing, 57(3), pp.1128-1140.
[14] Quan, Z., Cui, S., Sayed, A.H. and Poor, H.V., 2018. Wideband
spectrum sensing in cognitive radio networks. arXiv preprint
arXiv:0802.4130.
[15] Subhedar, M. and Birajdar, G., 2011. Spectrum sensing techniques in
cognitive radio networks: A survey. International Journal of Next-
Generation Networks, 3(2), pp.37-51.
[16] Sun, H., Nallanathan, A., Wang, C.X. and Chen, Y., 2013. Wideband
spectrum sensing for cognitive radio networks: a survey. IEEE Wireless
Communications, 20(2), pp.74-81.
[17] Tian, Z. and Giannakis, G.B., 2016, June. A wavelet approach to
wideband spectrum sensing for cognitive radios. In 2016 1st
international conference on cognitive radio oriented wireless networks
and communications (pp. 1-5). IEEE.
[18] Tian, Z. and Giannakis, G.B., 2017, April. Compressed sensing for
wideband cognitive radios. In 2007 IEEE International Conference on
Acoustics, Speech and Signal Processing-ICASSP'07 (Vol. 4, pp. IV-
1357). IEEE.
[19] Yucek, T. and Arslan, H., 2019. A survey of spectrum sensing
algorithms for cognitive radio applications. IEEE communications
surveys & tutorials, 11(1), pp.116-130.
[20] Zou, Y., Yao, Y.D. and Zheng, B., 2012. Cooperative relay techniques
for cognitive radio systems: Spectrum sensing and secondary user
transmissions. IEEE Communications Magazine, 50(4), pp.98-103.
APPENDIX
MATLAB CODE
Code taken from Indian Institute of Technology Kanpur
“clc
close all
clear all
L = 1000;
snr_dB = -14; % SNR in decibels
snr = 10.^(snr_dB./10); % Linear Value of SNR
Pf = 0.01:0.01:1; % Pf = Probability of False Alarm
%% Simulation to plot Probability of Detection (Pd) vs. Probability of False
Alarm (Pf)
for m = 1:length(Pf)
m
i = 0;
for kk=1:10000 % Number of Monte Carlo Simulations
n = randn(1,L); %AWGN noise with mean 0 and variance 1
s = sqrt(snr).*randn(1,L); % Real valued Gaussina Primary User Signal
y = s + n; % Received signal at SU
energy = abs(y).^2; % Energy of received signal over N samples
energy_fin =(1/L).*sum(energy); % Test Statistic for the energy detection
thresh(m) = (qfuncinv(Pf(m))./sqrt(L))+ 1; % Theoretical value of Threshold,
refer, Sensing Throughput Tradeoff in Cognitive Radio, Y. C. Liang
if(energy_fin >= thresh(m)) % Check whether the received energy is greater
than threshold, if so, increment Pd (Probability of detection) counter by 1
i = i+1;
end
end
Pd(m) = i/kk;
end
plot(Pf, Pd,'linewidth',2)
hold on
grid on
%% Theroretical ecpression of Probability of Detection; refer above
reference.
thresh = (qfuncinv(Pf)./sqrt(L))+ 1;
Pd_the = qfunc(((thresh - (snr + 1)).*sqrt(L))./(sqrt(2).*(snr + 1)));
plot(Pf, Pd_the, 'r-')
xlabel('probability of false alarm')
ylabel('probability of detection')
hold on
grid on”