Al-Mukhtar Journal of Sciences 00 (0): 00-00, 2021

ISSIN:

Performance Metrics in Radio Networks

Received: 2021/ Accepted: 2021

© Al-Mukhtar Journal of Sciences 2021
Doi:

Abstract: In cognitive radio networks CRN the main aim is to allow the secondary users SUs to
identify the empty bands and use them to transmit or receive data opportunistically. Primary users
PUs have the priority to use a channel while the secondary users must vacant this channel once a
primary user request it. An attack known in the cognitive radio as a primary user emulation
attacks (PUEA), its aim is to prevent the SU from using the empty bands.
In this paper, an analytical and experimental approach is presented to mitigate the PUEA, this
approach is based on obtaining the probability density functions PDFs of the received powers at
the secondary users from malicious nodes and also from the primary transmitter in the cognitive
network. Then, these obtained PDFs are used in Neyman-pearson composite hypothesis test to
measure the performance metrics (probability of false alarm and miss detection in the network).
The results proved that the performance metrics greatly influenced by the network area where the
secondary user presented and the threshold value λ that is used in the decision rule. Also there are
boundaries for the λ choices that can not be overtaken.

Keywords: Cognitive Radio CR; Probability Density Function (PDF); Primary User Emulation
Attack (PUEA).

INTRODUCTION communication, recently the operational

aspects of CR and its security aspects have
The four main functions of cognitive radio
gained a great deal of attention (FCC 2003).
are spectrum sensing, spectrum decision
The reasons that make the cognitive radio
spectrum sharing and spectrum mobility, via
vulnerable to new kind of security threats are
these functions the secondary users in the CR
as following:-
must be able to distinguish between the
- The open and dynamic features of the
available channels and used channels. For
cognitive radio network that makes the CR
example, if a TV transmitter acts as a primary
systems more vulnerable to various malicious
user, then the users that can use the white
attacks. The new threats such as jamming,
spaces (available channels) in the TV band
primary user emulation (PUE) and spectrum
are called cognitive users ( Buddhikot 2005 ).
sensing data falsification SSDF (Jin 2009).
The major challenge in spectrum sensing is
- Because the cognitive radio network shares
that the ability of the cognitive radio to
some features of the conventional wireless
recognize the spectrum bands that are not
networks , so we have to deal with the
used by the primary users so the secondary
conventional wireless security risks in
users can coexist with the primary users
addition to the threats targeted the cognitive
without any interference to their

*Corresponding Author: Mahmoud Ammar, m.ammar@zu.edu.ly, Department of Computer Engineering, University of Zawia, Zawia, Libya

1
Al-Mukhtar Journal of Sciences 00 (0): 00-00, 2021
ISSIN:

radio features, the conventional risks include A. Selfish attacks

MAC spoofing , Denial of Service …etc. The attacker’s motive is to acquire more
(Akyildiz 2006). spectrum for its own use by preventing others
Based on these vulnerabilities, from competing for the channels and unfairly
countermeasures are needed to make the occupying their share. In this type of attack,
cognitive radio networks robust and secure adversaries will defy the protocols and
against any kind of threat. policies only if they are able to benefit from
Cooperative spectrum sensing can be them (Bhattacharjeea 2013, Mathur 2007).
vulnerable when some malicious nodes share B. Malicious attacks
false local sensing reports to others. In such
The attacker’s only objective is to create
cases, the fused decision may be altered, hindrance for others and does not necessarily
hence the reliability of cooperative spectrum aim at maximizing own benefits. They do not
sensing. Such phenomenon where local have any rational objective and identify
sensing result is manipulated is known as protocols and policies to just induce losses to
Spectrum Sensing Data Falsification (SSDF) others (Jakimoski 2008).
or Byzantine attack. A malicious radio can
advertise ‘occupied’ as ‘available’ inducing a IMPACT OF PUE ATTACKS ON CR
policy violation or advertise ‘available’ as NETWORKS
‘occupied’ causing denial of spectrum usage.
Environment and changing the parameters in The presence of PUE attacks causes a number
of troubles for CR networks. The list of
order to improve the quality of service are
potential consequences of PUE attacks is:
achieved based on the main functions of the
cognitive radio. In adversarial, military, and 1. Bandwidth waste: The ultimate
heterogeneous competitive networks, such objective of deploying CR networks
actions are not surprising where an adversary is to address the spectrum under-
wants to cripple the operation of others in the utilization that is caused by the
network. Apart from this, there are also cases current fixed spectrum usage
where a node’s permanent spatial orientation policy. By dynamically accessing
is such that its reports are not suitable for use the spectrum “holes”, the SUs are
able to retrieve these otherwise
by other nodes. The adversary may vary its
wasted spectrum resources (Cabric
attack strategies based on different objectives. 2004).
Hence there is a need to evaluate the 2. QoS degradation: The appearance
trustworthiness of radios before considering of a PUE attack may severely
their local spectrum sensing reports (Chen degrade the Quality-of-Service
2006). (QoS) of the CR network by
destroying the continuity of
OBJECTIVE OF ADVERSARIAL secondary services (Cabric 2004).
3. Connection unreliability: If a real-
ATTACKERS time secondary serviceis attacked
The objectives of an attacker have a direct by a PUE attacker and finds no
correlation with the way the attacks are available channel when performing
launched, and therefore they determine the spectrum handoff, the service has to
nature of attacks. be dropped. This real time service
is then terminated due to the PUE
attack. In principle, the secondary

*Corresponding Author: Mahmoud Ammar, m.ammar@zu.edu.ly, Department of Computer Engineering, University of Zawia, Zawia, Libya

2
Al-Mukhtar Journal of Sciences 00 (0): 00-00, 2021
ISSIN:

services in CR networks inherently Figure 1. Radio network users distributions

have no guarantee that they will
have stable radio resource because
of the nature of dynamic spectrum
access. The existence of PUE
attacks significantly increases the
connection unreliability of CR
networks.
If M denotes the number of malicious users
in the system , P t : Primary transmition power
,Pm : Malicious transmition power , σ p :
Variance of Primary users , σ m :Variance of
Malicious users. The simulation model novel
parameters are as in the table I below.

SIMULATION MODEL Table I. The simulation model parameters

In practice , the SU aims to identify the signal
source, whether it is from a primary
transmitter (PT) or malicious users. This Parameter dp R Ro M Pt Pm σp σm
scenario model simulated as shown in Figure
1 where the SUs and malicious users are Value 160 300 30 20 100 10 8 5.5
located in a circular area with radius R. KM m m KW W dB dB

Some assumptions a are considered as follow:

The primary user is at a distance dp from the MATHEMATIC FUNCTIONS OF THE
secondary user . The malicious users are
distributed randomly around the good user as RECEIVED SIGNALS
in Figure 1. this model aims to analyze the To calculate the probability density functions
performance metrics under long distance d p = PDFs and use it in the simulated model, the
160 km to the SU. In the same time the received power Pr is determined first based
network radios R and Ro are too small R = on the relation below:
300 m and R0 = 30 m. such parameters for
analysis in CR networks are chosen as a novel  1
Pr d
study of the model under these low
-parameters to prove the efficiency of the γ is the path loss exponent. Thus, the
analytical model . received power from the primary is :

2
p( p )  P d  k 2
r t p

p/10
K 10
Where k is the path

loss
Thus the probability density function of the
Good User
Malicious
Primary received power is
user
user

*Corresponding Author: Mahmoud Ammar, m.ammar@zu.edu.ly, Department of Computer Engineering, University of Zawia, Zawia, Libya

3
Al-Mukhtar Journal of Sciences 00 (0): 00-00, 2021
ISSIN:

1
2
(10 log10 x−μ p ) The results obtained using Matlab simulation
Pr
P ( x)=
Aσ p √ 2πx (
exp −
2σ 2p ) 3 are presented in this section, the performance
of the cognitive network under the PUE
attack is studied in terms of probability of
 miss detection and false alarm.
µp , σp are the mean and variance of the
distribution , and given by Both the probability of miss detection and
false alarms are calculated for 1000 times of
  10 log p  20 log d simulations to be averaged to perform
p 10 t 10 p accurate results. The result obtained first
based on the threshold value λ=0.45 . The
Secondly, the received power from each network is assumed as it’s under a high
malicious user P (mi) at the secondary user is number of malicious attackers M=20. In this
calculated according to the equation below: case it is noticed that when λ is too small ,the
( mi) −4 achieved false alarm probability is very small
P =( PT ) d i K 4 which in this case is about 0.05 as shown in
mi
Figure 2 below. In the same time, the miss
Where k is the path loss . d i is the distance
detection probability is high and is about
from the secondary user.
0.325 as shown in Figure 3. These values are
Based on the above equations , the decision averaged out of 1000 runs.
variable Λ that is used to find the false alarm
and miss detection probabilities can be
calculated as follow
m Pr
Λ= P (χ) / P (χ) 5

Pr
Where P ( χ ) and P ( χ ) is the pdf of
m

received powers from the primary and from

all malicious users respectively.
After performing Λ , it is compared with a
threshold λ that can be chosen to guarantee a
fine false alarm and miss detection
probabilities as follow:
IF Λ ≤ λ then Figure 2. False alarm probability for λ=0.45
The decision is that the transmission is from a
primary user
IF Λ>λ then
The decision is that the transmission is from
malicious user.
RESULTS DISCUSSION AND
ANALYSIS

*Corresponding Author: Mahmoud Ammar, m.ammar@zu.edu.ly, Department of Computer Engineering, University of Zawia, Zawia, Libya

4
Al-Mukhtar Journal of Sciences 00 (0): 00-00, 2021
ISSIN:

between λ and miss detection while a positive

correlation between λ and the false alarm i.e.
as λ decreases, false alarm decreases and
miss detection increases.
To sum up , choosing the value out λ of the
boundaries leads to increasing one of these
probabilities, which leads to a wrong decision
about presence of the primary user.

CDF OF FALSE ALARMS AND MISSED

DETECTIONS
In order to testify these results, the
Figure 3. Miss detection probability for λ=0.45 cumulative distribution function (CDF) is
calculated to display both the false alarms and
Secondly , the model is justified under high the missed detection probabilities on the same
value of λ, λ=2.2 , thus the false alarm graph as shown in Figure 4 below. Notably,
probability is high which in this case is the CDF plot is a non-decreasing function and
0.43 ,opposed to the miss detection this indicates that the parameters and
probability which is very small and is 0.17. assumptions that were considered in the
These results summarized in table II as simulation are well chosen and the results are
follow. accurate because these cumulative
distributions functions in Figure 4 match and
follow the general appearance of the CDF of
Table II. Miss detection and false alarm for different λ
the continuous variable of the simulated
Paramete False Alarm Miss Detection
model.
r Probability Probability
Averaged for 1000 Averaged for 1000 runs
runs
λ =0.45 0.05 0.325
λ =2.2 0.43 0.17

These results mean that the threshold value of

the simulated mathematics model must be
chosen so that it doesn’t exceed 2.2 and not
less than 0.45. These boundaries of the
threshold value λ can guarantee acceptable
values of the performance metrics (miss
detection and false alarm). In addition, it has
proved that at least one of these performance
metrics is low because of the mathematic
relationship, which forms an inverse
Figure 4. Cumulative distribution functions for miss
correlation. i.e. there is inverse correlation
detection and false aslarm

*Corresponding Author: Mahmoud Ammar, m.ammar@zu.edu.ly, Department of Computer Engineering, University of Zawia, Zawia, Libya

5
Al-Mukhtar Journal of Sciences 00 (0): 00-00, 2021
ISSIN:

CONCLUSION I am also so grateful to all those people who

have supported me and had contributions in
making this work possible, my upmost thanks
Because of the limitations and problems in to the engineering faculty staff, and Zawia
the conventional spectrum sensing and radio University in general for the valuable
network approaches, this paper focused on suggestions and support.
one of the major threats in the radio networks
which is the primary user emulation attack. In Finally, I wholeheartedly thank my great
our proposed model and using an analytical parents, my wife and my lovely family for
approach, the security against primary user their understanding, encouragement and
emulation attacks in radio networks is support.
discussed. The proposed analytical model
and its impact on the PUEA attack is REFERENCES
investigated.
Akyildiz, I. (2006). Next generation/dynamic
The performance metrics results (the spectrum access/cognitive radio
probability of false alarm and missed wireless networks. A survey,” (Elsevier
detection in the network) proved that the Journal), on computer Networks,
number of malicious nodes in the system has 50(13), 2127-2159.
a great impact on the network and this has led
to a reduction in the quality of service due to Bhattacharjeea, S. (2013). In cognitive radio
the transmission from a high number of networks. The International Journal for
malicious users. the Computer and Telecommunications,
Also it has proved that at least one of the 01(36) 1387-1398 .
performance metrics is low as result of the
Buddhikot, M., & Ryan, K. (2005). Spectrum
inverse correlation between λ and miss
management in coordinated dynamic
detection probability. In other words, there is
spectrum access. Proc. of IEEE DySpan,
inverse correlation between λ and miss
10-1109.
detection while a positive correlation between
λ and the false alarm, this means if λ Cabric, D. & Brodersen, R. (2004).
decreased, false alarm decreases and miss Implementation issues in
detection increases. For each model there is a spectrumsensing for cognitive radios. in
boundary for value of λ depends on the Proceedings of the Thirty-eight
parameters used in the model . Asilomar Conferenceon Signals,
Systems, 0-7803-8622-1.
ACKNOWLEDGEMENTS
Chen, R. & Park, J. (2006). Ensuring
First and foremost, I would like to thank my trustworthy spectrum sensing in
department team at the engineering faculty cognitive radio networks. Proceedings,
for the great guidance, advices and IEEE Workshop on Networking
encouragement over the years. My sincere Technol. for Software Defined Radio
gratitude to all academics and colleagues Networks (SDR), 10–119.
within our research group.

*Corresponding Author: Mahmoud Ammar, m.ammar@zu.edu.ly, Department of Computer Engineering, University of Zawia, Zawia, Libya

6
Al-Mukhtar Journal of Sciences 00 (0): 00-00, 2021
ISSIN:

Federal Communications Commission FCC.

(2003). NPRM - Facilitating
Opportunities for Flexible, Efficient,
and Reliable Spectrum Use Employing
Cognitive Radio Technologies. FCC,
03-322.

Jakimoski, J. (2008). Denial-of-service attacks

on dynamic spectrum access networks.
IEEE CogNets Workshop, IEEE
International Conference on
Communications (ICC), 10.1093.

Jin, Z. & Anand, S. (2009). Detecting primary

user emulation attacks in dynamic
spectrum access networks,” In
Proceedings of the IEEE International
Conference on Communications,
Dresden, Germany, 04-2009.

Mathur, C. (2007). Security issues in cognitive

radio networks. Cognitive Networks:
Towards Self-Aware Networks, John
Wiley & Sons, Ltd, pp. 271–291.

*Corresponding Author: Mahmoud Ammar, m.ammar@zu.edu.ly, Department of Computer Engineering, University of Zawia, Zawia, Libya

7
 ‫مجلة المختار للعلوم ‪2018 ,00-00 :)0( 00‬‬

‫شبكات الراديو الذكية االستدراكية المعتمدة على ثقة المستخدمين‬

‫تاريخ االستالم‪ 15 :‬مايو ‪ / 2021‬تاريخ القبول‪2021 :‬‬

‫© مجلة المختار للعلوم ‪2021‬‬
‫‪:Doi‬‬

‫الملخص‪ :‬في شبكات الراديو المعرفية)‪ Cognitive Radio Networks ( CRN‬الهدف الرئيسي هو السماح للمستخدمين الثانويين‬
‫(‪ Secondary users )SUs‬بتحديد‪ L‬النطاقات الفارغة واستخدامها لنقل البيانات أو استقبالها بشكل انتهازي‪ .‬يتمتع المستخدمين‬
‫االساسيين )‪ Primary Users (Pus‬باألولوية في استخدام قناة او تردد ما بينما يجب على المستخدمين الثانويين إخالء هذه القناة‬
‫بمجرد أن يطلبها المستخدم األساسي‪ .‬هناك هجوم امني ُيعرف في الراديو المعرفي بأنه هجوم محاكاة المستخدم األساسي ‪Primary‬‬
‫)‪ ، user emulation attack (PUEA‬ويهدف إلى منع ‪ SU‬من استخدام النطاقات الفارغة‪.‬‬

‫في هذه الورقة ‪ ،‬تم تقديم نهج تحليلي وتجريبي للتخفيف من ‪ ، PUEA‬ويستند هذا النهج إلى الحصول على وظائف الكثافة‬
‫االحتمالية ‪ .( PDF) Probability density function‬بعد ذلك ‪ ،‬يتم استخدام ملفات ‪ PDF‬التي تم الحصول عليها في اختبار‬
‫الفرضية المركبة ‪ Neyman-pearson‬لقياس مقاييس األداء (احتمال اإلنذار الخاطئ واكتشاف األخطاء في الشبكة)‪ .‬أثبتت النتائج‬
‫أن مقاييس األداء تتأثر بشكل كبير بموقع الشبكة حيث قدم المستخدم الثانوي وقيمة العتبة المستخدمة في قاعدة القرار‪ .‬كما أن هناك‬
‫حدودا لالختيارات ال يمكن تجاوزها‪.‬‬
‫ً‬

‫الكلمات المفتاحية‪ :‬الراديو المعرفي ‪CR‬؛ دالة الكثافة االحتمالية (‪ )PDF‬؛ هجوم محاكاة المستخدم األساسي (‪.)PUEA‬‬

‫‪*Corresponding Author: Mahmoud Ammar, m.ammar@zu.edu.ly, Department of Computer Engineering, University of Zawia, Zawia, Libya‬‬

‫‪8‬‬