ATALLAH Michael

Physical Layer Security Solutions Against Passive and Colluding
Eavesdroppers in Large Wireless Networks and Impulsive Noise

Environments
by
Michael ATALLAH
MANUSCRIPT-BASED THESIS PRESENTED TO ÉCOLE DE

TECHNOLOGIE SUPÉRIEURE
IN PARTIAL FULFILLMENT FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
Ph.D.
MONTREAL, AUGUST 8, 2019
ÉCOLE DE TECHNOLOGIE SUPÉRIEURE

UNIVERSITÉ DU QUÉBEC
Michael Atallah, 2019

This Creative Commons license allows readers to download this work and share it with others as long as the author is credited. The cont
BOARD OF EXAMINERS
THIS THESIS HAS BEEN EVALUATED
BY THE FOLLOWING BOARD OF EXAMINERS
M. Georges Kaddoum, Thesis Supervisor

Department of Electrical Engineering, École de Technologie Supérieure
M. Julien Gascon-Samson , President of the Board of Examiners

Department of Software Engineering and Information Technologies, École de Technologie
Supérieure
M. Chamseddine Talhi, Member of the jury

Department of Software Engineering and Information Technologies, École de Technologie
Supérieure
M. Yousef R. Shayan, External Independent Examiner

Department of Electrical and Computer Engineering, Concordia University
THIS THESIS WAS PRESENTED AND DEFENDED
IN THE PRESENCE OF A BOARD OF EXAMINERS AND THE PUBLIC
ON JULY 17, 2019
AT ÉCOLE DE TECHNOLOGIE SUPÉRIEURE

ACKNOWLEDGEMENTS
Firstly, I would like to express my sincere gratitude to my advisor professor Dr. Georges
Kaddoum, for the continuous support of my Ph.D study and related research, for his patience,
motivation, and immense knowledge. His guidance helped me in all the time of research and
writing of this thesis.
I thank my labmates for the stimulating discussions, for the sleepless nights we were working
together before deadlines, and for all the fun we have had in the last four years.
Last but not the least, I would like to thank my parents, my brothers and my best friend
Michèlle, for supporting me spiritually throughout writing this thesis and in my life in gen-
eral.
Solutions de Sécurité de Couche Physique Contre les Récepteurs Indistincts Passifs et
Complexes dans les Grands Réseaux sans Fil et les Environnements à Bruit Impulsif
Michael ATALLAH
RÉSUMÉ
Les réseaux sans fil ont connu des évolutions rapides vers la durabilité, l’évolutivité et
l’interop- érabilité. Les sociétés en réseau futures conduisent l’économie numérique à une
commu- nauté plus globale d’infrastructures intelligentes et de services connectés pour une
société plus durable et plus intelligente. En outre, une énorme quantité d’informations
sensibles et confidentielles, telles que les dossiers médicaux, les supports électroniques, les
données financières et les fichiers des clients, est transmise via des canaux sans fil. La mise
en œuvre de la distribution et de la gestion des clés de couche supérieure a été mise au défi
par l’émergence de ces nouveaux systèmes avancés. Afin de résister à divers abus
malveillants et attaques de sécurité, la sécurité de couche physique (PLS) est devenue une
alternative attrayante. Le concept de base derrière PLS est d’exploiter les caractéristiques des
canaux sans fil pour la confidentialité. Son objectif est d’aveugler les oreilles indiscrètes de
sorte qu’ils ne puissent en extraire au- cune informations confidentielles des signaux reçus.
Cette thèse présente des solutions et des analyses pour améliorer le PLS dans les réseaux sans
fil.
Dans le deuxième chapitre, nous examinons les performances de capacité de confidentialité

d’un réseau à double saut amplifier et transférer (AF) pour les techniques de formation de
faisceau réparti (DBF) et de relais opportuniste (OR). Nous calculons la capacité de mise
à l’échelle pour deux grands ensembles; des relais fiables et des relais agressifs peu fiables
coopérant avec un dépisteur visant à intercepter le message. Nous montrons que l’échelle de
capacité dans le DBF est délimitée par une valeur qui dépend du rapport entre le nombre de
relais agressifs dignes de confiance et ceux qui ne le sont pas, alors que la mise à l’échelle de
la capacité de OU est limitée dans le haut par une valeur qui dépend du nombre de relais ainsi
que du rapport signal sur bruit (SNR).
Dans le troisième chapitre, nous proposons une nouvelle technique de multidiffusion par
local- isation, destinée aux grands réseaux AF bi-phase, qui vise à améliorer la sécurité en
présence d’écoutes indiscrètes passives non-collupères. Nous démontrons analytiquement
que la technique proposée augmente la sécurité en réduisant la probabilité de re-choisir un
secteur qui a des oreilles indiscrètes, pour chaque temps de transmission. De plus, nous
montrons également que la capacité de confidentialité de notre technique est la même que
pour la radiodiffusion. Ci-après, les limites inférieure et supérieure de la probabilité de
défaillance du secret sont cal- culées et il est montré que les performances de sécurité sont
remarquablement améliorées par rapport à la technique de multidiffusion classique.
Dans le quatrième chapitre, nous proposons un nouveau protocole de coopération pour les
réseaux de capteurs sans fil à amplification et transmission doubles à phase double, visant à
améliorer la sécurité de la transmission tout en tenant compte des capacités limitées des
nœuds de capteurs. Dans un tel réseau, une partie des K relais peut être de potentiels
oreilles indis-
VIII
crètes passives. Pour réduire l’impact de ces relais non fiables sur la sécurité du réseau, nous
proposons un nouveau protocole de transmission, dans lequel la source accepte de partager
avec la destination une information CSI (Channel State Information) donnée de source
sécurisée lien relais-destination pour encoder le message. Ensuite, la source utilisera à
nouveau cette CSI pour mapper le bon message sur un certain secteur tout en transmettant de
faux messages aux autres secteurs. L’adoption d’un tel protocole de sécurité est prometteuse
en raison de la disponibilité d’un grand nombre de capteurs électroniques bon marché dotés
de capacités de calcul limitées. Pour le schéma proposé, nous avons dérivé la probabilité de
coupure du secret (SOP) et démontré que la probabilité de recevoir les informations codées à
droite par un relais peu fiable sont inversement proportionnelles au nombre de secteurs. Nous
montrons également que le comportement agressif des relais non fiables coopérants n’est pas
efficace par rapport au cas où chaque relais non sécurisé tente d’intercepter individuellement
le message transmis.
Enfin, nous examinons les performances de sécurité de la couche physique sur les canaux à
évanouissements de Rayleigh en présence de bruit impulsif, telles que rencontrées par
exemple dans les environnements de électrique intelligent. Pour ce schéma, les métriques de
performance de confidentialité ont été prises en compte avec et sans brouillage assisté par
destination du côté de l’espionneur. D’après les résultats obtenus, il est vérifié que la POS,
sans brouillage assisté par destination, est un revêtement de sol avec un rapport signal sur
bruit élevé valeurs et qu’il peut être considérablement amélioré avec l’utilisation du
brouillage.
Mots-clés: Couche physique, brouillage, transmission sectorielle, bruit impulsif.

Physical Layer Security Solutions Against Passive and Colluding Eavesdroppers in
Large Wireless Networks and Impulsive Noise Environments
Michael ATALLAH
ABSTRACT
Wireless networks have experienced rapid evolutions toward sustainability, scalability and
in- teroperability. The digital economy is driven by future networked societies to a more
holistic community of intelligent infrastructures and connected services for a more
sustainable and smarter society. Furthermore, an enormous amount of sensitive and
confidential information, e.g., medical records, electronic media, financial data, and customer
files, is transmitted via wireless channels. The implementation of higher layer key distribution
and management was challenged by the emergence of these new advanced systems. In order
to resist various malicious abuses and security attacks, physical layer security (PLS) has
become an appealing alternative. The basic concept behind PLS is to exploit the
characteristics of wireless channels for the confidentiality. Its target is to blind the
eavesdroppers such that they cannot extract any confidential information from the received
signals. This thesis presents solutions and analyses to improve the PLS in wireless networks.
In the second chapter, we investigate the secrecy capacity performance of an amplify-and-

forward (AF) dual-hop network for both distributed beamforming (DBF) and opportunistic
relaying (OR) techniques. We derive the capacity scaling for two large sets; trustworthy
relays and untrustworthy aggressive relays cooperating together with a wire-tapper aiming to
intercept the message. We show that the capacity scaling in the DBF is lower bounded by a
value which depends on the ratio between the number of the trustworthy and the
untrustworthy aggressive relays, whereas the capacity scaling of OR is upper bounded by a
value depending on the number of relays as well as the signal to noise ratio (SNR).
In the third chapter, we propose a new location-based multicasting technique, for dual phase
AF large networks, aiming to improve the security in the presence of non-colluding passive
eavesdroppers. We analytically demonstrate that the proposed technique increases the
security by decreasing the probability of re-choosing a sector that has eavesdroppers, for each
transmission time. Moreover, we also show that the secrecy capacity scaling of our
technique is the same as for broadcasting. Hereafter, the lower and upper bounds of the
secrecy outage probability are calculated, and it is shown that the security performance is
remarkably enhanced, compared to the conventional multicasting technique.
In the fourth chapter, we propose a new cooperative protocol, for dual phase amplify-and-
forward large wireless sensor networks, aiming to improve the transmission security while
taking into account the limited capabilities of the sensor nodes. In such a network, a portion
of the K relays can be potential passive eavesdroppers. To reduce the impact of these
untrustworthy relays on the network security, we propose a new transmission protocol,
where the source agrees to share with the destination a given channel state information (CSI)
of source-trusted relay-destination link to encode the message. Then, the source will use this
CSI again to map
X
the right message to a certain sector while transmitting fake messages to the other sectors.
Adopting such a security protocol is promising because of the availability of a high number
of cheap electronic sensors with limited computational capabilities. For the proposed scheme,
we derived the secrecy outage probability (SOP) and demonstrated that the probability of re-
ceiving the right encoded information by an untrustworthy relay is inversely proportional to
the number of sectors. We also show that the aggressive behavior of cooperating untrusted
relays is not effective compared to the case where each untrusted relay is trying to intercept
the transmitted message individually.
Fifth and last, we investigate the physical layer security performance over Rayleigh fading
channels in the presence of impulsive noise, as encountered, for instance, in smart grid envi-
ronments. For this scheme, secrecy performance metrics were considered with and without
destination assisted jamming at the eavesdropper’s side. From the obtained results, it is veri-
fied that the SOP, without destination assisted jamming, is flooring at high signal-to-noise-
ratio values and that it can be significantly improved with the use of jamming.
Keywords: Physical Layer, Jamming, Sectoral Transmission, Impulsive Noise.

TABLE OF CONTENTS
Page
INTRODUCTION.....................................................................................................................1
CHAPTER 1 LITERATURE REVIEW................................................................................5

1.1 Physical Layer Security Concept...................................................................................5
1.2 Physical Layer Security Techniques..............................................................................6
1.2.1 Artificial Noise and Artificial Fading.............................................................7
1.2.2 Spoofing........................................................................................................10
1.2.3 Multi Antenna and Beamforming Based Techniques...................................11
1.2.4 Relay and Cooperative Methods...................................................................14
1.2.4.1 Cooperative Jamming..................................................................18
1.2.4.2 Artificial Jamming Signals Types................................................19
1.2.4.3 Jamming Policies.........................................................................20
1.2.4.4 Cooperative Jamming with Power Allocation.............................22
1.2.5 Game Theory for Security............................................................................23
1.2.6 Key Generation Technique...........................................................................25
1.3 Unrealistic assumptions...............................................................................................26
CHAPTER 2 SECRECY CAPACITY SCALING WITH

UNTRUSTWORTHY AGGRESSIVE RELAYS
COOPERATING WITH A WIRE-
TAPPER........................................................................................................29
2.1 Abstract........................................................................................................................29
2.2 Introduction.................................................................................................................29
2.3 System Model..............................................................................................................31
2.3.1 Distributed Beamforming.............................................................................33
2.3.2 Opportunistic Relaying.................................................................................34
2.4 SCALING LAW OF SECRECY CAPACITY............................................................35
2.4.1 Scaling Law of Distributed Beamforming....................................................35
2.4.2 Scaling Law of Opportunistic Relaying........................................................38
2.5 Conclusions.................................................................................................................39
CHAPTER 3 SECRECY ANALYSIS IN WIRELESS NETWORK WITH

PASSIVE EAVESDROPPERS BY USING PARTIAL
COOPERATION
................................................................................. 41
3.1 Abstract........................................................................................................................41
3.2 Introduction.................................................................................................................42
3.3 System Model and Problem Formulation....................................................................43
3.4 Lower and Upper Bounds of Secrecy Outage Probability..........................................47
3.5 Scaling Law of Secrecy Capacity................................................................................51
3.6 Simulation Results.......................................................................................................53
3.7 Conclusions.................................................................................................................56
XII
CHAPTER 4 DESIGN AND PERFORMANCE ANALYSIS OF SECURE

MULTICASTING COOPERATIVE PROTOCOL FOR WIRELESS
SENSOR NETWORK APPLICATIONS.....................................................57
4.1 Abstract........................................................................................................................57
4.2 Introduction.................................................................................................................58
4.3.1 Non Colluding Eavesdropping Relays..........................................................62
4.3.2 Colluding Eavesdropping Relays.................................................................63
4.4 Secrecy Outage Probability.........................................................................................64
4.6 Conclusions.................................................................................................................68
CHAPTER 5 SECRECY ANALYSIS OF WIRELESS SENSOR NETWORK

IN SMART GRID WITH DESTINATION ASSISTED
JAMMING
................................................................................. 69
5.1 Abstract........................................................................................................................69
5.2 Introduction.................................................................................................................69
5.3 Related Work...............................................................................................................70
5.5 Secrecy Outage Probability Analysis..........................................................................75
5.5.1 Secrecy Outage Probability Analysis with Jamming....................................75
5.5.2 Secrecy Outage Probability Analysis without Jamming...............................78
5.7 Conclusions.................................................................................................................81
CONCLUSION AND RECOMMENDATIONS.....................................................................83
BIBLIOGRAPHY..................................................................................................................118
LIST OF FIGURES
Page
Figure 1.1 Wireless wiretap system model........................................................................6
Figure 1.2 Normalized average secrecy capacity versus γ m , for chosen values
of γ w , in Rayleigh and Gaussian wiretap channels..........................................7
Figure 1.3 The spoofer is transmitting a deceiving signal to a legitimate

receiver...........................................................................................................10
Figure 1.4 Representation of a general MIMO wiretap channel......................................12
Figure 1.5 Representation of trusted (distinct relay and eavesdropper) relay

network...........................................................................................................15
Figure 1.6 Representation of untrusted (co-located relay and eavesdropper)

relay network..................................................................................................16
Figure 1.7 Eavesdroppers’ passive behavior....................................................................17
Figure 1.8 Eavesdroppers’ colluding behavior................................................................18
Figure 1.9 Representation of a network with a jammer...................................................19

Figure 2.1 System model.................................................................................................32
Figure 2.2 Ergodic secrecy capacity: ρ Δρs = ρt = ρu = ρd = 5 dB, σ 2
=
= 1
2
σ 2 = 1, and U = T...........................................................................................40
Figure 3.1 System model
clustered consisting
in G sectors, of a multi-antennas
a destination source s, T relays
d and an eavesdropper e. In
this figure, T = 9, K = 3 and G = 3..................................................................44

the proposed system
3.2 with jamming: hs,k and= simulated
Analytical hk,d = SOP 1, R = Figure
performances of . .2 . lower .2 bound
1bps/Hz, and K = 5.........................................................................................53
proposed system3.3
and OR with jamming:
Analyticalhand = hk,d SOP
s,k simulated = 1, and Figure
performances of the . .2upper . bound.2
R = 1bps/Hz....................................................................................................54
2 2
=Δ = ρ = ρ = 10 dB.............................................................................55
Figure 3.4 ρ Simulated secrecy capacity scaling: .hk,d . = .hs,k. = 1, and
ρs d k
XIV
Figure 4.1 In the 1st hop of each transmission, s multicasts the useful message
xtr and the fake ones xi ×t=r ’s towards N sectors. In the 2nd hop, the
K relays retransmit their received messages towards d..................................60
Figure 4.2 SOP with passive untrusted relays: R = 3 bps/Hz, M = 4, σs =
σk = 0.95 and μs = μk = 1................................................................................66

Figure 4.3 SOP with aggressive untrusted relays: R = 2 bps/Hz, M = 4,
σs = σk = 1.1 and μs = μk = 0.69.....................................................................67

Figure 5.1 The source s transmits xs to the destination d, while the
eavesdropper e is trying to intercept xs. In the case where d is
jamming, d is provided with two independent antennas; (1) is for
receiving xs. (2) is for jamming with artificial noise signal xd........................72
2
system without jamming: γ = γ , Γ = Γ = 1000, h = Figure
5.2 Analytical and simulated SOP performances . . of
the proposed d0 e0 d e s,d
2 . 2.
|hs,e| = hd,e =
2
1, and R = 1bps/Hz...............................................................79
2
2
system with jamming: hs,d = |hs,e| = hd,e = 1, Γd = Γe = Figure 5.3
Analytical
. . and simulated .SOP. performances of
the proposed
2
100, γd0 = γe0, γ j = 1 γd0, and R = 1bps/Hz.................................................80
LIST OF ABREVIATIONS
AWGN Additive White Gaussian Noise
AF Amplify-and-Forward
CB Cooperative Beamforming
CDF Cumulative Distribution Function
CJ Cooperative Jamming
CRN Cognitive Radio Network
CSI Channel State Information
D2D Device-to-Device
DBF Distributed Beamforming
ESR Ergodic Secrecy Rate
FDJ Full Duplex Jammer
HDJ Half Duplex Jammer
IJ Intended Jamming
IoBT Internet-of-Battlefield-Things
MC-DCSK Multi-Carrier Differential Chaos Shift Keying
MIMO Multiple-Input Multiple-Output
MISO Multiple-Input Single-Output
NOMA Non-Orthogonal Multiple Access
OR Opportunistic Relaying
XVI
PDF Probability Density Function
PLS Physical Layer Security
PU Primary User
RSS Received Signal Strength
SINR Signal-to-Interference-plus-Noise Ratio
SNR Signal-to-Noise Ratio
SOP Secrecy Outage Probability
SPCA Sequential Parametric Convex Approximation
SU Secondary User
INTRODUCTION
Wireless communication is an integral part of our lives; it also has significant social repercus-
sions. Privacy and confidentiality with respect to the transmitted information over the
wireless medium is vital, especially for applications concerning medical information, e-
banking, and e-commerce. However, wireless communications are often vulnerable to
eavesdropping and signal interception Mavoungou et al. (2016); Neshenko et al. (2019);
Hong et al. (2013). Many security requirements are considered in the design of wireless
networks, like integrity, confidentiality checks, authentication and spectrum access control
Lou & Ren (2009); Shiu et al. (2011). Integrity ensures that the information that has been
transmitted is utilized and modified by the legitimate user. Confidentiality refers to the
prevention of unauthorized information disclosure. Authentication refers to the individuality
of different terminals’ confirmations. Spec- trum access control refers to the prevention of
denial-of-service type attacks. Usually, these security tasks are mostly undertaken in the
protocol stack of the network’s upper layers with the usage of cryptographic encryption and
decryption methods. When employing symmetric- key cryptosystems, the two users have to
share a common private key that is encrypting and decrypting the private message Hong et
al. (2013). However, this requires a secure channel or protocol for the secret keys sharing.
The secret key management and distribution has its own difficulties Schneier (1998); they
lead to security vulnerabilities in wireless systems. As a substitute, the cryptosystems of the
public key allow the use of two different keys; a public one for encryption and a separate
private one for decryption. The first one is also available to all users since the private key is
only known by the receiver. Therefore, cryptographic methods rely on the hardness of the
computation to decrypt the message to achieve security when there is no availability of the
secret key. As the computation power increases, e.g., with the development of quantum
computers, the computational hardness of some mathematical problems, which is the basis of
the decryption and encryption, will not hold, resulting in many cryptosystems’ break down
Hong et al. (2013). Moreover, in future networks, more devices will
2
be connected to nodes with different power and computational capabilities. Furthermore, due
to the decentralized nature of the networks, devices join or leave the network in random time
instants, which renders the management and distribution of cryptographic keys a challenging
task. Therefore, many signal and coding processing techniques have been developed in the
physical layer to enhance and to support security in wireless systems. Many contributions
have been made to find alternative security solutions to fit the requirements of current and
emerging wireless networks Goel & Negi (2008); Gopala et al. (2008); Shannon (1949);
Bloch & Bar- ros (2011). Therefore, the security of the physical layer can facilitate the
cryptographic keys’ distribution to enhance the security. Even though the fast variations of
the channel and the broadcast nature of the wireless medium may cause additional
challenges to their design, the physical layer security techniques also exploit the wireless
transmissions’ properties to better protect the communication channel Hong et al. (2013).
Contributions and Outline
The first chapter is the literature review that browses briefly the applied security techniques
in the physical layer. The contributions of our thesis are summerized as follows:
In Chapter 2, the secrecy capacity scaling was investigated in the presence of untrustworthy
aggressive relays that are cooperating between each other to intercept the message. More-
over, destination assisted jamming was applied. Two transmission strategies were studied:
opportunistic relaying and distributed beamforming techniques. The secrecy scaling bounds
were calculated for both DBF and OR. For DBF, it is shown that its secrecy scaling is lower
bounded by a value related to the number of the trustworthy and the untrustworthy relays in
the network, and that intended jamming, when applied, remarkably enhances the security.
More- over, the DBF showed better security performance compared to OR, which gives DBF
the priority to be applied in large wireless networks when the security is demanded.
3
In Chapter 3, to reduce the probability that an eavesdropper would have a continuous access
to the transmitted message, the legitimate transmitter decided to change its transmission from
broadcasting to a location-based multicasting technique, in the presence of destination as-
sisted jamming. In this way, if it is not in the covered sector, the eavesdropper cannot access
the transmitted message. The secrecy capacity scaling was calculated and showed that this
location-based multicasting technique scales similar to the broadcasting one. Moreover, ana-
lytical expressions of the lower and upper bounds of the secrecy outage probability were also
provided. The proposed protocol was shown to be secure and confusing to the eavesdropper
since the later cannot have access to the transmitted information all the time.
Chapter 4 proposes a novel protocol that implements the location-based multicasting

protocol, to transmit the useful information in one sector and fake information towards the
other sectors. The main advantage of this technique is its immunity towards the presence of
aggressive relays when they plan to cooperate between each other to intercept the message.
The results showed that this aggressive cooperation by the eavesdroppers will hardly increase
the amount of the stolen information. Also, it is proved that by increasing the number of
multicasted sectors, the security performance is enhanced. No jamming was applied in this
scenario; however, the performance of the proposed protocol overcomes the secrecy
performance of the conventional jamming technique.
In Chapter 5, new secrecy capacity expressions in the presence of impulsive noise and des-
tination assisted jamming are proposed. This new alternative approach in reformulating the
secrecy capacity expressions allows the other researchers to analyse their proposed system
models easily in the presence of impulsive noise. Analytical expressions for the secrecy
outage probability, with and without jamming, were provided. From the obtained results, it
was shown that the SOP without destination assisted jamming is flooring at high SNR values,
and that it could be enhanced remarkably by adding destination assisted jamming techniques.
4
Author’s Publications
Here, we list the published and submitted journals and conference papers denoted by J and C
respectively.
– J1: M. Atallah and G. Kaddoum, "Secrecy Analysis in Wireless Network with Passive
Eavesdroppers by Using Partial Cooperation," in IEEE Transactions on
Vehicu- lar Technology, Apr. 2019. DOI: 10.1109/TVT.2019.2913934
– J2: M. Atallah and G. Kaddoum, "Secrecy Capacity Scaling With Untrustworthy Aggress-
ive Relays Cooperating With a Wire-Tapper," in IEEE Wireless Communications
Letters, vol. 5, no. 4, pp. 376-379, Aug. 2016. DOI:
10.1109/LWC.2016.2561285
– J3: M. Atallah and G. Kaddoum, "Design and Performance Analysis of Secure Multicas-
ting Cooperative Protocol for Wireless Sensor Network Applications,"
submitted to IEEE Wireless Communications Letters March. 2019
– J4: M. Atallah; M. S. Alam; and G. Kaddoum: ’Secrecy Analysis of Wireless Sensor Net-
work in Smart Grid with Destination Assisted Jamming’, IET
Communications, 2019, DOI: 10.1049/iet-com.2018.5344 IET Digital Library,
https://urnals/ digital-library.theiet.org/content/jo 10.1049/iet-com.2018.5344
– J5: G. Kaddoum, H. Tran, L. Kong and M. Atallah, "Design of Simultaneous Wireless

Information and Power Transfer Scheme for Short Reference DCSK
Communi- cation Systems," in IEEE Transactions on Communications, vol.
65, no. 1, pp. 431-443, Jan. 2017. DOI: 10.1109/TCOMM.2016.2619707
– C1: M. Atallah and G. Kaddoum, "Secrecy Analysis of Cooperative Network with Un-
trustworthy Relays Using Location-Based Multicasting Technique," 2017 5th
International Conference on Future Internet of Things and Cloud Workshops
(FiCloudW), Prague, 2017, pp. 206-210. DOI: 10.1109/FiCloudW.2017.74
– C2: M. Atallah, G. Kaddoum and L. Kong, "A Survey on Cooperative Jamming Applied
to Physical Layer Security," 2015 IEEE International Conference on
Ubiquitous Wireless Broadband (ICUWB), Montreal, QC, 2015, pp. 1-5. DOI:
10.1109/ICU WB.2015.7324413
CHAPTER 1
LITERATURE REVIEW
1.1 Physical Layer Security Concept
As shown in Fig. 1.1, a generic wireless communication network model consisting of three
nodes, namely a legitimate transmitter (Alice), an intended receiver (Bob) and an
eavesdropper (Eve), is taken into consideration. We call the link between Alice and Bob the
main channel, while the link between Alice and Eve is called the wiretap channel. This
model exemplifies the specific features of most multi-user secure communication systems.
The vital concept of the secrecy capacity relies on goal of maximizing the legitimate channel
capacity or minimizing the capacity of the illegitimate channels, which is attainable via the
usage of the dynamic nature of the wireless channels, otherwise it is equal to zero Gopala et
al. (2008). In Bloch et al. (2008), the secrecy capacity over an additive white Gaussian
noise (AWGN) channel Cs,A and Rayleigh fading channel Cs,R are respectively given by
Σ . Σ
1 P
Cs,A = , (1.1)
1+
m2
. 1 + σw2 ΣΣ+
2 1 P
σ −2 +
Σlog2 Σ log2 P|hw|2 ΣΣ
P|hm|2 σw2 , (1.2)
−
Cs,R = .1 + σm2 log2 .1 +
where [x]+ = max {0, x}, P represents the transmitted power, σm 2 and σ
w 2 are the noise power
of
the main channel and wiretap channel, respectively. Moreover, hm and hw are the instantaneous
channel coefficients of the main channel and wiretap channel, respectively. Also, the received
signal-to-noise ratios (SNRs) at Bob and Eve are defined as γm
andP|hm|2 = P|hw|2 , re-
= σm σw
γw
spectively. To achieve security, our aim is to keep the secrecy capacity Cs strictly positive, i.e.
Cs > 0. In Fig. 1.2, the average secrecy capacity of a Rayleigh fading channel (1.2) is
compared with that of a Gaussian wiretap channel (1.1). Strikingly, one can observe that the
secrecy capacity over Rayleigh fading channels is higher than over AWGN channels. In
other words, we can use the fading property of the physical layer to decrease the SNR of the
wiretap channel.
6
Figure 1.1 Wireless wiretap system model
Besides using the fading characteristics of the wireless channel, many other methods to im-
prove the secrecy performance of the wireless communication systems have been suggested.
In Shiu et al. (2011), physical layer security methods are classified into five major
approaches: multiple-input-multiple-output (MIMO) channel, theoretical secrecy capacity,
coding schemes (channel coding and network coding), power allocation, and signal design
(artificial noise). Additionally, cooperative relay Han et al. (2015); Wang et al. (2013a); Chen
et al. (2013), cooperative jamming Atallah et al. (2015); Ibrahim et al. (2015); Jameel et al.
(2018), inter- leaving and spreading in frequency and time to secure Multi-Carrier
Differential Chaos Shift Keying (MC-DCSK) Kaddoum et al. (2012) and energy harvesting
Xing et al. (2014) are other useful methods. In the following section, we will describe the
widely used methods in physical layer security.
1.2 Physical Layer Security Techniques
In this section, we will explore the most commonly used techniques to enhance the security
in the physical layer.
0.9
Rayleigh, Av.w = 0dB
0.8Rayleigh, Av.w = 10dB Rayleigh, Av.w = 20dB
0.7 Gaussian, Av. = 0dB

w
Gaussian, Av.w = 10dB
0.6 Gaussian, Av. = 20dB
s
Av. secrecy capacity C
0.5
0.4
0.3
0.2
0
−10 −5 0 5 10 15 20 25 30
0.1 Av.m (dB)
Figure 1.2 Normalized average secrecy capacity versus γ m ,

for chosen values of γ w , in Rayleigh and Gaussian wiretap
channels
1.2.1 Artificial Noise and Artificial Fading
A. Artificial Noise
In multi-antenna systems, Artificial noise is one of the most popular techniques to guarantee
security at the physical layer Goel & Negi (2008); Khisti & Wornell (2010). The basic idea
behind artificial noise technique is that the channel state information (CSI) of the main
channel is unknown by the eavesdroppers. Thus, they will be distracted and unable to
decode the
transmitted information-bearing symbols. In Fig. 1.1, the source sends its signal
x = u + v, (1.3)
where u issignal
Then, the the message and
received byvthe
is the added artificial
legitimate receivernoise.
Bob isv is chosen such that hm v = 0.
yb = hm x + nb = hm(u + v)+ nb (1.4)
yb = hm u + nb,
whereas the signal received by the eavesdropper Eve is
ye = hw x + ne = hw u + hw v + ne (1.5)
ye = hw u + hw v + ne,
hence, the secrecy capacity is obtained as
Σ . u m .
P |h |2 Pu |hw|
Cs = Σlog2 1 + 2 − 2 1+
b 2 Pv |hw|2 + σ ΣΣ+ , (1.6)
σ log2 e
where Pu and Pv are the transmitted power of u and v respectively, σ and σ 2 are the noise
2
b e
power at the legitimate receiver and the eavesdropper, respectively. hm and hw are the channel
coefficients of the main and wiretap links, respectively. We can see from (1.6) that the
secrecy capacity is improved by adding the artificial noise compared to (1.2). The authors in
Lin et al. (2013b) proposed a generalized scheme for injecting artificial noise to a legitimate
channel. Their scheme was shown to be efficient under various channel conditions. Their
simulation results showed that their algorithm outperforms other previous algorithms in
enhancing the secrecy capacity. In Zhang et al. (2016b), an efficient algorithm was proposed
to study the optimal resource allocation for maximizing the weighted sum secrecy rate with
a new frequency domain artificial noise aided transmission strategy. Also, in Zeng et al.
(2019), the authors proposed a strategy to secure the confidential information of massive
MIMO-NOMA networks,
where the base station, based on the estimated CSI, precodes the confidential information and
injects artificial noise.
B. Artificial Fading
Different from the artificial noise, the main idea of artificial fading is to weight the
transmitted information symbol s randomly by a weighting coefficient k. The transmitted
signal could be written as
x = ks, (1.7)
with a constraint that h mk = 1, where hm is the channel coefficient between the transmitter and
the legitimate receiver. Therefore, the received signal at the legitimate receiver becomes
ym = hmks + nm = s + nm, (1.8)
where nm is the AWGN at the receiver. Therefore, the receiver will be able to decode its
received signal directly without any channel coefficient hm, whereas the signal received at the
eavesdropper becomes
yw = hwks + nw, (1.9)

where hw is the channel coefficient between the transmitter and the eavesdropper, and nw is the
AWGN at the eavesdropper. The authors in Wang et al. (2015c) compared between artificial
noise and artificial time-varying multiplicative noise that they named it artificial fast fading
scheme since this scheme results in an equivalent fast fading channel for the eavesdropper.
They concluded that the artificial noise scheme achieves a larger secrecy rate when the trans-
mitter has more antennas than the eavesdropper. Otherwise the artificial fast fading is
superior. Motivated by their results, they proposed a hybrid artificial noise-artificial fast
fading scheme to achieve a better secrecy performance than either schemes. As mentioned in
Wang & Yang (2012), the unwanted wireless communication links can deliberately be
corrupted by double beam switching of the smart antenna array as a novel concept of artificial
fading. In Wang et al. (2014a), artificial fast fading was applied by randomly weighting
the information sym-
bols at different transmitting antennas in a special way so that the eavesdropper’s channel is
a fast fading channel while that of the intended receiver is an additive white Gaussian noise
channel. In Song (2018), the researchers proposed a novel cross layer design by combining
artificial fast fading with secret-keys in the upper layer crypto-system to nullify the
information leakage for any number of antennas at the eavesdropper. For both artificial noise
and artificial fading, it is assumed that the transmitter and the receiver are acquainted of the
main channel. Subsequently, the legitimate channel’s security performance becomes easily
assailable by the eavesdropper. Moreover, the mobility of the legitimate nodes, with the
artificial noise and artificial fading techniques, which adds complexity due to the rapid
changes in the characteristics of the legitimate channels, has not yet been investigated.
1.2.2 Spoofing
Figure 1.3 The spoofer is transmitting a deceiving signal to

a legitimate receiver
In wireless networks, a spoofing attack, depicted in Fig.1.3, is a situation in which a node

transmits deceiving signals to a legitimate receiver by acting as if it is a legitimate
transmitter. Spoofing attacks studies have investigated the detection of the spoofers’ location,
which can be done by measuring the received signal strength (RSS) transmitted by the
attacker. Mathemati-
cally, RSS is given
by
RSS(dB)= Ptx +ρ −PL, (1.10)
where Ptx is the transmitted power, PL is the path loss and ρ is the gain of the transmitting
antenna. To locate the spoofer, many receivers should work collaboratively to measure the
RSSs Wang & Yang (2013). Another way to estimate the spoofer’s location is by measuring
the angle of the arrived signal Chen et al. (2012), this method is more accurate than the
former one Wang & Yang (2013). In Liu et al. (2019), the authors proposed a new pilot
spoofing attack detection scheme by employing another node as a trusted user, which also
cooperates in the uplink training process and helps to detect pilot spoofing in multiple-input
single-output (MISO) systems. Also, to prevent spoofing attacks, finger prints or link
signatures are used as useful methods Zhang et al. (2008). Link signature is extracted from
the channel impulse response as a function of time delay and the magnitude of the impulse
response. By applying the latter, the legitimate nodes are able to distinguish each other since
the channels between them are identified and are well known between them. Therefore, being
in a different location, a spoofer could be easily detected from the link signature of its
channel. The weakness of the link signature method lies in the complexity of the key
signature calculation at the legitimate nodes when they are changing their locations.
Essentially, the available work in this particular field has mainly focused on locating the
spoofers. Therefore, to establish a protocol on how to protect the legitimate nodes or even
attack the spoofers, further work needs to be done. Very few works studied the mobility of
the legitimate nodes in the context of spoofing attacks. Consequently, an investigation is
needed in this field to provide a certain level of security when nodes are moving or for fast
varying channel conditions.
1.2.3 Multi Antenna and Beamforming Based Techniques
To enhance the security in the physical layer, multiple antenna techniques are widely ap-
plied Yang et al. (2013); Zhang et al. (2015); Li et al. (2014a); Wang et al. (2014b); Ba-
nawan & Ulukus (2014); Xing et al. (2014); Vishwakarma & Chockalingam (2014); Og-
gier & Hassibi (2011). Fig 1.4 shows a general MIMO wiretap channel, where the source,
Figure 1.4 Representation of a general MIMO wiretap channel
transmits an n×1 complex transmitted signal X with covariance matrix Kx = E{ XX H }, satis-

destination, and wiretapper are equipped with n, nm and nw antennas, respectively. The source
fying the power constraint Tr(Kx) ≤ P. Therefore, the signals received by the destination and
the wiretapper are
Ym = Hm X + Nm, (1.11)
and
Yw = Hw X + Nw, (1.12)
respectively, where Nm and Nw are respectively nm ×1 and nw ×1 complex white Gaussian

additive noise vectors. Therefore, the secrecy capacity of MIMO wiretap channels is given in
Oggier & Hassibi (2011) by
CS = max logdet(I + Hm Kx HmH ) − logdet(I + Hw Kx HwH ) (1.13)
Kx≥0,Tr(Kx)=P
where Hm and Hw are respectively nm ×n and nw ×n fixed channel matrices. Moreover, (.)H
is the Hermitian and I denotes the identity matrix. It is clear from (1.13) that the secrecy
capacity is enhanced by increasing the numbers of antennas at the destination. The secrecy
performance of MIMO wiretap channels was analysed in Kong et al. (2016) and Kong et al.
(2018a). The authors in Yang et al. (2013) assumed a scenario where multiple legitimate
users
are receiving multiple independent data streams from a base station; during the transmission,
many eavesdroppers with multiple antennas are interested in the transmission stream of the
base station. Colluding or not, the eavesdroppers may also use receiving beamforming
method to maximize the signal-to-interference-plus-noise ratio (SINR) of the streams they
are wire- tapping. To guarantee a confidential transmission between the legitimate users and
the base station, the cooperative jammer will work on keeping the SINR at the eavesdroppers
below a certain threshold level. Another scenario in Li et al. (2014a) studied the Gaussian
wiretap channel’s secrecy capacity aided by an external jammer. While the jammer and the
eavesdropper are equipped with multiple antennas, each of the receiver and the transmitter
have a single one. The authors in Wang et al. (2014b) revealed a scenario for secure
transmission within a two-hop amplify-and-forward relay network scheme, such that for a
large number of antennas, the ergodic secrecy capacity (ESC) is independent of the number
of antennas at the source and dependent on the number of antennas at the destination. In
Kong et al. (2018b), the analysis of the secrecy performance in MIMO wireless networks was
provided for two schemes: the nearest user and the best user based on its SNR. The
researchers in Tran et al. (2019) proposed two transmit antenna selection solutions in MIMO
NOMA networks. Their study showed that increasing the number of antennas at the
legitimate nodes only has an impact on low and medium range of transmitted SNR values.
Beamforming, a technique used to direct the signal transmission or reception, is also an
efficient method and it is applied in many works with the cooperative jamming technique
Wang et al. (2013a); Tran & Kong (2014); Wang et al. (2013b); Han et al. (2015);
Vishwakarma & Chockalingam (2014). In Wang et al. (2013a), a scheme with joint
cooperative jamming and beamforming was proposed to raise the security level of a
cooperative relay network, where part of the nodes use a distributed beamforming
mechanism while the others are simultaneously jamming the eavesdropper. In Tran & Kong
(2014), another beamforming scheme was proposed; by preventing the eavesdroppers from
using the beamformers to suppress the jamming signals. It also uses two orthogonal
dimensions for transmitting and receiving signals. Moreover, a hybrid cooperative jamming
and beamforming scheme was proposed in Wang et al. (2013b) also; the idea behind this
work is that in both hops of a cooperative transmission, some intermediate nodes relay the
signals to the legit-
imate receivers by adopting the beamforming distribution, while the other nodes are jamming
the eavesdropper, which eventually leads in protection of the transmitted data. The authors
in Han et al. (2015) developed an optimal relay assignment algorithm to solve a problem to
maximize secrecy capacity, and an algorithm on smart jamming was also proposed to
increase the system’s secrecy capacity. In Alsaba et al. (2019), the authors studied a null-
steering beamforming technique to enhance the security in NOMA systems by injecting a
jamming signal and directing it toward the malicious node while being suppressed in the
direction of the legitimate users. Also, the researchers in Akhlaghpasand et al. (2019)
proposed a framework to protect the uplink transmission from jamming attacks in massive
MIMO systems by sup- pressing the jamming interference during the detection of the useful
information sent by the legitimate users. However, one of the challenges in beamforming is
when the nodes are moving, which makes it difficult to track and direct the beams towards
these nodes, besides that when working in high frequencies, these beams could be easily
blocked, even by hand.
1.2.4 Relay and Cooperative Methods
In the context of relay networks, we can divide the security issues in two broad categories,
namely trusted relays and untrusted relays.
A. Trusted Relays:
The eavesdropper and the relays are two separated network entities. Fig. 1.5 shows one of
the most frequently used relay-based wiretap scenarios. To counteract external
eavesdroppers, the relays can play various roles. They may be acting as traditional relays or
both jamming partners as well as relaying components in order to strengthen the secure
transmission. The concept of trusted relays was used in Arafa et al. (2018) to secure
downlink NOMA systems. In Atallah & Kaddoum (2019), the source and the destination
share the CSI of source-trusted relay-destination link to encode the messages and to map the
transmission. The authors in Dahmane et al. (2017) introduced a weighted probabilistic and
trust-aware strategy to provide high security and integrity level with less relays. In addition,
in Waqas et al. (2018), secret key
Figure 1.5 Representation of trusted (distinct relay and
eavesdropper) relay network
generation was investigated in D2D communications in the presence of trusted and untrusted
relays. Another novel noise-forwarding strategy called deaf helper phenomenon was also
proposed in Lai & El Gamal (2008); to confuse the eavesdropper, dummy codewords
independent of the secret message are sent by the full-duplex relays. This strategy was also
investigated in Bassily & Ulukus (2013, 2012). In Li et al. (2013), a security-oriented
transmission scheme was proposed in cognitive radio network CRN with the aid of multiple
relays. To maximize the secondary user (SU) link secrecy capacity, both cooperative
jamming techniques and beamforming are used to improve the performance of the SU while
providing a good protection to the primary users (PUs). The proposed scheme contributes in
securing the SU’s transmission while the SNR attenuation at the PU receiver is kept
acceptable. Another interesting security scheme, in a centralized cognitive radio network
(CRN), was proposed in Wen et al. (2019), where the base station is communicating with a
PU in the presence of an eavesdropper, while the SU acts as a friendly jammer. This jammer
could be fully trusted or untrusted when it does not send jamming signals all the time for
selfish reasons. Therefore, a selection criterion was adopted to evaluate the trust degree of
this jammer and its effect on the secrecy performance.
B. Untrusted Relays:
Unlike the aforementioned case, the relay itself is sometimes considered an untrusted user; it
Figure 1.6 Representation of untrusted (co-located relay and
eavesdropper) relay network
acts both as an eavesdropper and a traditional relay, i.e., as shown in Fig. 1.6 the relay node
and the eavesdropper are co-locating. First studied in Oohama (2007) for general relay
channels, this type of model implies that the source desires to use the relay to communicate
with the destination while it intends to shield the relay of the message. Under the assumption
that some of the messages that have been transmitted are confidential to the relay, coding
problems asso- ciated with the relay-wiretap channel are studied. In Shrestha et al. (2019),
the authors studied the secrecy performance of a multi-hop ad-hoc wireless network in the
presence of untrusted and trusted relays in each hop. To perform the transmission, the most
secure relay will be chosen in each hop to deliver the message. The researchers in El Shafie
et al. (2017) proposed a new scheme to secure a wireless network in the presence of
untrusted relays. The destination and another cooperative jammer inject artificial noise to
jam these relays for two reasons; to prevent them from intercepting their received messages
and to help them harvest energy to charge their batteries. In Atallah & Kaddoum (2017) and
Atallah & Kaddoum (2019), new location-based multicasting techniques were proposed to
reduce the possibility of an untrustworthy relay intercepting the whole transmitted message.
As an interesting result, the use of an untrustworthy relay can still be beneficial in increasing
the secrecy capacity Yener & He (2010); Jeong et al. (2012); Sun et al. (2012); Kuhestani et
al. (2016). In the following section, we will explore a very important cooperating method in
physical layer security; the cooperative jamming technique, which is a promising method and
has attracted significant attention. This method was proposed originally for a multiple access
wiretap channel, where multiple legiti-
mate users wish to establish secure communications with an intended receiver in the presence
of an eavesdropper. Since an eavesdropper could be part of a wireless network as an
untrusted relay, we will explore here some major types of the eavesdroppers’ behaviors.
Active behavior:
Here, as an example, the eavesdropper could attack the wireless system by sending jamming
signals. In this case, it is possible to locate this active eavesdropper and change the current
strategies to avoid this type of attacks.
Passive behavior:
In this type, each eavesdropper will work individually to intercept the message, without doing
Figure 1.7 Eavesdroppers’ passive behavior
any action that could lead to detect its real identity or its place. Therefore, it’s hard to locate
this kind of eavesdroppers compared to the active one, Fig.1.7.
Colluding behavior:
For this type of behavior, the eavesdroppers will collude together to intercept the message,
by sending all their received signals towards another wiretapper. In the literature, this
behavior was called the colluding, cooperating or aggressive behavior, Fig1.8.
Figure 1.8 Eavesdroppers’ colluding behavior
1.2.4.1 Cooperative Jamming
To confuse the eavesdropper, a special technique called cooperative jamming can be used
where an artificial noise is introduced by a helpful interferer. The secrecy performance
analysis with cooperative jamming was studied in the presence of the impulsive noise Atallah
et al. (2019), aggressive relays Atallah & Kaddoum (2016), mixture Gamma distribution
Kong & Kaddoum (2019), αμ fading channels Kong & Kaddoum (2019), Hybrid Millimeter
Wave Networks Vuppala et al. (2018) , and device-to-device (D2D)-enabled cellular
networks Tolossa et al. (2018). In the following section, we will introduce the cooperative
jamming techniques which are used to increase the physical layer security. To improve the
secrecy capacity, we should either increase the legitimate receiver’s SNR or decrease the
eavesdropper’s SNR. A natural approach to achieve the latter is to introduce interferers into
the system, Fig. 1.9.
Figure 1.9 Representation of a network with a jammer
1.2.4.2 Artificial Jamming Signals Types
Cooperative jamming relies on creating the interference at the eavesdropper’s side, many arti-
ficial jamming signals are used and could be divided into four categories Long et al. (2014):
1. Gaussian noise: which is similar to additive noise at the receiver Atallah & Kaddoum
(2016, 2017); Atallah & Kaddoum (2019).
2. Jamming signals which are priory known at the legitimate receivers, and thus only
impact the eavesdropper’s performance. This type of signals is better than the previous
one because the jamming signals don’t affect the legitimate receiver Long et al. (2013);
Dong et al. (2011).
3. Random codewords of a public codebook which is known by all the nodes including the
eavesdroppers, so the legitimate receiver has the ability to decode and cancel the
jamming signals, even though it requires a complicated self-interference cancellation
receiver to decode the codewords Pierrot & Bloch (2011).
4. Useful signals for the other legitimate nodes; signals of multiple simultaneous source-
destination pairs Sheikholeslami et al. (2012), or signals of the invited cognitive radio
users Stanojev & Yener (2011) and Stanojev & Yener. (2011); this type is difficult to
apply because of the change in the multiple transmission pairs.
5. Random fake signals that the legitimate transmitter sends to confuse the eavesdroppers.
The legitimate receiver uses self-interference cancellation to cancel these fake messages
Atallah & Kaddoum (2019).
In the following subsection, we will explore some applied policies with the cooperative jam-
ming technique to enhance the performance and increase the security.
1.2.4.3 Jamming Policies
Several policies were proposed for relay selection to secure the communication Liu et al.
(2015); Sun et al. (2015); Hui et al. (2015); Jameel et al. (2018). In Liu et al. (2015), four
relay selection policies are proposed and compared, particularly random relay and random
jammer, best relay and best jammer, random jammer and best relay, and also best relay and
no jammer. This work characterizes the proposed relay selection policies, impact and the
power of interference constraint on the secrecy performance by deriving new exact closed-
form expressions for the secrecy outage probability; it is shown that the jammer’s absence
raises the outage saturation phenomenon. In Hui et al. (2015), selection methods for the
relay and the jammer were developed in order to minimize the secrecy outage probability; in
these selection methods, the knowledge of the jammer and relay set is kept secret to the
eavesdropper while each intermediate node knows its own role. As a result, the
maintenance of the privacy of the selection greatly improves the SOP performance of the
system. This work assumes a decode and forward relay system, in which through
intermediate nodes in the presence of numerous passive eavesdroppers, the destination can
communicate with the source. To prohibit the eavesdroppers from the interception of the
signal of interest, the intermediate nodes act as jammers or as conventional relays. To
determine whether they will be serving as relays or jammers, the intermediate
nodes take the decision based on the receiving information from the destination. Addition-
ally, the eavesdropper is unaware of the selection result to null the interference towards it. In
Luo & Yin (2018), a new scheme was provided in two-hops wireless networks. The source
communicates with the destination via N relays in the presence of a wiretapper. In each trans-
mission, one of these relays will be selected to jam the wiretapper, while the other N − 1
relays are retransmitting their received messages from the source towards the destination
using distributed beamforming (DBF). Another scheme was provided in Liu et al. (2013)
where in the first phase, the information bearing signal is transmitted by the source
simultaneously as it is cooperating with the destination in jamming the eavesdropper without
interference at the relay. In the second phase, a relay is selected optimally, which transmits
the decoded source signal. Meanwhile, this relay is cooperating with the source to jam the
eavesdropper without creating interference whereat the destination is located. The authors in
Lin et al. (2013a) proposed a new transmission scheme, where the relaying group and the
jamming group are constructed together. The jammers send the jamming signal and the
useful message in the same time. This scheme enables to confuse the wiretappers and
increase the signal-to-noise ratio at the legitimate receiver’s side. In Chen et al. (2013),
attack strategies were investigated in a multi-relay network that consists of both malicious
and cooperative relays, where the malicious relays have the freedom to listen to the
transmitter in the first hop (so that they can send interference signals in the second hop). The
direct emission of jamming signals in both hops is also investigated. Subsequently, it is
shown that the malicious relays should attack in both hops rather than just listening in the
first hop and attacking in the second hop. On the other hand, the opportunistic cooperative
jamming and the opportunistic relay chatting schemes were compared in Ding et al. (2011).
It is shown that the chatting scheme where the relay nodes jam the eavesdropper in the both
phases, is better than the cooperative jamming scheme in which the eavesdropper is only
jammed in the first phase. In Alibeigi & Taherpour (2019), the authors proposed a security
scheme in two-hops D2D communications, based on making use of other cellular users as
friendly jammers to jam an eavesdropper while this latter is trying to intercept the transmitted
message. According to their simulation results, a better secrecy performance is achieved
when the number of cellular users or the distance to the eavesdropper is increased. The
researchers
in Chen (2018) investigated the security over a two-users Gaussian interference channel,
where each source communicates with its corresponding destination. When one of the
destinations receives the other destination’s message, it will treat it as interference. They
showed, for a symmetric case, that the optimal secrecy rate is achieved as long as the
interference-to-signal
ratio in decibel is no more than 2/3. Otherwise, cooperative jamming is needed to achieve
the optimal secrecy rate in their system. Furthermore, another scheme was studied in Mobini
et al. (2019) to secure a source-relay-destination link in the presence of an eavesdropper and
an external cooperating jammer. Two cooperating protocols were investigated: the full duplex
jammer protocol (FDJ), where all the nodes are half duplex except the relay is full duplex,
and the half duplex jammer protocol (HDJ), where all the nodes are half duplex. It is shown
that, from a secrecy perspective, FDJ is superior to HDJ.
1.2.4.4 Cooperative Jamming with Power Allocation
Since the system’s performance in cooperative jamming highly depends on the jamming
strategy and power level Park et al. (2013), three power allocation strategies were derived in
Park et al. (2013) for the SOP to be minimized. Moreover, three kinds of jamming power
allocation schemes are proposed according to the available CSI at the destination to limit
the outage probability. In He et al. (2019), the researchers proposed three user-pair selection
schemes for untrustworthy relay networks with multiple source-destination pairs, namely
opportunistic, greedy and genie-aided user pair selection schemes. They showed that the
greedy user-pair selection scheme overcomes the other schemes from a secrecy perspective,
due to the cooperation by the source that adds flexibility to the network. The authors in
Zhang et al. (2015) investigated the cooperative jamming in MISO channels in which the
legitimate receiver splits the received power for energy harvesting and information decoding.
Another power allocation method is analysed in Long et al. (2014) in which the source
nodes send jamming signals as a part of their power instead of hiring extra nodes to jam the
eavesdropper. Two types of jamming signals are analysed; a priori known jamming signals at
the source nodes, and unknown jamming signals at the source nodes. A major finding
reported in this work is that, if
the jamming signals are known a priori at the source nodes, the secrecy capacity is improved
significantly when compared to the scenario in which the jamming signals are unknown. In
Yang et al. (2014), besides applying cooperative jamming technique, the base station utilizes
a linear precoding scheme, which exploits transmission diversity by weighting the
information stream. When the number of the friendly jammer’s antennas is no less than the
total number of the eavesdropping antennas, an optimal solution is obtained. The authors in
Wang et al. (2015a) proposed a sequential parametric convex approximation (SPCA) based
algorithms to address the power allocation optimization and maximize the ergodic secrecy
rate (ESR) lower bound, and show that the secrecy capacity is improved by the optimized
power allocation that tends to allocate jamming signals more power. An optimal relay
selection criterion and power allocation strategy were derived in Wang & Wang (2014)
between the jamming signals and the confidential information for the ESR maximization.
Another study in Deng et al. (2015) showed that a helper node should allocate its power as a
jammer or as a helper depending on the locations of the helper and the eavesdropper. In Do
et al. (2019), the authors studied the optimal transmit power in the presence of an active
eavesdropper that is jamming the destination. The destination tells the source when to
transmit the data and when to harvest energy depending on the source’s power and the
existence of jamming attacks by the eavesdropper. The aim of their policy is to optimize the
security and the allocated power at the source when it is transmitting data under the energy
harvesting constraint that is applied to the source.
1.2.5 Game Theory for Security
As an effective framework for the design of security mechanisms for wireless networks,
Game theory, traditionally applied in the areas of sociology, economics, biology, political
science, and resource allocation in wireless systems, has recently emerged. Moreover,
jamming policies using game theory methods were proposed in Fakoorian & Swindlehurst
(2013); Chen et al. (2013); Stanojev & Yener (2013); Li et al. (2014b). In Chen et al. (2013),
a multi-relay network, consisting of both malicious and cooperative relays, applies Nash
equilibrium game strategy on its scheme, by modelling the sets of malicious relays and
cooperative relays as two
players in a zero sum game with the maximum achievable rate as the utility. The authors in
Fakoorian & Swindlehurst (2013) proposed a scheme of two users MISO Gaussian interfer-
ence channel, where the transmitters aim to maximize the difference between their secrecy
rate and that of the others. In this scheme, the weaker link tries to minimize the gap between
its secrecy rate and that of the other transmitter, while the transmitter with the stronger link
tries to maximize this gap. This paper used Nash equilibrium strategy as a solution in its
scheme. In Houjeij et al. (2013), using the non-cooperative game theory framework, the
interactions in CRNs between secondary users (SU)s and eavesdroppers were analysed. A
novel secure channel selection algorithm has been proposed to solve the formulated game;
the eavesdroppers and the SUs are enabled to take distributed decisions in order to reach a
Nash equilibrium point. As showed by the authors, in terms of the average secrecy rate per
SU, the proposed approach yields significant performance improvements especially when
compared to a classical spectrum sharing scheme. The researchers in Stanojev & Yener
(2013) proposed another game-theoretic model, Stackelberg game, with the legitimate parties
being the owners of the spectrum acting as a game leader, and the set of the assisting
jammers become the followers. They showed that as the number of potential jammers
increases, a chosen jammer’s utility will decrease because of the aggressiveness of the game
leaders, i.e. the legitimate parties. In Li et al. (2014b), it was shown that the strategies of the
legitimate transmitters quickly learned by a smart jammer would lead to an adjustment of the
jammer’s strategy to damage the legitimate transmission. Meanwhile, the existence of the
smart jammer is well known from the transmitters. This scenario of anti-jamming is
modelled as a Stackelberg game, where the leader is the source node and the follower is the
jammer. It is shown that the obtained optimal power control strategies from the Stackelberg
equilibrium game can minimize the effect of the damage caused by the jammer. As proposed
by the authors in Zhu et al. (2010), and based on a reversed Stackelberg game, a secure
cooperative spectrum trading scheme in CRNs is applied; the illegal actions of the SU are
automatically supervised by the PU, who will adjust it’s strategies according to the actions of
the SU. In Badia & Gringoli (2019), the authors studied a game theory scenario in the
presence of a malicious node for two scenarios: when there is only one friendly jammer and
when there are multiple friendly jammers. Their study showed that even though the exis-
tence of multiple friendly jammers enhances the security, it is still not effective enough to
stop the malicious node from trying to perform unauthorized transmissions because of the
lack of coordination between these jammers. Another interesting dynamic psychological
game study, between a soldier and an attacker, was investigated in Hu et al. (2019), where the
soldier tries to finish his mission in a certain time to pass through the battlefield while
keeping connected to the Internet of Battlefield Things (IOBT). In the mean time, the attaker
is trying to delay the soldier’s connection time with the IOBT by using jamming. Their
results showed that by using their proposed Bayesian updating algorithm, the soldier and the
attacker can reach ε-like psychological self-confirming equilibrium strategies for their
proposed psychological game. However, more studies are still needed to investigate the
schemes where there are multiple de- fenders versus multiple attackers. Additionally, in most
of the studies, the assumption of that the defender and the attacker can detect the system state
with no error, needs to be relaxed.
1.2.6 Key Generation Technique
Key generation is a technique where two legitimate nodes extract secret symmetric key bits
by exploiting the fluctuations and the random characteristics of wireless communication
channels. It is a low cost solution since it does not require complex operations. For the
model shown in Fig. 1.1 where Bob and Alice want to establish a secure key while the
eavesdropper is listening to the legitimate channel between them. Bob, Alice, and Eve
can get correlated
observations Yn = (Y1, ...,Yn), Xn = (X1, ..., Xn) and Zn = (Z1, ..., Zn), respectively. Over the
legitimate channel, Bob and Alice will exchange a message s while Eve is trying to eavesdrop
it.
gB(For a sufficiently large n and any ε > 0, R is the achievable key rate if there exists KB =
Yn, s) and KA = gA(Xn, s) making the key generation satisfy the following Zhang et al.
(2016a):
• Bob and Alice are generating the same key with high probability Pr(K A ×K=B) < ε.
• No information is leaked to the eavesdropper which means guaranteeing the generated key’s
secrecy 1 I(KA; s, Zn) < ε where I(.) is the mutual information.

• The key rate R satisfies R > 1n H( K A )+ ε.
1 1
n n
A
notesthe
• Finally, thegenerated
alphabet of
keythe uniformly distributed log |K | < H .K . + ε, where K de-
is generated
key.
Moreover, the key capacity which is the largest key rate achieved is given by
CK = min[I(X ;Y ), I(X ;Y | Z)] , (1.14)

Most of works studied the key generation technique based on the characteristics of wireless
channels, while a few works have investigated schemes with static channels. The authors in
Guillaume et al. (2015) proposed a new scenario to generate random keys in static channels
by using a moving third party to exploit the channels’ characteristics between him and the
legitimate nodes to generate the key. In Madiseh et al. (2012), another scheme is proposed
to generate keys in static environments by employing random beamforming. Furthermore,
Huang & Wang (2013) also proposed a key generation scheme aided by frequency diversity
and opportunistic beamforming for long coherence time channels. In Felkaroski & Petri
(2019), the authors generated their keys from the CSI that was extracted from multiple
mmWave subcar- riers. This generation method yields a very fast bit generation rate, which
enabled the communicating legitimate nodes to establish and refresh the shared generated
secret key in a very short period of time. Also, the authors in Zhang et al. (2018) proposed a
two-way secret key generation method, where each legitimate node shares its random signal
with the other legitimate node through the reciprocal channel. Then, each node will
generate keys benefiting from the randomness that comes from multiplying its received signal
by its local signal. By applying this method, there is no need to rely on the CSI to generate
keys. The results showed the effectiveness of this method, not only theoretically, but also
practically.
1.3 Unrealistic assumptions
In the literature, most of the security methods rely on knowing the eavesdropper’s channel,
location or both of them. Additionally, for our best of knowledge, the eavesdroppers in all the
scenarios were receiving the whole transmitted message all the time. Many studies
investigated in the eavesdroppers’ passive behavior, and few considered the colluding one.
Also, key generating techniques didn’t exploit the presence of having many untrusted
nodes, and the secrecy performance in impulsive noise invironments wasn’t been analysed
yet. Therefore, we covered each of these aforementioned points in our following articles.
CHAPTER 2
SECRECY CAPACITY SCALING WITH UNTRUSTWORTHY AGGRESSIVE

RELAYS COOPERATING WITH A WIRE-TAPPER
Michael Atallah1, and Georges Kaddoum1
1
Department of Electrical Engineering, École de Technologie Supérieure,
1100 Notre-Dame Ouest, Montréal, Québec, Canada H3C 1K3
Paper published in IEEE Wireless Communications Letters, August 2016.
2.1 Abstract
In this paper, we investigate the performance of the secrecy capacity in an amplify-and-

forward (AF) dual-hop network for both distributed beamforming (DBF) and opportunistic
relaying (OR) techniques. We derive the capacity scaling for two sets; U untrustworthy
aggressive relays cooperating together with a wire-tapper to intercept the message, and T
trustworthy relays, for a large number of nodes. We prove that the capacity scaling in the
DBF is bounded by a value depending on a ratio between the number of the trustworthy and
the untrustworthy aggressive relays. Finally, we show that DBF is better than OR whose
capacity scaling is proved to be upper-bounded by a value tending to zero when the
untrustworthy relays are aggressive. Simulation results confirm our analytical derivations.
Keywords: Physical layer security, cooperative jamming, distributed beamforming, oppor-

tunistic relaying, amplify and forward.
2.2 Introduction
Security has always played a critical role in wireless cooperative communication systems de-
sign. The basic notion of physical layer security is to increase the legitimate links capacity
while decreasing the capacity of the illegitimate links, which is achievable via the utilisation
of the dynamic nature of wireless channels Liang et al. (2008); Gopala et al. (2008). Many
con-
30
tributions have been recently made to increase the secrecy capacity by combining advanced
strategies in wireless communications like beamforming, multiple antenna schemes, game
theory techniques and power allocation methods Atallah et al. (2015). Because of the
broadcast nature of the wireless network’s medium, all the users could be potential
eavesdroppers within the transmission range. Considering this point, recent works show that
the secrecy rate could be enhanced when treating the untrusted nodes as relays instead of
treating them as eavesdroppers Jeong et al. (2012); Yener & He (2010). In Jeong et al.
(2012), this scenario is extended to MIMO scheme with beamforming strategy. Furthermore,
asymptotic analysis is often seen in research as the “end of the line”; the attaining of a result
that cannot be dramatically improved upon. Therefore, asymptotic analysis is a method of
describing limiting behaviour in systems when they are very large. Hence, many works study
the scaling performance from a security perspective Sun et al. (2012); Kim et al. (2015).
Using the opportunistic relaying OR scheme as described in Sun et al. (2012), the scaling law
of the secrecy capacity is investigated for multiple untrustworthy relays. The researchers in
Kim et al. (2015) have studied the maximum capacity scaling according to the number of the
untrustworthy relays by considering all relays as untrustworthy passive nodes. Hence, the
capacity scaling where the untrustworthy relays are aggressive by sending their messages to a
wire-tapper has not yet been studied in the literature. Therefore, we investigate the following
questions in this paper:
• What is the minimum secrecy-capacity scaling according to the number of untrustworthy

aggressive relays U and trustworthy relays T ?
• Which scenario would be better, DBF or OR?
Our considered cooperative network contains one source, two sets of relays using AF tech-
nique, and one destination. Hence, the first set includes untrustworthy relays which are col-
laborating together to eavesdrop the transmitted messages through an external wire-tapper,
and the other set includes trustworthy relays. Moreover, to reduce the eavesdropping capacity
of untrustworthy relays, the destination acts as a jammer and transmits a jamming signal to
the relays during the first phase of communication. In this paper, an investigation of the
31
asymptotic
performance of DBF and OR using AF technique is performed. The contributions presented
in this paper are; 1) The secrecy capacity scaling is provided by showing that DBF has the
scaling of 1 log2( T + 1) at either the absence or the presence of the intended jamming (IJ) by
2 U
the destination. 2) In OR, the secrecy capacity scaling is upper bounded by a value tending to
zero for large number of T and U .
Notations:
able (r.v.) XE[X ] and VAR[X
. Furthermore, ] denote
fx(.) and Fxthe mean expectation
(.) denote anddensity
the probability the variance of a(PDF)
function random
andvari-
w.p.1
the cumulative distribution function (CDF) of X , respectively. More, → denotes the con-
vergence with probability one, [A,]+ denotes max{A, 0}. For , two functions f (x) and g(x), f
(x) ∼ g(x) means that limx→∞ f (x) g(x)= 1 and lim1/x→0 f ( 1 ) g( 1 ) = 1. For a r.v. X , the
x x
notation X ∼ Nc(a, b) denotes that X is a complex Gaussian r.v. with mean a and variance b.
2.3 System Model
Consider a two-hop wireless network which consists of a source s communicating with a des-
tination d through a set of amplify-and-forward relays Rk = {1, 2, ..., K} divided into two
sets, an untrustworthy set Ru = {1, 2, ...,U } and a trustworthy set Rt = {1, 2, ..., T} of
relays, where Ru ∪R t = Rk. The trustworthy relays are considered as an essential part of the
network, whereas the untrustworthy relays are the nodes that login to the network for a long
enough period of
time. Moreover, the destination broadcasts the jamming signal toward all the relays. In our
network, each relay has a single antenna operating in half-duplex mode, as shown in Fig.
2.1. On the other hand, the external wire-tapper cooperates with the untrustworthy relays to
decode the transmitted symbols of the source thanks to the different replicas of the source’s
signal relayed by these latter. Moreover, there is no direct link between the source and the
destination in our system, i.e. all transmitted information must pass by relays. In our analysis,
the channels are assumed to be uncorrelated reciprocal frequency-flat block-fading with the
coefficient between nodes i and j being denoted by hi, j and being modelled as a complex
Gaussian random variable
. .2 . .2
where (i, j ∈ {s, Rk, d}). Therefore, the channel gains h and hk,d are independent and
exponentially distributed r.v.’s whose means are σ 2 ands,kσ 2 respectively. We assume that the
1 2
Figure 2.1 System model
noise variance N0 is the same in the first and the second hop, and the channel state
information CSI is known by the receiving nodes. The source broadcasts the message signal
xs in the first hop of transmission to the relays while the destination sends a jamming signal
xd towrds the relays. In the first hop, the received signal at the kth relay is given by
√ √
yk = hs,k Ps xs + hk,d Pd xd + nk . (2.1)
Where nk is a complex additive white Gaussian noise at the kth relay with zero mean and
variance N0. The transmitted powers of the source and the destination are denoted by Ps and
Pd respectively. It is assumed that at the kth relay, the received signal-to-interference-plus-
noise ratio SINR would be
ρs .hs,k.
2
γk = . (2.2)
ρ .
.
d hk,d
2
+1
Δ
,
where the signal to noise ratio SNR is denoted by ρ J = N0 , i ∈ {s, d}, ⊂ ρ where
Δ i
Pi ρ J
,
ρ j = Pj N0 , j ∈ {s, t, u, d}. In the following subsections, we will derive the secrecy capac-
ities of DBF and OR respectively.
2.3.1 Distributed Beamforming
second xi = ai wi yi . where wi, i ∈ {t, u} represents the optimized beamforming

weight. hop is the
Using DBF strategy at the relays’ side, the retransmitted signal by the ith relay in
the The normalized amplifying coefficient ak for the kth relay is as follows
1
a
.hs,k. 2
.hk,d 2 .
s
k = .ρ + ρd .
+ N0
The destination receives a signal transmitted by each of the relays, which can be expressed as
yd = hk,dak wk yk + nd . (2.3)
Similarly, the wire-tapper receives a signal transmitted by each of the untrustworthy aggressive
relays
ye = hu,eau wu yu + ne . (2.4)
Where hu,e is the channel coefficient between the untrustworthy relay and the external wire-
tapper, nd and ne are complex AWGN with zero mean and variance N0 at the destination and
at the wire-tapper respectively. After removing the jamming signal xd at the destination, the
received SINR becomes
ρ . T ρs |hs,t | ρt .ht,d .
DBF U . s |hs,u|2ρu hu,d 2
u=1 s s, s,
2
ud u u 2 2
γd =∑
ρ |h +∑
t=1 s s,
2
d t t,d
| + (ρ + ρ ) | 2
| + (ρ + ρ ).h 2 . (2.5)
t
| +1 ρ| .+1
h
h
Whereas the recieved SINR at the wire-tapper becomes
U
ρs |hs,u|2ρu |hu,e|2
γ
=∑
e u=1 ρuρd .h u, .2 |hu, |2 + ρs |hs, |2+ ρ d.h 2
. (2.6)
u,
.2 + ρu|h u,
d e u
e |+1 d
Therefore, the instantaneous secrecy capacity of the DBF could be written as
.
DBF Σ1 ρs |hs,t |2 ρt .ht,d .2
T
.
log
2 2 1 + t=1 ∑
CS = ρ| | + (ρ + ρ ).h 2
+1
2
s s, d t t,d
h t
2 2
+∑
.
U ρs |hs,u| ρu .hu,d Σ
1 .
2 . | + 2(ρd
u=1
| 1 + e+ 1
ρ
s −
2
hs,u+ ρ u) .h d
u,
log2 γ Σ+
Σ
d
Σ Σ+
= CDBF −Cw . (2.7)
The data rate between the wire-tapper and all the untrustworthy relays cooperating with it is
denoted by Cw. Since the relays are half-duplex, we use the rate-loss factor value of 1/2.
2.3.2 Opportunistic Relaying
During the second hop, only the best relay b that has the maximum SNR at the destination
retransmits the signal. Through the kth relay, the end-to-end SINR is as follows
2 2
ρs .hs,k. ρk .hk,d .
γs,k, = . (2.8)
ρ . .2 + (ρd k,
.2 + 1
d
s d
Hence, the best relay is selected as
hs,k + ρk ). h
b = argmax {γs,k,d . (2.9)

k∈Rk }
Therefore, the achievable secrecy capacity becomes
Σ . γ Σ
OR 1 1 Σ+
CS = 1 + max( s,k,d) .1 + e Σ
k − γ
2 2
log2
d w
= ΣCOR − C Σ+ . (2.10)
2.4 SCALING LAW OF SECRECY CAPACITY
2.4.1 Scaling Law of Distributed Beamforming
This subsection shows that the secrecy capacity scaling of untrustworthy aggressive relays in
DBF is not the same as the maximum secrecy capacity for trustworthy relays. Without loss of
=
generality, we assume in our analysis that ρ Δ ρs = ρt = ρu = ρd .
Theorem
the 1. When T → ∞ and U → ∞ with any finite ρ, the ergodic secrecy capacity of
S S 2 2 U
DBF
C = E{C DBF
} is lower bounded by 1 log ( T + 1).
1
scaling through
It is trustworthy relays ineta al.
dual(2006)
hop network
that forisTupper
→ ∞ bounded
and any by
finitelogρ,2(Tthe
).
But Proof. shown in Bolcskei
capacity considering the aggressive behaviour of the untrustworthy relays, the secrecy
capacity becomes 2
lower bounded by
S
DBF . S
.Σ1 C
. =T E CDBF
ρ |hs,t 2| ρ .ht,d2.
=E log2 1 + | + 2ρ .h 2 + 1
2 Σ ∑ ρ |h s,t 2 t,d .
2
t=1
2 Σ
+∑
U ρ |hs,u| ρ2 .hu,d. γ e)
s, .2 + 1 − log2(1 Σ+Σ
u=1 ρ | u 2
u,
1
|d + 2ρ .h +
ρ |hs,t 2| ρ .ht,d2.
≥ E Σ . .1 + ∑
(a) 1 2 .2 + 1
t=1 ρ |
s,
2 t
| + 2ρ .ht,d
T
U ρ |hs,u|2 ρ .hu,d. 2
ΣΣ . γ ΣΣ+
1
+∑ 2 . .2
ρ |h s,u | + 2ρ h u,d + 1 −E
2
log2(1 + e)
u=1
Σ . d Σ Σ+
= E CDBF −E {Cw} . (2.11)
where (a) follows from the fact that E{max(X1, X2 )}≥ max(E{X1}, E{X2}). Let
Δ
. s,u.| 2|ρ hu,d 2 , (2.12)
2 2 2 ρ h2 2 2
Nu =
ρ .hd,u. |hu,e| + ρ |hs,u| + ρ |hu,e| + ρ .hu,d. + 1
and
Δ
.| |ρ ht,d 2 .
. s,t 2
(2.13)
2ρ h 2
Mt =
ρ |hs,t | + ρ .ht,d. + 1
VAR[Mt ]
M satisfies the Kolmogorov conditions i.e. T < ∞ and μ = 1 T E [ ] < ∞ are true
∑ Mt
t 2 t T ∑
t=1 t t=1
for any finite ρ Bolcskei et al. (2006). Nu also satisfies Kolmogorov conditions since Nu < Mt ,
so we can apply the following theorem Serfling (1980):
U U
Nu E [Nu] w.p.1
∑ ∑ → 0. (2.14)
− U
U
u=1 u=1
. Σ 1 1
U
w.p.1 DBF
Therefore, γdDBF → , and C
u
∼ 2 log2(T +U ), where μu = U
U E
μ
d
∑
E [Nu] < ∞.
u=1
DBF Σ . Σ Σ+
C ≥ E CDBF −E {C }
S d w
Σ Σ+
1 1
∼ log2(T + U ) − log2(U )
2 2
2 U
1 T
= log2 . + 1Σ. (2.15)
It is clear from (2.15) that the lower bounded secrecy capacity depends on the ratio between
T and U . For example, to maintain a certain level of secrecy capacity in a wireless network,
the maximum number of untrustworthy aggressive relays should not exceed:
T
U (2.16)
≤
DBF
S
. 22C
Let’s assume that the wire-tapper could receive more signals from other trustworthy relays T J
signals
where Tfrom
J
≤ Twill be U J , the
. Hence, where
totalU number
J
= T J + of
U .the
In relays
this case,
that and
the by followingcould
wire-tapper the same stepsthe
combine in
the proof of Theorem 1, the secrecy capacity scaling in equation (2.15) will be
Σ
DBF 1
CS j Σ
log2(T + U ) − 1 log2(U +
Σ2 2
)1 1
= +U
1 log2(T U)−
T +U j) Σ+
= 2log . Σ. 2 (2.17)
2 log22(T
TJ+U
which tends to zero for T J = T .
Moreover, we will show in Theorem 2 that even when the destination is not jamming, the
secrecy capacity scaling will tend to the same bound value of DBF with IJ as given in
equation (2.15).
Theorem 2. When T → ∞ and U → ∞ with any finite ρ, the ergodic secrecy capacity of a
DBF
without IJ tends to the value 1 log2( T + 1).
2 U
Proof. Considering that the ergodic secrecy capacity of a DBF without IJ is given by
DBF
DBF Σ
CS,NoJam . S,NoJam
.Σ1 =E
. Σ
T ρ |hs,t | ρ U ρ |hs,u| ρ .hu,d.
.ht,d.
2C
t=1 s, 2 2
t
t,d
1 +ρ∑|h
=E log
2
2 2
2 + − 2 .
| + ρ .h ∑ |+ρ
2
+1 u=1 s, u,
u d
. +1 .h
1
log2(1+ ∑U ρs | | ρu | Σ+Σ . (2.18)
2 2
hs,u hu,e|
2 2 )
2 u=1 ρs |hs,u| + ρu |hu,e| + 1
It can be shown by following the same procedure as that in Theorem 1 that E ,CDBF
S,NoJam
,∼
1 T
2 log2( U + 1).
2.4.2 Scaling Law of Opportunistic Relaying
It was shown that the lower bound for ergodic secrecy capacity tends to zero as the total
number of the relays K → ∞ Sun et al. (2012). Considering the aggressive behaviour of
the untrustworthy relays, we calculate here the upper bound value for the secrecy
capacity.
Proof. Let
1
X = 1 + max( ρ .hs,k.2 2ρ .hk,d .2
),
2 2 2
k ρ .hs,k. + 2ρ .hk,d2. + 1
2
U ρ |hs,u| ρ .hu,d. 2,
Y=1+∑ 2
u. |hu,e
2 2 2
2 s u u
, |+ρ | , |+ρ | , ,
u=1 ρ
.h d u | + ρ .h e d
2 h h . +1
and
Z max( ρ|h
k |s k2
=1+ ,
). Then, the secrecy capacity will be upper bounded by
OR .Σ Σ+ Σ
E{CS } = 1 log2( X
2 Y
E 1 ) X
< E . log2 (1 + )Σ
2 Y
1 1
= E Σ log2(X + Y ) − log2(Y )Σ
2 2
1 1
(a)
≤ log2 .E [X ]+ E [Y ] Σ −
log 2 .E [Y ] Σ
2 2
< 2 2
2 2
1 1
(b) log .E [Z]+ E [Y ] Σ − log
.E [Y ] Σ
1
(c)1 ρ
∼ log2 . 2log2 K + U Σ − log2(U )
2 2
1 ρ log K
∼ 2 log2 . 2 2 + 1Σ, (2.19)
U ρ|h s k | 2ρ|hk d |
where (a) follows from Jensen’s inequality, (b) follows from the fact that , 2 2 , 22
ρ|hs k| +2ρ|hk d| +1 <
2 2 , ,
ρ|hs k| 2ρ|hk d|
,
ρ|h, s,k .2
.hs,k , .hk, .2 .2
.hs,k , (c) follows from the fact
)≤
2 ≤ min(ρ that
,
. | .2
2
| +2ρ|hk d ρ
E{maxk(ρ hk,d )} ∼ ρ log2 K + O(log2 log2 K) Sharif & Hassibi (2005) and E(Y )= U by
following the same steps in our proof of Theorem 1. However, when T → ∞ and U → ∞
with
any finite ρ, equation (2.19) tends to zero.
Therefore, based on the results given in equations (2.15) and (2.19), we can conclude that DBF
guarantees better security than OR technique.
Assuming that ρ Δ ρs = ρt = ρu = ρd = 5 dB and U = T , we do the performance comparison
of the ergodic secrecy
= capacity between the lower bound DBF, OR and DBF with and
without
IJ and we show the outcome in Fig. 2.2. Moreover, we assume that the relays are located near
the middle of the source and the destination, and the variances σ 2 = σ 2 = 1. It is observed in
1 2
Fig. 2.2 that increasing the total number of relays gives better performance for DBF network,
but it reduces the secrecy capacity of OR to zero even at the presence of IJ. Moreover, Fig.
2.2 shows a secrecy capacity gap between DBF without IJ and the one with IJ. Therefore, it
can be seen that with the absence of IJ, DBF tends to reach its secrecy capacity scaling
quickly.
2.5 Conclusions
The capacity scaling of secure cooperative relaying with DBF and OR through trustworthy
and untrustworthy aggressive relays has been investigated in this paper. Considering the
aggressiveness of the untrustworthy relays in DBF, we conclude that 1) Secrecy capacity
scaling is bounded by a value that depends on the ratio between the number of the
trustworthy and the untrustworthy aggressive relays, this value is reached quickly with the
absence of IJ. 2) Based
Ergodic secrecy capacity [bps/Hz]
1.6
DBF with intended jamming
1.4 DBF without intended jamming OR with intended jamming
OR without intended jamming DBF secrecy capacity scaling
1.2
0.8
0.6
0.4
0.2
0
020406080100120
Number of relays K
Figure 2.2 Ergodic secrecy capacity: ρ Δ ρs = ρt = ρu = ρd = 5 dB,

=
σ 2 = σ 2 = 1, and U = T
1 2
on our results, OR is not recommended for security issues, leading to the priority being
handed over to DBF strategy.
CHAPTER 3
SECRECY ANALYSIS IN WIRELESS NETWORK WITH PASSIVE

EAVESDROPPERS BY USING PARTIAL COOPERATION
1
Paper published in IEEE Transactions on Vehicular Technology, 2019.
3.1 Abstract
This paper proposes a new location-based multicasting technique, for dual phase amplify-
and- forward (AF) large networks, aiming to improve the security in the presence of non-
colluding passive eavesdroppers. These eavesdroppers could also be part of this cooperative
network as relays. In order to reduce the impact of these eavesdroppers on the network
security, we propose a new transmission strategy where, for the first hop of each transmission
time, while the destination is jamming, the source randomly chooses a different subset K of
the total T relays, to transmit its message towards the destination. For practical
implementation, sectoral transmission can be achieved with analog beamforming at the
source’s side. In the second hop, using the distributed beamforming technique, the K AF
relays retransmit the received signal to the destination. We analytically demonstrated that the
proposed technique decreases the probability of choosing the same sector that has certain
eavesdroppers again, for each transmission time, to K/T . Moreover, we also show that the
secrecy capacity scaling of our technique is still the same as for broadcasting. Hereafter, the
lower and upper bounds of the secrecy outage probability are calculated, and it is shown
that the security performance is remarkably enhanced, compared to conventional multicasting
technique.
Keywords: Physical layer security, jamming, secrecy outage probability, amplify and
forward, secrecy capacity, scaling.
42
3.2 Introduction
Physical layer security (PLS) is considered a promising approach for strengthening the
security in wireless communication. One of the most important tools to measure the security
performance in PLS is the security rate, in which the channel capacity of the legitimate
links should be higher than the capacity of the illegitimate ones. Otherwise, it is equal to zero
Gopala et al. (2008). In order to achieve a positive secrecy rate, many techniques have been
proposed, such as cooperative jamming (CJ) Lee et al. (2018); Atallah et al. (2015), multi-
antenna scenarios Chen et al. (2017), beamforming Guo et al. (2017), game theory Silva &
Cordero (2017), and power allocation schemes Atallah et al. (2015). In the literature, the
aforementioned techniques were sometimes combined to achieve better security. In Cumanan
et al. (2017), a CJ technique, by multiple jammers, was combined with an optimal power
allocation technique to achieve a better security rate in the presence of multiple
eavesdroppers. The authors in Wang et al. (2013a) combined cooperative beamforming (CB)
and CJ techniques to achieve higher security. The combination of CB and CJ was studied
again in Wang & Wang (2015) in the presence of multi-antenna eavesdroppers. Due to the
nature of the wireless medium, nodes can join and leave the network frequently. These nodes
could be beneficial to the network, when being used as relays, and could also be considered
as potential eavesdroppers. However, as demonstrated in Jeong et al. (2012) and Yener & He
(2010), treating these nodes as relays could be more beneficial to the wireless network, from
a security perspective, than treating them as eavesdroppers. In cooperative relaying networks,
two main scenarios were studied in the literature, the opportunistic relaying (OR) one,
where the best relay is chosen to retransmit the message, and the distributed beamforming
(DBF) one, where all the relays retransmit their received messages towards the destination
using the beamforming transmission. In A. El-Malek et al. (2017), the secrecy performance,
considering OR networks, was studied after applying power allocation and jamming
techniques, in the presence of interference and many eavesdroppers. Also, the secrecy outage
probability (SOP) lower and upper bounds, in OR networks, were investigated in Mabrouk et
al. (2017) using CJ under outdated channel state information (CSI). Moreover, hybrid
schemes that contain DBF and CJ were investigated in Wang & Xia (2015). Hereafter, a
joint cooperative beamforming, jamming, and power-allocation scheme was proposed in
Wang et al. (2015b) to enhance the security performance in cooperative relay networks. In
this direction, the authors in Kim et al. (2015) studied the secrecy capacity scaling and the
asymptotic performance of a two-hops network, with untrustworthy relays, for both OR and
DBF. There- after, these scenarios were extended in Atallah & Kaddoum (2016) to study the
secrecy scaling laws for dual phase large networks, with wiretappers that are cooperating
between each other to intercept the messages. The majority of the proposed techniques in the
literature assumed that the wiretappers are receiving the data all the time, which can harm the
security. To tackle this challenge, in this paper, we propose a new location-based multicasting
technique, based on sending just a part of the information to a certain sector for a certain
transmission time, and then switching to other sectors randomly to send the other parts. This
transmission strategy can be practically implemented thanks to analog beamforming at the
source’s side. We mathe- matically demonstrate that our proposed technique reduces the
possibility of an eavesdropper intercepting the whole message, since it’s getting just a part of
it. Also, we show that the secrecy capacity scaling converges to the same value of the
broadcasting method in Kim et al. (2015). Moreover, an analysis of secrecy outage
probability (SOP) lower and upper bounds is provided, and shows a remarkable
improvement compared to the conventional multicasting scenario and OR techniques in A.
El-Malek et al. (2017) and Mabrouk et al. (2017).
Notations: VAR[X(r.v.)
a random variable ] andX E[X ] respectively
. Also, FX (.) and fXdenote the variance
(.) denote and thedistribution
the cumulative mean expectation
functionof
w.p.1
(CDF) and the probability density function (PDF) of X , respectively. Moreover, →
denotes
the convergence with probability one, and [A ]+ denotes max{A, 0}. For a r.v. X , the notation
X ∼ Nc(a, b) denotes that X is a complex Gaussian r.v. with variance b and mean a.
3.3 System Model and Problem Formulation
G = T/K ∈
Consider N+ clusters, where
a multi-antennas K is
access the number
point of relays in each cluster,
s, T amplify-and-forward a destination
(AF) relays clusteredd into
and
a passive eavesdropper e, that could also take part of this cooperative network as a relay.
Each
Figure 3.1 System model consisting of a multi-antennas source s, T relays
clustered in G sectors, a destination d and an eavesdropper e. In this figure,
T = 9, K = 3 and G = 3
relay has a single antenna, as shown in Fig. 3.1, and operates in a half-duplex mode. It is
assumed that there is no direct link between the source and the destination, which means that
all the transmitted information should be forwarded by the relays. This scenario can represent
a D2D cooperative network. As shown in Fig. 3.1, in the first hop of each transmission time,
while d is jamming the relays, s will multicast the signal xs,g to the gth cluster that contains K
relays in it, where 1 ≤ g ≤ G. In the second phase, the K relays will forward the received
message towards d, using the distributed beamforming technique, which has been proven to
outperform the opportunistic one Kim et al. (2015), Atallah & Kaddoum (2016). After each
transmission time, s will choose another set of K relays to transmit towards d. Hence, the
received signal expression, at a kth relay, where 1 ≤ k ≤ K, is given by
√ √
yk = Ps hs,k xs,g + Pd hd,k xd + nk, (3.1)
where nk ∼ Nc (0, N0) is the complex additive white Gaussian noise (AWGN) at the kth relay,
Ps
and
that Pd are the transmitted powers from the s and d respectively. In our analysis, we assume
the channels are quasi-static block Rayleigh channels, i.e. the channel coefficients hl, j , where
l ∈ {s, k, d} and j ∈ {k, e, d}, are constant during the transmission time of one message,
but may change independently to different values thereafter. Moreover, we assume that the
. h.l,2j follow independent
CSI is known by the receiving nodes. Accordingly, the channel gains
exponential distributions. It is also assumed that the noise variance N0 has the same value in
the first and the second phases. Consequently, the received signal-to-interference-plus-noise
ratio (SINR) at the kth relay becomes
2
ρs .hs,k. , (3.2)
γk =
ρ .
Δ , .
d hd,k
ρ N0, and i ∈ {s, k, e, d}. If e is not 2in+1
the sector covered by the antenna beam of
where i =
the transmitter,
Pi then it will not receive the message xs,g. Thus, the received signal expression
at e can be expressed as
√ √
ye = b. Ps hs,e xs,g + Pd hd,e xd + ne, (3.3)
where b is a Bernoulli r.v. that takes the value 0 when e is in the uncovered sector, and 1 when
e is in the covered sector. Hence, when b = 0, the received signal at e is given by
√
ye = Pd hd,e xd + ne . (3.4)
Since the probability of e being in the covered sector is equal to K/T , then the probability
mass function of b is expressed as
K
Pr(b = 1) = p = , (3.5)
1
T
K
T
Pr(b = 0) = p0 = 1 − p1 = 1 − . (3.6)
From (3.3), the SINR at e is obtained as
b. ρs |
ρ h|2 2 +1
hs,e , (3.7)
γeJ = d . d,e.
The expected value of the SINR at the eavesdropper’s side becomes

1
∑ Pr (b = i) γe b=i J (3.8)
γe = i=0
K
=0+ ρs |hs,e|2
. .
T ρ h d 2 +1
2
K ρs |hs,e| . (3.9)
=
. .dd,e
T ρ h 2 +1
In the second hop, the retransmitted message from
d,e the kth relay will take the following form
xk = αk wk yk, where wk is the optimized beamforming weight, and αk represents the normalized
amplifying coefficient. From (3.1), αk can be calculated as αk = . 1 . In the
ρ | s hs,k + ρ
2 2
second hop, the received message, at d, from the K relays, is given as| d |hd,k| + N0
K
y
d= ∑ hk,d
k=1 αk wk yk + nd, (3.10)
where nd ∼ Nc (0, N0) is the complex AWGN at d. After receiving y d, the destination will
extract xs,g, after removing the jamming signal x d. From Kim et al. (2015) and the references
therein, the SINR at the destination’s side becomes
. . .hk,d .
ρs hs,k 22
γd = ∑k=1
K
ρs .
2 ρ
k
ρ .2 +1
. (3.11)
k,
k d
hs,k . CS
From (3.11) and (3.8), the secrecy capacity .he]+ will be given by
+( = [+Cd −)C
Σ1 .
K . . k .hk,d .
s hs,k Σ
CS = 1 + ∑k=1
ρ .s ρ 2 s, d ρ k,
2 2
h 2 ρk .2 +1k d
. +( + ).h
1 . K ρs | 2 ΣΣ+
hs,e|
.d
hd,e , (3.12)
− log2 1 + .2 +1
2 T
ρ s, to d and e, respectively. The loss rate, 1/2, was used
where Cd and Ce are the data rates from
in (3.12) due to the constraint of operating the half-duplex mode at the relays. In the following
section, we will calculate the SOP lower and upper bounds of CS.
3.4 Lower and Upper Bounds of Secrecy Outage Probability
From (3.12), the SOP expression can be written as
pout = Pr[CS < R] (3.13)

⎡1 + ∑K 1 γ γs,k 2 γk,d
⎤
= Pr⎣ +2 γk,d +1 < u⎦

γs,e
s,k
k=1 K2 ⎤
⎡1 + ∑1K + T1γγγd,es,k+1
2 γk,d
k=1 2 s,k +2
∼ Pr⎣ 1 + K γe s, < u⎦ , (3.14)
T γd,e +1
2R
where R isupper
the threshold andthe
u =SOP
2 .Mabrouk
We will use
et al.the following inequality
lower and bounds of (2017). For any two r.v.toXcalculate
and Y the
1 XY
min {X,Y} ≤ ≤ min {X,Y}. (3.15)

2 X+Y
From (3.14) and (3.15), we have
1 γ 2γ
. Σ k, . Σ
min γs,k , 2γ k, d ≤ ≤ min γs,k , 2γ k, d .
2 γs,k +2 d
s,k
Thus, the SOP lower and upper bounds can be respectively expressed as
. Σ
k=1 min γ γs,k , 2 γk,d ⎤⎦
4 = Pout,UB,
Pout ≤ Pr s,e
<
K
⎡ 1 + 1 ∑K1 + Td,e +1 u
⎣
γ
2 k=1
Pout ≥ Pr .γs, Σ< = Pout,LB.
K + K
1
⎡ 1 + 1 ∑ min γ γ
s,k , 2 k,d u ⎤
T d,e +1
⎣ ⎦
We evaluate the general expression of γthe SOP bounds as follows
. Σ
k=1 min γs,k , 2 γk,d ⎤⎦
Pout,B = Pr θ
K
T γs,
< , (3.16)
⎡ 1 + 1 ∑K1 + d,e
+1
⎣ u
γ
where θ takes the values 2 and 4 for the SOP lower and upper bounds, respectively.
Theorem 3. The secrecy outage probability lower and upper bounds, of our proposed tech-
nique, are given by
p K
=1 λ0 s s (3.17)
− s ( 2−
out,B Γ(K)
( 4 − 5)) , 3
where s3, s2, s4, and s5 are respectively given in (3.25), (3.26), (3.29) and (3.30).
Proof. From (3.16), the SOP is bounded by
. Σ
k=1 Σ
min γs,k , 2 γk,d
Pout,B = Pr ⎡⎣ 1 + θ1 ∑
K
γs, < ⎤⎦
K
1+ T u
e
1
γθ d,e +1
= Pr Σ 1 +u A <
B
θ +A
= 1 − Pr ΣB < Σ, (3.18)
γs, K . Σ
e
∑ min s, , 2 γk, .
k θu d
where B = 1 + KT γd,e +1 k=1 γ
, and A =
θ θu
+A
We have B>1⇒ > 1 ⇒ A > θ u −θ . (3.19)
From (3.18) and (3.19), and assuming that A and B are independent, the SOP bounds are
obtained as
. Σ
pout,B = 1 − ∫ B θ +A (a) da. (3.20)
F
∞
A
f θu
θu−θ
Lemma 1. The PDF of A and the CDF of B are respectively given by

K
λ0
f a aK−1 e −λ a
0 (3.21)
,
A ( )= Γ(K)
KT λ2
FB (b)= 1 − exp Σ− λ1 (b− 1) Σλ1 (b− 1) KT + λ2 , (3.22)
∫ 0∞
where Γ(X )= t X−1 e−t dt, is the Gamma function,
1 1 1 1
λ1 = , λ2 = , and λ0 = + 2. (3.23)
ρ σ12 ρ σ2 2 ρ σ1 2 2ρ σ
2
Where σ . .2 . . respectively.
1 and σ2 are the means of hs,k 2and hk,d
2
2
Proof. Please refer to the Appendix I.
By using Lemma 1’s results, the bounds in (3.20) can be expressed as
pout B = 1 − ∫∞ .1 − exp Σ− λ1 .θ + A − 1Σ T Σ
, θu K
θu−θ
K
λ2 Σ aK−1 e −λ a λ0 da
× . θ +A Σ 0
Γ(K)
λ 1 −1 T+
λ2 θu K
λ0K ⎡ ⎤
∫∞
⎣ s1da⎦ , (3.24)
θu−θ
λ1
θ uK K 1 u
λ2 λ1 T
where s3 = T exp . .1 − ΣΣ, (3.25)
∫
∞
θu− aK−1 e−λ0 a da

s2 =
θ
= λ0−K Γ(K)+ λ0−K [−Γ(K)+ Γ(K, λ0 (θu−θ ))] , (3.26)

and
Kθu
λ1 T
aK−1 exp .−a. λ¸ θuK
0 +xs λ2 ˛
s1 = a +( θ − θ u )
+
λ1 T
s ˛¸
φ
aK−1 exp ( aβ x)
= − . (3.27)
∫ a+φ
Thus, to calculate θu−θ s1 da in (3.24), we will integrate by parts as follows
∞
∞
∞ ∞
∞
∫ s1da = ∫ UdV = VU . − ∫ VdU, (3.28)
θu−θ
θu−θ θu−θ s ˛s4¸ x θu θ
s − ˛s5¸ x
exp(−aβ )
where dV = a+φ
da ⇒ V = exp (βφ ) Ei (−β (a + φ )) , and U = aK−1 ⇒
∫∞ e
dU = (K − 1) aK−2 da. Here, Ei (X ) = − − t −t dt, is the exponential integral of the r.v. X .
X
From (3.28), s4 and s5 are calculated as
s4 = − (θu−θ )K−1 exp (βφ ) Ei (−β (θu−θ + φ )) , (3.29)

∞
s5 = (K − 1) eβφ ∫ aK−2 Ei (−β (a + φ )) da
θu−θ
(†)
= (K − 1) e s7 . (3.30)
βφ
(−β (θu−θ
E1 (β (θu−θ + φ )) s6 −e
+φ ))
(†) follows from using the integral in [Ng (1969), Section 4.1, Eq. 7] after employing the
following property E1 (X )= − E i (−X ) . Thus, s6 and s7 in (3.30) are defined as

K−2
(−1)m (K − 2)!(θu−θ )K−m−2 (θu−θ + φ )m+1
∑ ,
s6 =
m=0 (K − 2 −m)!(m + 1)!
(K − 2)!(θu−θ )K−m−2
s7 = ∑ K−2
m=0 (K − 2 −m)!(m + 1)!βm+1
× ∑mj=0 (−1) j(m− j)! (β (θu−θ + φ )) .
j
Substituting (3.25), (3.26), (3.29), and (3.30) in (3.24), we completed the proof.
3.5 Scaling Law of Secrecy Capacity
In this section, we will show that the secrecy capacity of our proposed technique converges to
1
log2(T ). To prove this, we will show that, for large K and T , the lower and upper bounds of
2
CS
scale to 1 log2(T ). Without loss of generality, in our analysis, we assume that ρ Δ ρs = ρd = ρ k ,
2 =
but the extension using different values of ρi is straightforward.
Theorem 4. The ergodic secrecy capacity CS = E {CS}, for any finite ρ, is lower bounded by
2
1
log2(T ).
Proof.
.Σ . K .hs,k. .hk,d . Σ
1 2
1+∑ ρ
E {CS} = E .h . 2ρ .2 +1
2log2 s,k +2 k,
1 k=1. . d2
s,
K e.
h ρ| | ΣΣ+Σ
− log2
2 T h
1+ .hd,e
2
+1
.hs,k. .hk,d . ΣΣ
2
≥ ΣE . ρρ
log2 .1 + ∑K s, 2ρ
2
(†) 12 k .2 +1
.1 . K ρ| ΣΣΣ+
k=1 . .hs+2
,e|.hk,d
2
T. .
h 1+ hd,e
−E log2 ρ
2 2
+1
+
= [E {Cd} −E {Ce}] , (3.31)
(†) follows from
X2)}applying
≥ max(EJensen’s
{X1}, Einequality
E {max(X1, on theBolcskei
{X2}). From convex et
function max(we
al. (2006), X1,have
X2),that
which is
Kol-
mogorov conditions are satisfied for
ρ .hs,k. ρ.hk,d .
2 2
Δ
M = k,
d .
k 2
ρ .hs,k .2 +1
. + .h
i.e., 1 ∑K E Δ
∞ and ∑ K
VAR[γ ] ∞ are true. Since γ ρ |hs k|2 ρ |hk d|2
k [Mk] < , 2
k
< , dk = 2, , 2 <
K =1 k=1 ρ |hs,k| +2 ρ |hk,d| +1
k
Mk ,⇒ γdk also satisfies the Kolmogorov conditions, which are μ Δ
= K1 ∑Kk=1 E [γdk] < ∞, and
K
∑
k2γdk]
VAR[
k=1 < ∞. Hence, we can apply the following theorem [(Serfling, 1980, 1.8.D)]
K γk K E [γk] w.p.1
→ 0. (3.32)
w.p.1
∑k=1 K − ∑k=1
γ
K
Resultantly, k → Kμ. Substituting in (3.31), we get
CS ≥ [E {Cd} −E {Ce}]+
Σ1 1. ΣΣ+
ρ |h
K
|
= (log2 K + log2 μ) − log2 + log2 2

T 2 2 s,e
Σ Σ+ ρ .hd,e. +1
1 1 K
∼ log2 K − log2
12 2 T
= log2 T. (3.33)
2
From Bolcskei et al. (2006), the asymptotic capacity scaling upper bound of dual phase
relaying networks through trustworthy relays, where the source is broadcasting towards the
relays
in the first hop, is shown to be 1/2 log2 T . Considering, for K → ∞, that our technique acts
like broadcasting, the result in Bolcskei et al. (2006) will be the asymptotic upper bound of
CS. Thus, by showing that both the lower and upper bounds coincide with 1/2 log2 T , we
completed the proof.
3.6 Simulation Results
100
10-1
out,LB
10-2
P
T=10 Sim, our scenario, K=5, T=15 Theory, our scenario, K=5, T=15 Sim, our scenario, K=5, T=20 Theory, o
10-3
10-4
0 5 10 15 20
Eb/N0 [dB]
2 2
proposed system with jamming: .hs,k. = .hk,d . = 1, R = 1bps/Hz, and K =5
Figure 3.2 Analytical and simulated SOP lower bound performances of the
100
10−1
Theory, our tech. K=4,T=8 Sim, our tech. K=6,T=12 Theory, our tech. K=6,T=12 Sim, our tech. K=8,T=16 Theory, ou
out,UB
P
10−2
10−3
0 5 10 15 20
Eb/N0 [dB]
proposed system and OR with jamming: .hs,k. = .hk,d . = 1, and R = 1bps/Hz

Figure 3.3 Analytical and simulated SOP upper
2 bound 2 performances of the
In this section, we evaluate the performance of our scheme by means of analytical and simula-
tion results of the SOP and the secrecy capacity scaling. It’s assumed that ρ =Δ ρs = ρd = ρk,
. . . .
2 2
R = 1bps/Hz and hs,k = hk,d = 1.
Fig. 3.2 shows a comparison, of the SOP lower bound, between the broadcasting (the green
curve), and our proposed technique, which is shown to be, from a security perspective,
remarkably better than the former technique. In the simulation results, it is assumed that K
is fixed and equal to 5. We can see that the analytical expression, given in (3.17), perfectly
matches the SOP lower bound, for different values of T . Also, we can notice that the SOP is
improved when
4.5
4
Secrecy Capacity [bps\Hz]
3.5
2.5
2
Chan. capacity sim; Broadcasting on trusted relays
1.5 Sec. capacity sim; Broadcasting with untrusted relays
Sec. capacity sim; Our proposed technique Sec. capacity sim; Normal Multicasting
1
50 100 150 200 250
Number of Relays K
2 2
=Δ = ρ = ρ = 10 dB
Figure 3.4 Simulatedρsecrecy capacity scaling: .hk,d . = .hs,k. = 1, and
ρs d k
the total number of the relays T increases. In fact, the higher the T is, the less the probability
that an eavesdropper receives a signal from s, and the better the secrecy performance
becomes.
Fig. 3.3 shows a comparison between the SOP upper bound for our technique and the OR
technique that was proposed in A. El-Malek et al. (2017) and Mabrouk et al. (2017), with
jamming, for different values of K and T . It is shown that the secrecy performance is
remarkably enhanced with adding more relays to the network. Also, a noticeable
improvement is shown in our proposed technique compared to the OR technique proposed in
A. El-Malek et al. (2017) and Mabrouk et al. (2017).
Δ ρ
broadcasting,
Fig. 3.4 showsand
the the conventional
secrecy multicasting
capacity scaling transmission,
comparison betweenfor s = ρdtechnique,
ourρproposed = ρk = 10the
dB
=
and T = 3K. Moreover, we show the performance for the case where all the relays are trusted
(the black curve). From Fig. 3.4, we can see that 1) our technique scaled similar to, even
slightly better than, the broadcasting technique, which gives an advantage to our proposed
technique, since the eavesdropper receives just a part of the message, whereas it receives all
the transmitted message when the broadcasting scenario is applied Kim et al. (2015). 2) A
security enhancement was achieved compared to the conventional multicasting technique.
3.7 Conclusions
In this paper, we proposed a new location-based multicasting cooperation strategy that takes
advantage of the locations of all the nodes to enhance the security. We provided an analytical
study for the SOP, and we showed that the secrecy capacity scaling, of the proposed
technique, converges to values similar to the broadcasting case. Moreover, it was shown that
the SOP is improved when the total number of relays T increases. Our results also displayed
remarkable security performance improvement, compared to the conventional multicasting
technique. As future work, we will further improve the current study by considering
aggressive eavesdroppers.
CHAPTER 4
DESIGN AND PERFORMANCE ANALYSIS OF SECURE MULTICASTING

COOPERATIVE PROTOCOL FOR WIRELESS SENSOR NETWORK
APPLICATIONS
1
Paper under revision in IEEE Wireless Communication Letters, 2019.
4.1 Abstract
This paper proposes a new security cooperative protocol, for dual phase amplify-and-forward
large wireless sensor networks. In such a network, a portion of the K relays can be potential
eavesdroppers. The source agrees to share with the destination a given channel state infor-
mation (CSI) of a source-trusted relay-destination link to encode the message. Then, in the
first hop, the source will use this CSI to map the right message to a certain sector while trans-
mitting fake messages to the other sectors using sectoral transmission. In the second hop, the
relays retransmit their received signals to the destination, using the distributed beamforming
technique. We derived the secrecy outage probability and demonstrated that the probability
of receiving the right encoded information by an untrustworthy relay is inversely proportional
to the number of sectors. We also showed that the aggressive behavior of the cooperating
untrusted relays is not effective compared to the case where each untrusted relay is trying to
intercept the transmitted message individually.
Keywords: Physical layer security, secrecy outage probability, amplify and forward, secrecy
capacity.
58
4.2 Introduction
In wireless networks, nodes can join and leave frequently, which increases the risk of the ma-
licious nodes that are penetrating the wireless network. Therefore, the demand for security
solutions in the physical layer is becoming more and more essential. One of the important
metrics that evaluate the security performance in the physical layer is the secrecy rate, which
is the difference between the channel capacity of the legitimate links and the channel
capacity of the illegitimate ones Gopala et al. (2008). Many techniques have been proposed
to achieve a positive secrecy rate, such as multi-antenna scenarios, beamforming, game
theory, power allocation schemes and cooperative jamming Wang et al. (2014b), Kuhestani
et al. (2018a), Kuhestani et al. (2016) and Atallah et al. (2015). A wireless network could
benefit from the new joining nodes, by using them as relays, or by treating them as potential
eavesdroppers. However, as shown in Kuhestani et al. (2016) and Kuhestani et al. (2018b),
taking advantage of these nodes and using them as relays could be more useful to the wireless
network, from a security perspective, than treating them as eavesdroppers. The authors in
Kim et al. (2015) studied the secrecy performance for the case of multiple passive untrusted
relays, where each passive untrusted relay is trying to intercept its received message
individually. In Atallah & Kaddoum (2016), the authors studied the secrecy capacity scaling
with aggressive untrusted relays. We define the aggressive behavior as when the untrusted
relays are cooperating between each other by sending their received messages to an external
wiretapper. Both Kim et al. (2015) and Atal- lah & Kaddoum (2016) considered two
transmission schemes, namely opportunistic relaying (OR) and distributed beamforming
(DBF). They also demonstrated that DBF outperforms OR technique from a secrecy
perspective. In Atallah & Kaddoum (2017), a new location-based multicasting technique was
proposed considering both passive and aggressive untrusted relays behaviors. It was shown
that this technique enhances the security compared to Kim et al. (2015) and Atallah &
Kaddoum (2016).
On the other hand, the randomness of the channel has been exploited for different purposes,
whether to enhance the reliability or to secure the communication system as it was used to
generate keys in Li et al. (2005). Therefore, in this paper, we combine the channel
randomness
with multicasting transmission to propose a new location-based multicasting protocol in two-
hops wireless sensor networks (WSN). The goal of this protocol is to increase the security of
these networks while taking into account that wireless sensor nodes have limited capabilities.
In the proposed protocol, the source and the destination share the channel state information
(CSI) to map the source’s transmission by sending the useful encoded message towards a
specific sector, while sending other fake messages, similar to the useful one, towards the
other sectors to confuse the eavesdroppers. Thus, we propose two strategies: the first one is to
prevent the eavesdropper from receiving the transmitted message all the time by
multicasting the signal to a different sector in each transmission time. Hence, for an
eavesdropper located in a certain sector, the probability that it would be in the right sector is
inversely proportional to
the number of sectors, p = 1/N 1. This eavesdropper can still know when there is a trans-
mission towards it and when there isn’t. Also, it can know to which sector this transmitted
signal is multicasted when this eavesdropper cooperates with other eavesdroppers located in
other sectors. Therefore, we came up with the second strategy which is based on sending fake
messages towards the other sectors to increase the entropy and the confusion, related to being
in the right sector, at the eavesdroppers. We provide analytical expressions for the secrecy
outage probability (SOP) of both passive and aggressive untrusted relays. Our numerical re-
sults show how our technique enhances the security performance and how immune it is
against the aggressive behavior of the untrusted relays. Finally, adopting such a security
protocol by allowing a part of the nodes to forward fake messages is promising because of
the availability of high number of cheap electronic sensors with limited computational
capabilities.
Consider a source s equipped with multi-sectoral antennas, K amplify-and-forward (AF)

cooperative relay sensor nodes with limited capabilities, and a destination d, provided with
sectoral antennas. Out of the K relays, there are U untrustworthy relays that could be
potential eavesdroppers. Each relay is equipped with a single antenna and works in a half-
duplex mode, as shown in Fig.3.1. It is assumed that there is no direct link between s and d,
i.e. all the transmit-
message xtr and
Figure 4.1 In the
the fake onesofxeach
1st hop i t=r ’s towards N sectors. In the 2nd
transmission, s multicasts the
useful hop, the K relays retransmit× their received messages towards d
ted information should be forwarded by the relays. To perform the proposed security method,
s and d should share the CSI knowledge of the source-trusted relay-destination link, which is
the kernel of our developed security method. This CSI is considered to be the main cause of
randomness and it is completely mapped into a vector V of digital values. It should be noted
that this security algorithm is implemented just before the communication process starts, and
it can be renewed at any time s and d agree on to keep refreshing the source of security and to
make it as strong as possible. Moreover, since the received signal-to-noise ratio (SNR) in an
AF two-hop wireless network over the first hop is higher than the received SNR after two
hops. Therefore, by considering the case where the eavesdropper is in the first hop, we are
studying the worst case security scenario. In the first hop, the source will encode the data
prior to the transmission by using the vector V . Then, s will use this vector again to map its
transmission
of the different messages xi’s towards N different sectors, where 1 ≤ i ≤ N , N ∈ N+. We will
denote the desired encoded signal by xtr , whereas the other signals xi×t=r are the fake ones
that
are transmitted over the other sectors. Without the knowledge of V , each untrusted relay e
will try to randomly guess the useful signal with a probability 1/N. Even if it succeeds in
guessing
and receiving the useful message, the untrusted relay would still need the vector V to decode
it. In the second phase, all the K relays will resend their received messages towards d using
the DBF technique. Since it has the same vector V , after removing the interference coming
from the fake messages by using self-interference cancellation (SIC), the destination will be
able to know from which sector the useful message is coming and decode it using V . The
received signal, at the kth relay, where 1 ≤ k ≤ K, is given by
√
yk = Pi hs,k xi + nk, (4.1)
.
where nk ∼ Nc 0, σ 2 is the complex additive white Gaussian noise (AWGN) at the kth relay,
Σ variance σ 2, P is the transmitted power from s towards the ith sector. We
with mean 0 and i
assumed that the channels are quasi-static block log-normal channels, i.e. the channel coef-
v
ficient hv,r ∼ lnN (μv, σ 2), where {v, r} ⊂ {{s, k},{k, d}}, is considered as constant during
the transmission time of one message, but it may change independently thereafter, the CSI is
known by the receiving nodes, and the noise variance N0 has the same value in the first and
the second phases. It is important to note that adopting such security solution by allowing a
part of the nodes to forward fake messages is feasible due to the availability of a high number
of electronic sensors with limited capabilities. Consequently, the received signal-to-noise
ratio (SNR), at a kth relay, is expressed as
Δ , . .= 2
ρ γk ρs hs,k (4.2)
will be χk== αk wk yk, where wk is the beamforming

, weight, and αk represents the normalized
where j N0, j ∈ {s, k, e}. In1 the second hop, the retransmitted message from the kth
Pj relay . The received useful messages at d will be written as
2
amplifying coefficient αk = . s |hs k| +
ρ
,
N0
M
∑ hm,d αm wm ym + nd, (4.3)
yd = m=1
d ∼ Nc (0, N0) is the complex AWGN at d. After optimizing the beamforming weights from
nwhere M is the number of the relays in the sector that receives the right message. 1 ≤ m ≤ M,
Kim et al. (2015) and the references therein, the SNR at the destination is obtained as
2
M ρs |hs,m|2 ρm hm,d
γd = .
∑
m=1 . .
M ρs | s, 2 .
m | + 2
+1
h hm,d
ρm
= ∑ γ m. (4.4)
m=1
Σ
The channel capacity at d will be 1
log(1 + , (4.5)
2 +
Σ
Cd = γd )
where [ξ ]+ denotes max{ξ, 0}.

4.3.1 Non Colluding Eavesdropping Relays
In this scenario, there are two different hypotheses H1 and H2 as follows :

Hypothesis
knows how H to1:recover
the untrusted relay is the
V and decode in the right sector with a probability p1 = 1/N and it
message.
Hypothesis H2: the untrusted relay is in a wrong sector, with a probability p0 = 1 − p1 =

1 − 1/N. Then, this relay will not impact the security and the channel capacity at the eaves-
dropper e will be equivalent to zero from a security point of view. Considering the aforemen-
tioned two hypotheses, the channel capacity at e will be expressed as
⎧
(4.6)
⎨
12
log(1 + γe) H1
Ce = ⎩ 0
H2,
2
where γe = ρs |hs,e| is the SNR of the useful message at e.
4.3.2 Colluding Eavesdropping Relays
Assuming aggressive untrusted relays, cooperating between each other and sending their mes-
sages towards an external wire-tapper A, the received useful signal at A will be written as
U1
yA = ∑ hu,A αu wu yu + nA, (4.7)
u=1
messages
where U1 xistr,the 1≤u≤
andnumber 1 ≤ U . Moreover, nA ∼ Nc (0, N0) is the complex AWGN at A.
of Uthe untrusted relays that are in the right sector and sending the
useful Hence, the SNR at A will become
U1 ρs |hs,u|2 ρu hu,A 2
γ . .
u=1U s s, u u,
.
= ∑
1
u A
ρ |γu.
A
=∑ u=1 h | + ρ .h
2 2
+1 (4.8)
We will define two hypotheses for A:

Hypothesis : A receives
H J decodes the right message with a probability p1 = 1/N and knows how to
recover V and the message.
1
Hypothesis Hj : the colluding relays are just in the wrong sectors, or A con not recover V ,
2
which means that A won’t have any impact on the security. Hence, the channel capacity at A
will be equivalent to
2 1
CA = (4.9)
⎩
⎨
log0(1 + γA ) H
1 j
⎧ H2j
,
We will define the worst security case as when e, (in the non colluding state), or A, (in the
colluding state), knows how to recover V and decode the message. Therefore, the channel
capacity at q, where q ∈ {e, A}, is given as
Σ
11
Cq = q) . (4.10)
. log(1
N 2
γ Σ+ +
From (4.5) and (4.10), the general secrecy capacity expression of the worst case is calculated
as
Σ
+ q
CS,q = C −Cq
Σ d 1 Σ
= Σ γ . 1 + Σ +. (4.11)
1log(1 + d ) − log γ
2 2N
4.4 Secrecy Outage Probability
Theorem 5. The secrecy outage probability expression of our proposed method CS,q, for both
passive and aggressive untrusted relays scenarios, is expressed as
Σ Σ . Σ
Pr C < R = Φ ..ln .22R 1 + eμ 1 −1Σ − μ Σ σ−1Σ
S
2 q N
3 d
, d
q
Φ.. . Σ μ Σ σ−1Σ
1 + N
+ ln 2 2R 1+eμ q q
−1 − d d
6 . √ Σ
( 3 )
σ
1
. Σ μ Σ σ−1Σ
Φ..
1 2R
− ln . 2
6 √ Σ
( 3σ
N
1+e − q
) −1 − d d . (4.12)
1
Proof. From (4.11), and for a threshold R, the SOP is defined as Kim et al. (2015)
2
Σ , Σ 2N . Σ
1 1
γ
Pr CS q < R = Pr Σγ log(1 + d ) − log 1 + γq < RΣ (4.13)
= Pr Σ d . ΣN
∞ < 2
2R
1 + γq 1 −1Σ
. Σ
d γ dγ . q
N
γ q q
q
=0
∫ Fγ
. Σ
.2
2R
1+γ 1 −1Σ f
Since γq and γd are following a log-normal distribution, (please refer to Appendix II for
the proof), then their probability density function (PDF) and cumulative distribution function
(CDF) are given as follows
1 (ln x−μ)2
−
fX (x; μ, σ )= √ e 2σ 2 , (4.14)
x σ 2π
ln x σ
−μ
FX (x; μ, σ ) = Φ. Σ, (4.15)
and Φ is the CDF of the standard normal distribution. Thus, the SOP in (4.13) is obtained as
Σ Σ
Pr CS,q < R = (4.16)
∞ ⎛ Σ ⎞
. . 1 (ln γq −μ q)2
∫ Σ
=
⎝ ⎠ √ d
Φ ⎟ γ q σ q γ2q.π
0 2 σq2
ln 22R σd − μd
⎜ N
−1 −
1+ e
γq
. Σ
Let β = ln γq , then γq = eβ , and d γq = eβ dβ . (4.17)
β is a normally distributed r.v. β ∼ N (μq, σ q2). Substituting (4.17) in (4.16), the secrecy
Σ Σ
outage probability Pr CS,q < R is written as
ψ(β )
xs
¸ ⎛
. 2R Σ Σ μ ⎞˛ 2
2. 1 + β −1 − d
∞ N
∫ ⎜ − (β−μq)
ln 1 e 2 σq2
e ⎟
Φ⎜ ⎟ q√ dβ. (4.18)
0 ⎝ σd ⎠ σ 2π
It is noticed that (4.18) denotes the expectation of ψ (β ). We will use Holtzman √

tool Holtz-
man (1992) to approximate E [ψ (β )] in terms of three points located at μq, μq + 3 σq and
√
μq − 3 σq as follows
Σ Σ
Pr CS,q < R = E [ψ (β )] =
−
3 . Σ 61 √ 6 √
2 1
ψ μq + ψ . μq + 3 σ q Σ − ψ . μq 3 σq Σ . (4.19)
Compensating ψ (β ) from (4.18) in (4.19) yields (4.12).
100
Secrecy Outage Probability
10−1
Sim, N=8
10−2
Theory, N=8 Sim, N=4
10Theory,
−3
N=4 Sim, N=2
Theory, N=2
10Sim,
−4
Jamming Tech. No security Tech.
Sim, Best Case, N=8
0 5 10 15 20
Eb/N0 [dB]
Figure 4.2 SOP with passive untrusted relays: R = 3 bps/Hz, M = 4,
σs = σk = 0.95 and μs = μk = 1
In this section, we demonstrate the validity of our derived results using MATLAB software.
accurately
Fig.4.2 showscharacterize
the SOP asthea simulation
function ofresults.
the SNR.It isItassumed thatthat
is noticed R =the
3 bps/Hz, = 4, σs =
derivedMexpressions
σk = 0.95 and μs = μk = 1. From Fig.4.2, we can see how the secrecy performance improves
when the number of sectors N is increased. For example, to keep the SOP level at 10−2,
the source has to increase the number of sectors N from 4 to 8, which will also reduces the
required SNR from 23dB to 17dB. Also, it is shown that our performed technique
outperforms the conventional jamming technique, where the destination jams the nodes while
the source is transmitting in the first hop. As we can see from Fig.4.2, the margin between the
worst and the
100
10-1
Sim, N=2, U1=3

Theory, N=2, U1=3 Sim, N=4, U1=3
10-2Theory, N=4, U1=3 Sim, N=4, U =1
10-3
1
Theory, N=4, U1=1
Sim, Jamming Tech, U1=3 Sim, N=8, U1=4
10-4Theory, N=8, U1=4
10-5
0 5 10 15 20
Eb/N0 [dB]
Figure 4.3 SOP with aggressive untrusted relays: R = 2 bps/Hz, M = 4,
σs = σk = 1.1 and μs = μk = 0.69

best case, when e does not know how to recover V , depends on e’s capability in recovering V
and decoding the message.
Fig.4.3 shows the SOP of our proposed technique for different values of N, when R = 2
bps/Hz, σs = σk = 1.1, and μs = μk = 0.69. It can be seen that the greater the number of
sectors, the better the secrecy performance. Moreover, we can see from Fig.4.3 that there
is not that
much of difference between the base of one and that of three aggressive untrusted relays. For
example, at SNR level of 18dB, the SOP just goes from 1.05 × 10−2 to 1.85 × 10−2 after
adding two extra aggressive untrusted relays, which means that our proposed technique is
immune towards adding more eavesdropping relays that are cooperating with each other.
Also, it is shown that the security performance is improved when our technique is applied
compared to the jamming technique. To evaluate the diversity order, we calculated the slope
at 20dB for the
following cases: when all the relays are trusted, the brown curve in Fig. 3.2, and when all the
relays are untrusted and aggressive, the blue curve in Fig. 3.4. For the first case, the slope is
1.1, whereas it becomes 0.7 for the second case.
4.6 Conclusions
In this paper, we proposed a new location-based multicasting protocol that is mapped by the
knowledge of a trusted link’s CSI in two-hops WSN. We provided an analytical study for the
SOP for the passive and the aggressive behaviors of the untrusted relays. The results showed
the immunity of our technique towards the untrusted relays aggressive behavior, and an im-
provement in the security compared to the conventional jamming technique.
CHAPTER 5
SECRECY ANALYSIS OF WIRELESS SENSOR NETWORK IN SMART GRID

WITH DESTINATION ASSISTED JAMMING
Michael Atallah1, Md. Sahabul Alam1, and Georges Kaddoum1
1
Paper published in IET communications, 2019.
5.1 Abstract
In this paper, we investigate the physical layer security performance over Rayleigh fading
channels in the presence of impulsive noise, as encountered, for instance, in smart grid en-
vironments. For this scheme, secrecy performance metrics are considered with and without
destination assisted jamming at the eavesdropper’s side. Specifically, we derive analytical ex-
pressions for the secrecy outage probability (SOP), at the legitimate receiver. Finally,
numerical results are provided to verify the accuracy of our derivations. From the obtained
results, it is verified that the SOP, without destination assisted jamming, is flooring at high
signal-to-noise- ratio values and that it can be significantly improved with the use of
jamming.
5.2 Introduction
Wireless Sensor Networks (WSN)’s are widely employed in oil, gas, and smart grid (SG)
mediums, since they have tremendously reduced the costs, increased the network coverage,
and reduced the deployment time Gungor et al. (2010); Akyildiz et al. (2002). For example,
in the context of designing a reliable smart grid, it is crucial to monitor and control the power
system parameters in the transmission and distribution segments as well as in substation de-
vices Fang et al. (2012). In order to allow such advanced functionalities and to avoid possible
disruptions in electric systems due to unexpected failures, a highly reliable, scalable, secure,
cost-effective, and robust communication network must be operational within the power grid
70
that convey data from monitoring sensors in the field to the access point. In this vein, the
most promising method of SG monitoring, explored in the literature, is based on WSN’s
Gungor et al. (2011). For such applications, the increase in the infiltration rate for smart
sensor networks has raised concerns regarding their security and privacy. Therefore,
creating a secure environment for communications, and guaranteeing the privacy of
customers, is becoming a significant challenge in SG environments. Since the infrastructures
tend to be highly diversi- fied, especially with the continuous deployment of small sensors
Baig & Amoudi (2013), the lower layers (physical and data link layer) are oblivious of any
security considerations. In this vein, to tackle the security issues, physical layer security
(PLS) was suggested as a potential solution Soosahabi & Naraghi-Pour (2012). Recently,
many security methods were studied in the PLS field, like game theory , multiple antenna
schemes , beamforming , cooperative jamming , and power allocation techniques Atallah et
al. (2015). Particularly, cooperative jamming strategies have been deemed efficient for
reliable secure transmission over wireless mediums Atallah et al. (2015); Atallah &
Kaddoum (2016, 2017); Liu et al. (2013).
5.3 Related Work
In the literature, multiple researches have been depicting the performance of cooperative jam-
ming strategies in the presence of additive white Gaussian noise (AWGN). However, the
noise characteristics usually observed in SG environments are remarkably non-Gaussian and
are in- herently impulsive Middleton (1977); Sarr et al. (2017); Neagu & Hamouda (2016).
For example, the noise, emitted from power equipments in a power substation, appeared to
be impulsive et al (2011). On the other hand, the performance of PLS techniques, in the
presence of impulsive noise, is not widely acknowledged. In Pittolo & Tonello (2013), the
secrecy rate was studied in narrowband power line communications (PLC) networks taking
correlated channels into consideration. Thereafter, Pittolo & Tonello (2014) evaluated the
security in PLC networks with multi-carrier and multi-user broadcast channels. Both Pittolo
& Tonello (2013) and Pit- tolo & Tonello (2014) showed that a higher secrecy rate could be
achieved when deploying wireless channels, rather than utilizing the PLC links. In Salem et
al. (2017), the researchers
proposed a PLC and wireless hybrid security scheme. However, the analysis in Salem et al.
(2017) is limited to AWGN only and the effect of impulsive noise is ignored. In Liu et al.
(2013), the authors studied two-hops wireless networks, with destination assisted jamming, in
the presence of an eavesdropper. In their work, both the destination and the transmitter are
jamming the eavesdropper. However, the practical case of impulsive noise, whether at the
legitimate receiver or at the eavesdropper’s side, has not been investigated.
To the best of the authors’ knowledge, no existing work has considered PLS in wireless sys-
tems, where the effect of the impulsive noise was involved. To fulfill this research gap, our
paper provides a mathematical framework to investigate the performance of PLS in the pres-
ence of impulsive noise. Here, we consider a Bernoulli-Gaussian (BG) noise model, to take
into account the impulsive behavior. This is motivated by the facts that the BG model is
tractable and can represent the amplitude distributions of real impulsive noise measurements
to a certain level of satisfaction Shongwey et al. (2014). We consider a single input single
output (SISO) communication network, which consists of a source, a destination, and a
passive eavesdropper. The passive eavesdropper is trying to intercept the transmitted message
between the source and the destination without interfering with the system. To study the
secrecy performance in this system, we consider two scenarios: 1) with destination assisted
jamming, where the destination is jamming the eavesdropper while the source is transmitting
its signal, and 2) without destination assisted jamming, which is the worst security case. In
this network, the wireless channels have a Rayleigh distribution and the noise is characterized
by a Bernoulli-Gaussian random process Ghosh (1996), to capture the combined effects of
the AWGN and the impulsive noise. Our main contributions summarized are as follows:
- We analyze the secrecy performance of the proposed network in smart grid scenarios, in
the presence of impulsive noise.
- We reformulate the secrecy capacity in an alternative approach aiming to make the deriva-
tion of the secrecy outage probability (SOP) tractable.
- We also analyze and compare the secrecy outage performance under two different scenar-
ios: i) with destination assisted jamming, where the destination assists by jamming the
passive eavesdropper while the source is transmitting its signal, and ii) without
considering destination assisted jamming.
- We provide accurate results, in terms of the achievable secrecy capacity and the SOP, for
these scenarios. Subsequently, numerical simulations are used to verify the accuracy of
our analytical expressions.
Figure 5.1 The source s transmits xs to the destination d, while the

eavesdropper e is trying to intercept xs. In the case where d is jamming, d is
provided with two independent antennas; (1) is for receiving xs. (2) is for
jamming with artificial noise signal xd
As shown in Fig. 5.1, a wireless communication system consists of a source s, a destination

d, and a passive eavesdropper e. In the demonstrated network, two scenarios can occur; in
the first one, the source is broadcasting its signal xs, and both the destination and the passive
eavesdropper are receiving it. The destination has two independent antennas; one is jamming
the passive eavesdropper, with an artificial noise signal xd, while the other one is receiving
the signal, coming from the source. In the second scenario, there is no destination assisted
jamming. Thus, the received signals, at the destination and at the eavesdropper’s sides, are
respectively expressed by
yd
√ √
j = hs,d Ps xs + hdinPd xd + nd , (5.1)
√ √
ye = hs,e Ps xs + hd,e Pd xd + ne , (5.2)
where Ps is the source transmit power and Pd is the artificial noise signal xd power. Also, hs,d
is the channel coefficient between the source and the destination and hdin is the channel co-
efficient between the jamming and the receiving antenna at the destination side. In addition,
nd and ne are respectively the noise terms at the destination and the eavesdropper, that cap-
ture the combined effects of AWGN and impulsive interferers. In a specific situation, where
no jamming is used at the eavesdropper’s side, Pd in (5.2) will be equal to zero. In our anal-
ysis, we assume that the channels are quasi-static block Rayleigh channels, i.e. the channel
coefficients hs,e, hs,d, hdin, and hd,e are considered as constant during the transmission time of
one message, but they may change independently thereafter. Accordingly, the channel gains
|hs,e|2, .hs,d .,222|hdin| , and .hd,e .follow independent exponential distributions. For this model,
the thermal noise component at node m, where m ∈ {e, d}, is considered complex Gaussian,
whereas the impulsive part is modeled as a Bernoulli-complex Gaussian random process
Ghosh (1996). Since it is a sum of two complex Gaussian random processes, nm qualifies as a
complex Gaussian noise and can be written as Dubey & Mallik (2015),
nm = nm0 +nm1, (5.3)
where nm0 is the AWGN component at node m, with zero mean and variance σm02 , and nm1 =
bm Am is the impulsive component. Moreover, Am is a complex white Gaussian noise, with zero
2
mean and variance σm1 , and bm is the Bernoulli process. The probability mass function of bm
is given by,
Pr(bm = 1) = pm1,
Pr(bm = 0) = pm0 = 1 − pm1, (5.4)

where pm1 and pm0 denote the probabilities of occurrence of the impulsive and the thermal
noise at node m, respectively. Thus, the noise variance of nm can be written as
Nm = σ 2 + bm σ 2 . (5.5)
0 m
At the destination side, due to the large power difference between xs and x d, .i.e. xd xs, the
destination will be able to use successive interference cancellation (SIC) to remove xd.
Hence, after using SIC, the signal at the destination will be given by
√
yd = hs,d Psxs + nd. (5.6)
Then, the received signal-to-interference-plus-noise-ratio (SINR), at node m, can be expressed
as
Ps |hs,m2|
γm =
Pd .h d,m .2 + N
m
2
Ps |
= 2 2 2
hs,m| + bm σm1
2
Pd .hd,m. + σm0
Ps | , (5.7)
= | 2
hs,m 2
Pd .hd,m. + σm0 (1 + bm Γm)

where σ2
, the SINR becomes
m0 . By dividing the nominator and the denominator by
= m1
Γm σ m0 γ
m
2
2 σ
γm0
γm = , (5.8)
γ
j +1 + bm Γm
. .P 2
where h
2
d d,e , and Ps | .
2
e0 σ hs,m|
γ j= γm0 σ2
m0
=
Thus, from (5.8), the channel capacity at node m becomes
Cm = log2 (1 + γm)
γm0
.1 + γ j +1 + bm Γm
= log2
Σ. (5.10)
Finally, the achievable secrecy capacity would be given by
Cs = [Cd −Ce]+
Σ . d0 Σ . γe0
1 + γ 1+
= log2 1 + bd − γ j +1 + be ΣΣ+ , (5.11)
Γd log2 Γe
+
where
the secrecy outage probability equations to study the impactsection,
[a] = max(a, 0). As a consequence, in the following we detail
of impulsive noisetheonderivations
the PLS. of
The analysis considers both cases: with and without destination assisted jamming.
5.5 Secrecy Outage Probability Analysis
It is important to mention that a secrecy outage event happens when the target secrecy R is
greater than the achievable secrecy capacity Cs, i.e., Cs < R. For the two scenarios, the analyt-
ical expressions of the SOP can be found as
Pr[Cs < R]= Pr[log2 (1 + γd ) − log2 (1 + γe) < R] . (5.12)
5.5.1 Secrecy Outage Probability Analysis with Jamming
In this subsection, we study the SOP after applying destination assisted jamming.
Theorem 6. With jamming, the secrecy outage probability in Rayleigh fading channels, in the
presence of impulsive noise, is given by
1 1 . Σ
p p a exp (a ) (a )+ γ γ
∑ E
k=0 1 2 1 3 d0 j
di ∑ , (5.13)
γd0γj
i=0
Pr[Cs < R]=
where a1,a2, and a3 are given in (5.19), (5.20), and (5.22), respectively. Also, E1 (x)
∫ ∞isexp(−tx)
the ex-
ponential integral function of the random variable (r.v.) x and defined as E1 (x)=
dt.
1 t
Proof. From (5.11) and (5.12), the SOP is demonstrated as

Σ
Σ +bγdd0Γd
Pr[Cs < R] = Pr 1+1
1 + γe γ <
d0
= Pr Σ.1 +
R
1 + bd Γd Σ < 2 (1 + e)Σ
e ΣΣ
γ
γd0
) < .1 +
= 1 − Pr Σ2R (1 + γ 1 + bd Γd
γd0
= 1 Σ Pr γ.−< 1 + Σ 2−R −1Σ
e
1 + bd Γd Σ
.. γd 1 1Σ
1 1 p ∫∞ 1 γ d (5.14)
0 f
− +
1+bdi Γd d0,
di Fϒe
= ∑ 0 R
2− ϒd0( d0)
i=0
where Fϒe(γe) is the cumulative distribution function (CDF) of γe, which is given as
γ j +1 + be Γe
Fϒe(γe) = Pr(ϒe < γeγ
)
e0
< γeΣ
.= Pr. . ΣΣ
= Pr γe0 < γe γ j +1 + be Γe
1 ∫∞ .
= ∑ pek Fϒe e
0
0 γ .
γ j +1 + bek Γe
k=0
ΣΣ ϒ j .
f γ
jΣ
d
j
γ
∫∞
. Σ ΣΣ
1 . exp .− γ j Σ
= γ +1 +
∑ 1− . e j bek Γe γ
exp γ γe0 γ jj
d
⎞ γ j
k=0 ⎛ γe0 exp .− γe (1+bek Γe ) Σ − γ j γe −γe0
0
1 p γe
= ∑ ek
⎝ ⎠ . (5.15)
−
k=0 γe γ j +
γe0
Since γm0 follows an exponential distribution, the probability density function (PDF) and the
CDF of γm0 are given by Alouini & Goldsmith (1999):
,
e−γm0 γm0 (5.16)
γm0
fϒm0 (γm0)=
,
γ
Fϒm0 γm0m
(γm0 ) = 1 − exp .− 0 Σ, (5.17)
where γm0 is the mean of γm0. Substituting (5.15) and (5.16) in (5.14), the SOP with jamming
can be written as ⎛
∫∞ − γ
1 exp . γ d0 Σ 1
− ⎝
Pr[Cs < R]= 1 ∑ pdi ∑ pek
γd0d0 ⎞⎞
⎛ i=0 Γ )0
γe0 exp .− γe (1+b ek e
Σ − γ j γe −γe0
k=0 γd0
γe
× ⎝− ⎠⎠ d
γe γ j +
1
1
γe0
1
= . Σ
∑ di k=0
γd0γ j i=0 p ∑ ek
p a1 exp (a 2 ) E 1 (a 3 )+ γ γ
d0 j , (5.18)
where a1 = −γe0 2R (1 + bdi Γd ) , (5.19)

z2 + bdi Γd γ j + bek Γe γd0 + γd0 + γ j
a2 = d0 j , (5.20)
. γγ
z2 = − γ j − γ e0 (1 + bdi Γd ) 2R, (5.21)
z3 + z4 − γ d0 γ j (1 + bek Γe)
a3 = Σ , (5.22)
j e0 d0
γγγ
. . Σ Σ
γ
z3 = e0 (1+ Γ )
bdi d e0 γ − γ j 4
R γ γ
+ d0 j (1 + Γ
bek e ) (5.23)
,2R
. Σ
z4 = bek Γe γd0 + bdi Γd γ j + γd0 + γ j γe0. (5.24)
5.5.2 Secrecy Outage Probability Analysis without Jamming
Here, we consider the worst security case, without destination assisted jamming, to measure
the security performance of our proposed scenario.
Theorem 7. Without jamming, the secrecy outage probability in Rayleigh fading channels, in
the presence of impulsive noise, is given by
1 1
Pr[Cs < R]= ∑ pdi ∑ pek
i=0 k=0
⎛ . Σ
1−2R(1+bdi Γd ) Γe) γd0 ⎞
exp γd0 (1 + e
k
× b1 − . (5.25)
⎝ (1 + bek Γe) γd0 + 2R (1 + bdi Γd ) γe0 ⎠
Proof. Without jamming, Pd = 0, then γ j in (5.9) is equal to zero, and (5.11) is further simpli-
fied. Revisiting (5.12), the SOP, under this condition, becomes
+bd Γd
Pr[Cs < R] = Pr Σlog2 . 1 γ Σ< Σ
1 +1 + γd0e R
Σ . Σ Σ
= Pr γd0 < 2R (1 + γ e ) − 1 (1+ bd Γd ) . (5.26)
By substituting (5.16) and (5.17) in (5.26), we can write the SOP as
1 1 ∫
Pr R ∞ γ
z f γ d (5.27)
]= ∑
pdi ∑p
ek Fϒd0 ( 1)
[Cs < i=0 k=0 0 ϒe0 ( e0) e0
1 1 ∫∞ . . Σ
. ΣΣ exp −e0 γe0 γ
1
1− −z d e0 (5.28)
= ∑ pdi ∑ pek γ γ γ
exp d0 e0
i=0 k=0
⎛
1 10 γd0 e
. Σ
p p 1−2R(1+bdi Γd ) k Γe) γd0 ⎞
exp (1 + b
=∑ di ∑
ek 1− (5.29)
i=0 k=0 ⎝ (1 + bek Γe) γd0 + 2R (1 + bdi Γd ) γe0 ⎠
where z1 in (5.27) becomes
γe0
z.= .21
R
1+
1 + bek Γe
Σ − 1Σ(1 + bdi Γd ). (5.30)
Although this paper considers the secrecy analysis for the case of memoryless impulsive
noise, modeled by a Bernoulli-Gaussian process, the analysis could be easily extended to
consider the presence of any kind of impulsive or Gaussian mixture noise.
Sim, Pd1=0, Pe1=0

0.85 Theory, Pd1=0, Pe1=0 Sim, Pd1=0, Pe1=0 .01
Theory, Pd1=0, Pe1=0 .01 Sim, Pd1=0.05, Pe1=0
Theory, Pd1=0.05, Pe1=0

0.8
0.75
0.7
0 5 10 15 20 25 30 35 40 45
Eb/N0 [dB]
system without jamming: γd0 = γe0, Γd = Γe = 1000, Figure 5.2

Analytical and simulated SOP performances of the
proposed
.hs,d .2 = |hs,e|2 = .hd,e.2 = 1, and R = 1bps/Hz

100
10-1
Sim, Pd1=0, Pe1=0

Theory, Pd1=0, Pe1=0 Sim, Pd1=0.05, Pe1=0.05
10-2 Pd1=0.05, Pe1=0.05 Sim, Pd1=0.1, Pe1=0.1
Theory,
Theory, Pd1=0.1,Pe1=0.1
10-3
0 5 10 15 20 25 30 35 40 45
Eb/N0 [dB]
system with jamming: h = |hs,e| = hd,e = 1, Γd = Γe = 100, d0 = e0, γ

Figure 5.3 Analytical. and.s,d
2simulated
1 γ 2 SOP performances
.= 1bps/Hz
.2 of the proposed
j = d0 , and R γ
γ
2
In this section, we present the SOP performances of SG networks, with and without
destination assisted jamming. In the simulation results, performed using MATLAB software,
it is assumed that the probability of having impulsive noise Pm1, ranges from 0.01 to 0.1, and
Γm takes the values 100 and 1000. These values are chosen to represent the characteristics of
the impulsive noise, as observed in SG environments Middleton (1977). On the other hand, it
is also assumed
. . 2 2 . .2
that the threshold R = 1bps/Hz, Eb/N0 = γd0 = γe0, γ j = 1 γd0, hs,d = |hs,e| = hd,e 2 = 1,
d0 σ= e0
and the background noise σ 2 2
= 1.
Fig.5.2 shows the analytical results derived in (5.25), and simulated SOP performances with-
out destination assisted jamming. From Fig.5.2, it is seen that the analytical results perfectly
match the simulations. Moreover, we can observe that the security performance is directly
proportional to the impulsive noise at the eavesdropper, and is inversely proportional to the
impulsive noise at the destination. We also exposed that when Eb/N0 > 25dB, the security
is flooring; due to the effect of the noise becoming negligible, and therefore, the SOP scales
towards a constant value.
Fig.5.3 shows the analytical results given in (5.13), and simulated SOP performances when
considering destination assisted jamming. This figure further confirms the correctness of the
analysis through the simulations. From Fig.5.3, We can see the degradation in the security
performance due to the presence of the impulsive noise. Subsequently, compared to Fig.5.2,
it is also observed that by adding jamming, the security performance is remarkably enhanced.
This confirms that the destination assisted jamming technique can be beneficial to SG
environments.
5.7 Conclusions
We have presented the SOP expressions, in wireless SG environments, with and without des-
tination assisted jamming. Our analytical expressions allow the measurement of the security
performance; when either both of the destination and the wiretapper or any of them is
affected by impulsive noise. From the obtained results, we verified that the analytical results
agree with the simulations. We also showed that the achievable security is enhanced when the
occurrence probability of impulsive noise is higher at the eavesdropper’s side than at the
legitimate receiver. Moreover, the results show that destination assisted jamming can
significantly enhance the security of the network, making it a promising security solution for
SG networks. In this paper, we didn’t consider any aggressive action by the eavesdropper,
which would be an interesting topic for future works.
CONCLUSION AND RECOMMENDATIONS
This thesis concentrated on the security in the physical layer in wireless networks. The en-
vironment’s characteristics were exploited to provide secure protocols in wireless systems. A
jamming technique was applied in most of the works, whether to analyse its presence, to add
more security to the studied systems, or to compare its performance with our novel protocols
to show the superiority of our protocols against the conventional jamming technique. The do-
main was exploited to design spatial transmission protocols. Rayleigh and log-normal fadings
were considered. Also, the presence of impulsive noise has been investigated. The proposed
protocols showed their strength against the eavesdroppers’ passive and colluding behaviors.
Finally, a novel approach to study the security performance in impulsive noise environments
is proposed and analysed. Hence, we investigated the secrecy capacity scaling with the
presence of colluding eavesdroppers in large wireless networks. To reduce the probability of
having access to the whole message by the eavesdroppers, we proposed new location based
protocols for that purpose. Our techniques were liberated from the need to know the CSI or
the location of the eavesdroppers. Also, by exploiting the presence of having many trusted
nodes in large wireless networks, we proposed a novel key generating and mapping
technique. In addition, we provided new secrecy capacity expressions for the environments
that are affected by the impulsive noise. Thus, from the second chapter, we showed how the
secrecy performance was enhanced after applying cooperative jamming technique, and how
the distributed beamforming overcomes the opportunistic relaying technique for large
number of relays. However, opportunistic relaying can still be used for small number of
relays and its performance could be enhanced by increasing the transmitted power. In the
third chapter, a novel location based protocol was proposed to reduce the probability that an
eavesdropper will receive the whole transmitted message. Even though the multicasting
technique is applied in each transmission time, the secrecy capacity scaling was not just
similar, but also slightly better than the broadcasting technique applied in the second
chapter. We applied cooperative jamming technique
along with our proposed protocol to boost the security. The results showed how increasing
the number of the sectors or the number of the relays could remarkably enhance the security
performance. Our proposed protocol could be exploited to optimize the selection strategy of
the sectors to avoid choosing the weak clusters and to focus on the stronger ones. Since the
passive behavior of the eavesdroppers was considered in this chapter, there was a need to in-
vestigate in the colluding behavior to provide the most secure protocol for wireless networks.
Thus, we proposed a new key mapping technique, in the fourth chapter, that was proved to
be immune towards the colluding behavior of the untrusted eavesdroppers. This protocol also
exploited the presence of the large number of trusted relays to generate keys to encode the
transmitted signals and to map the transmission towards different clusters. By applying this
protocol, the destination and the relays can benefit from the transmitted power in all the
sectors to harvest the energy. Here, the destination didn’t need to drain its energy by jamming
the relays. That’s due to the fact that our proposed protocol is strong enough to be operated
by itself and to achieve good security results. Furthermore, This protocol is promising due to
the availability of high number of cheap electronic sensor nodes. In another topic, since the
smart grid environment suffers from the impulsive noise, we provided in the fifth chapter the
secrecy analyses for two cases: before and after applying the destination assisted jamming
technique, with the presence of impulsive noise. Besides reformulating the secrecy capacity
equations, we showed in our results that the impulsive noise has an impact on the security
since it’s affect- ing both the destination and the eavesdropper, which means that
implementing good impulsive noise receivers could be taken into consideration to enhance
the security performance. Also, the results showed how the security was enhanced after
applying destination assisted jamming technique, especially for the systems that are
transmitting with high power. It should be noted that the simulations in this thesis were
performed using MATLAB software.
85
6.1 Future Work
As extensions to the current works, the imperfect CSI could be considered in the future
studies to cover more general cases. The full duplex mode could be investigated, whether at
the transmitter, the receiver or at the relays to see its impact on the security with our
proposed security protocols. Moreover, analysing and comparing the cooperation strategies
when the relays are decoding and forwarding or compressing and forwarding the information
could also be considered, especially since these relays will have different capabilities and
permissions on the access to the information. The aggressive behaviour of the eavesdroppers
when they cooperate together to intercept the transmitted messages could be analysed to
show its effect on some of the proposed protocols. Our proposed security techniques, that
rely on the space diversity, could be optimized by checking which sectors are the best to be
chosen and which sectors should be avoided whether from the power or the security
perspective. Multi-antennas techniques could be applied and investigated at both the
legitimate and illegitimate nodes. The case where the eavesdroppers are attacking the
network by jamming the legitimate nodes could also be investigated whether to find
alternative ways to route the transmitted information or other solutions to reduce the attack
effect on the legitimate nodes. Moreover, the impulsive noise environments could benefit
from a deeper investigation and analysis in the context of several physical layer security
techniques. Deep or reinforcement learning could be implemented to detect jamming
behavior. Additionally, more research efforts need to be focused towards exploiting relay
positioning and cross layered scenarios, which this latter could be used to gain more benefits
from secure cooperative schemes.
88
1. Proof of Lemma1
. Σ
1. Here, we will prove that A = ∑Kk=1 min γs,k , 2 γk,d has the PDF given in (3.21)
.
Lemma 2. Let Y = min γs,k, 2 γk,d ’s, then Y is an exponentially distributed r.v. with a
rate parameter λ0 = 1 2 Σ+ 1 2 .
ρ σ1 2ρ σ2
Proof. Since γs,k and 2 γk,d are exponentially distributed r.v.’s, their PDF and CDF will
be respectively given by the following equations Alouini & Goldsmith (1999)
fϒ ( γ )= λe −λγ , (A I-1)
Fϒ ( γ )= 1 − exp (−λ γ ) , (A I-2)

where λ is the rate of the exponentially distributed r.v. γ . Hence, to prove that Y has an
exponential distribution, we will first find the complement of the CDF of Y as follows; for
some v > 0,
. Σ
Pr (Y > v) = Pr min{γs,k , γk,d } > v
. Σ
= Pr γs,k > v, γk,d > v
. Σ. Σ
= 1 − Fγs,k (v) 1 − Fγk,d (v)
−λ1 v
=e e 2
λ2 v
−
= 2
. (A I-3)
λ2
−v.λ1 + Σ
e
Then, the CDF of y will be calculated as
λ2
F v 1 Pr Y −v.λ1 + Σ
1 e
v
Y ( )= − ( > )= − 2
,
where λ1 and λ2 are defined in (3.23). We can see from (A I-3) that the CDF of Y has
exactly the form of the CDF of an exponentially distributed r.v. given in (A I-2), with a
rate parameter of λ0 = λ1 + λ2 = 1 2 + 1 2 .
2 ρ σ1 2ρ σ2
Since A is a sum of K exponential r.v.’s, it has a Gamma distribution with a shape parameter
K and a rate parameter λ0, and its PDF is given by (3.21).
2. Since γs,e γd,e follow exponential distributions, the forms of their CDF and PDF are re-
γs,e is expressed
spectively given in (A I-1) and (A I-2). Thus, the CDF of a r.v. Ze = γ d
as
Σ . Σ Σ
Pr [Z < u] = Pr γs,e < γd,e +1 u
∫∞
.. Σ Σ . Σ
= Fϒs,e γd,e +1 u fϒ γd,e d γd,e
0 d,e
∫∞
. . . ΣΣΣ . γ Σ
= 1 − exp − λ1 u γd,e +1 λ2 e− λ1 d,e d γd,e
0 λ2
= 1 exp ( λ u) . (A I-4)
− −1
λ1 u + λ2
where λ1 and λ2 are defined in (3.23).
Since B = 1 + KTZ, the CDF of B is calculated as
K
FB (b) = Pr Σ1 + T Z < bΣ
T
= Pr ΣZ < (b− 1) K Σ
T λ2
= 1 − exp Σ−λ1 (b− 1) Σ . (A I-5)
K
Thus, we completed the proof.
K λ1 (b− 1) T + λ2
90
We will prove that γq and γd are following a log-normal distribution. First, we will define the
SNR γi, j as follows
2
γi, j ρi hi, j (A II-1)
. .=
where i ∈ {s, m} and j ∈ {m, e,
d}.
,
Lemma 3. Let X ∼ ln N (μ, σ ), then aX ∼ ln N (μ + ln a, σ 2), and X a ∼ ln N (aμ, a2 σ 2),
2
a ∈ R.
2 2
From Lemma 3, where a = 2, the channel gain .hi, j. ∼ ln N , γi, ). By using the prop-
j
(2μγi, j
,
erties in Lemma 3, we find that γi, j ∼ ln N (μγ i,
j γ2i,
), where μγi, j = 2 μi + ln(ρi), ln ( ρ i )=
j , 2
γi,
ln(Pi) − ln(N0), and σ = 4 σi2. Hence, γe ∼ ln N γs,e ).
2 (μγs,e
Now, we will find the distribution of γm (4.4) with the following approximation for high
SNRs, as follows
γ γ 1 1 (A II-2)
γs,m γm,d ≈ s,m m,d = = ,
γm = γs +γ +1 γ +γ 1 1 z
,m m,d +
s,m m,d γs,m γm,d
and z2= 1 .
s,m
where z = z1 + z2, z1 = γ m,d
Lemma 4. Let X j ∼ ln N (μ j , σ j2) are independent log-normally distributed variables with

varying σ and μ parameters, and Y = ∑nj=1 X j . Then the distribution of Y has no closed form
expression, but can be reasonably approximated by another log-normal distribution Z with
parametersFenton (1960)
APPENDIX II
PROOFS FOR CHAPTER 3

/2
μ = lnΣ∑ eμj+σ 2 Σ − σ2 , (A II-3)
j Z
Z 2
j j
Z = ln Σ 2μ μ+σ+σ 2/2σ
+ 1Σ. (A II-4)
2 ∑ e j j 2j (e) 22
(∑ −1)
σ
2 2
Form Lemma 3, where a = −1, we find that Z1 ∼ ln N , γs, ) and Z2 ∼ ln N , γm,d ).
m
(−μγs,m (−μγm,d
z
Also, from Lemma 4, Z ∼ ln N (μz, σ 2), where
.. . Σ Σ Σ
σz2 = ln exp σ 2 z1 − 1 /2 + 1 ,
. z zΣ
μz = ln(2 exp (μz ))
1
+ 0.5 σ 2 1−σ 2 .
Thus, from Lemma 3 and (A II-2), we get γm ∼ ln N (μγm , σγ 2 ), where a = −1, μmγ = − μ z ,
and σ 2 = σ 2. From (4.4), since γ is a sum of many γm, we will
m again use Lemma 4 to find
γm z d
that γd ∼ ln N (μγ , σ 2 ), where
d γd
.. . γΣ Σ Σ
σd2 = ln exp σ 2 m − 1 /M + 1 ,
. γ dΣ
μd = ln(M exp (μγm )) + 0.5 σ 2m −σ 2 .
Since the expressions of γu and γA in (4.8) are similar to γm and γd respectively, by following
m γ A γ
the same steps, we show that γu ∼ ln N , σm 2 ) and γA ∼ ln N , σA 2 ) where
(μγ (μγ
.. . γΣ Σ Σ
σA2 = ln exp σ 2 m − 1 /U1 + 1 ,
γ A
. m Σ
μA = ln(U1 exp (μγm )) + 0.5 σ 2 −σ 2 .
SUBMITTED PAPERS
APPENDIX III
A SURVEY ON COOPERATIVE JAMMING APPLIED TO PHYSICAL LAYER

SECURITY
Michael Atallah1, Georges Kaddoum1, and Long Kong1
1
Paper published in IEEE International Conference on Ubiquitous Wireless Broadband

(ICUWB) , November 2015.
1. Abstract
Security has always played a critical role in wireless cooperative communication systems de-
sign. Eavesdropping and jamming are two common threats to the information security in
wireless networks. However, jamming can be used in a cooperative manner to enable a
secure communication link between the legitimate transmitter and the intended receiver. This
paper presents a comprehensive survey on different jamming methods used to enhance the
physical layer security. This survey outlines first the underlying concept and challenges with
respect to security in wireless network design followed by a comprehensive literature review
and analysis of jamming techniques with their applications in this field. For each jamming
protocol, the paper categorizes different techniques within the existing literature by
elaborating on their application, and corresponding performances.
Keywords: Physical layer security, cooperative jamming, beamforming, power allocation,

artificial noise, multiple antennas, MIMO, game theory.
2. Introduction
Wireless communications is playing an integral part in our lives and also has a significant
social impact. Privacy and confidentiality with respect to the transmitted information over
the wireless medium becomes vital, especially for applications concerning medical informa-
94
tion, e-banking, and e-commerce. However, wireless communications are often vulnerable to
eavesdropping and signal interception Hong et al. (2013). Many security tasks are involved
in wireless networks design, like integrity and confidentiality checks, authentication,
spectrum access control Lou & Ren (2009); Shiu et al. (2011). Confidentiality refers to the
prevention of unauthorized information disclosure. Integrity ensures that the transmitted
information is utilized and modified by the legitimate user. Authentication refers to the
identity confirmation of different terminals. Spectrum access control refers to prevention of
denial-of-service type attacks. Conventionally, these security tasks are addressed mostly in
the upper layers of the network protocol stack using cryptographic encryption and decryption
methods. When employing symmetric-key cryptosystems, the two users have to share a
common private key to encrypt and decrypt the confidential message Hong et al. (2013).
However, for the secret keys sharing, this requires a secure channel or protocol. The
difficulties in secret key distribution and management Schneier (1998) lead to security
vulnerabilities in wireless systems. Alternatively, public-key cryptosystems allow the use of a
public key for encryption and a separate private key for decryption. The public key is
available to all users whereas the private key is known only to the receiver. However, the
cryptographic methods rely on the computational hardness on decrypting the message to
achieve security when the secret key is not available. As the computational power increases,
e.g., with the development of quantum computers, the computational hardness of certain
mathematical problems, for which the encryption and decryption are based on, may no longer
hold, causing many current cryptosystems to break downHong et al. (2013). Many coding
and signal processing techniques in the physical layer have been developed in the recent
years, to support and to further enhance security in wireless systems, many researchers have
made contributions to find alternative security solutions to fit the requirements of the current
and emerging wireless networks Goel & Negi (2008); Gopala et al. (2008); Shannon (1949);
Bloch & Barros (2011). Even though the fast channel variations and the wireless medium’s
broadcast nature may cause additional challenges, physical layer security technique exploits
the properties of the wireless transmissions to secure the communication channel in a better
wayHong et al. (2013).
Interference, in general, is regarded as undesired phenomenon in wireless communications.
But in secure communications, interference can benefit the system if it is used in a proper
way. The idea is to create an interference and put the eavesdropper in a disadvantage
comparing to the legitimate nodes Park et al. (2013). Several applications use interference to
increase the security in the physical layer, one of the security applications that has become a
very common and promising technique in the physical security field is the cooperative
jamming which is accomplished by the friendly terminals in which one of the legitimate
parties sacrifices his entire rate to jam the eavesdropper.
In this paper a continuation and update of the recent achievements in the field of physical
layer security is presented with emphasis on different jamming methods and protocols of
such schemes. Hence, our contribution can be summarized as follows:
1. Providing a brief overview of physical layer system model and the challenges in this field.
2. Developing a literature review of the different jamming techniques within the existing
recent literature with their advantages and disadvantages, followed by a discussion on
their subsequent application.
The remainder of this paper is outlined as follows. The concept of physical layer security is
depicted in Section 3. Cooperative Jamming and techniques to enhance physical layer
security via cooperative jamming are presented in section 4. Finally, the concluding remarks
are given in section 5.
3. Physical Layer security and cooperative jamming
3.1 Physical Layer Security
As shown in Fig. III-1, a generic wireless communication network model which consists of
three nodes is taken into consideration: the legitimate transmitter (Alice), the intended
receiver (Bob) and the eavesdropper (Eve). The link between Alice and Bob is called the
main channel,
while the link between Alice and Eve is named as a wiretap channel. This model exemplifies
the specific features of most multi-user secure communication systems. The secrecy capacity
is defined as the maximum achievable secrecy rate. In Bloch et al. (2008), the secrecy
capacity over additive white Gaussian noise (AWGN) channel Cs,A and Rayleigh fading
channel Cs,R are given by
Alice Bob
Eve
Figure-A III-1 Wireless wiretap system model
Σ .
1 Σ
P
Cs,A = 1+ .1 + ΣΣ+,
m2 w2 (A III-1)
2 1 σP
σ
+
Cs,R = σm2 Σ − 2 P|hw|2 ΣΣ
Σlog2 P| log2 σw2 , (A III-2)
2 −
.1 + h |
log2 .1 +
m
where P is the transmitted power, σm and σw are the noise power of the main channel and
wiretap channel. hm and hw are the Rayleigh fading coefficients of main channel and wiretap
+
channel respectively. [x] = max {0, x}. Also, the received signal-to-noise ratio (SNRs) at Bob
and Eve are defined as = P|hm|2 and γ w = P|hw|2 , respectively.
γm σm σw
In Fig. III-2, an average secrecy capacity of Rayleigh fading channel is compared (equation
(A III-2)) with that of Gaussian wiretap channel (equation (A III-1)). Strikingly, one can
observe
that the secrecy capacity over Rayleigh fading channels is higher than that of an AWGN
channel, in other words, we can use the fading property of the physical layer channel to
decrease the SNR of wiretap channel. Besides using the fading characteristics of wireless
channel, many other methods are applied to improve the secrecy performance of the wireless
communication systems. All the existing physical layer security methods in Shiu et al. (2011)
are classified into five major approaches: theoretical secrecy capacity, multiple-input-
multiple-output (MIMO) channel, coding schemes (channel coding and network coding),
power allocation, and signal design (artificial noise). Additionally, cooperative relay Han et
al. (2015); Wang et al. (2013a); Chen et al. (2013), cooperative jamming Ibrahim et al.
(2015) and energy harvesting Xing et al. (2014) are other useful methods. Among the
aforementioned methods, cooperative jamming is a promising technique and has attracted
significant attention. It was originally proposed for a multiple access wiretap channel, where
multiple legitimate users wish to have simultaneous secure communications with an intended
receiver in the presence of an eavesdropper.
3.2 Cooperative jamming
Cooperative jamming is a special technique where artificial noise is introduced by a helpful

interferer to confuse the eavesdropper.
In the following section, we will introduce the cooperative jamming techniques which are
used to enhance the physical layer security. To improve the secrecy capacity, we should either
increase the legitimate receiver’s SNR or decrease the eavesdropper’s SNR. A natural
approach by which to achieve the latter (decreasing the eavesdropper’s SNR) is to introduce
interferers into the system.
3.3 Artificial Jamming Signals types
Cooperative jamming depends on creating the interference at the eavesdropper’s side, many
artificial jamming signals are used and could be divided into four categories Long et al.
(2014):
0.9
Av. w = 0dB
0.8 Av. w = 10dB
0.7 Av.  = 20dB

s
Av. secrecy capacity C
0.6
0.5
0.4
0.3
0.2
0.1
0
−10 −5 0 5 10 15 20 25 30
Av.m (dB)
Figure-A III-2 Normalized average secrecy capacity versus γ m , for

selected values of γ w . The thicker lines correspond to the normalized
average secrecy rate capacity of Rayleigh fading channel while the
thinner lines correspond to the secrecy capacity of a Gaussian wiretap
channel
1. Gaussian noise: which is the same as the additive noise at the receiver Bassily & Ulukus
(2013); He & Yener (2013); Zhang et al. (2011).
2. Jamming signals which are priory known at legitimate receivers, which has an impact
only on the eavesdropper’s performance. This type of signals is better than the previous
one because the jamming signals don’t affect the legitimate receiver Long et al. (2013);
Dong et al. (2011).
3. Random codewords of a public codebook which is known by all the nodes including the
eavesdroppers, so the legitimate receiver has the ability to decode and cancel the
jamming signals, even though it requires a complicated self-interference cancellation
receiver to decode the codewords Pierrot & Bloch (2011).
4. Useful signals for the other legitimate nodes; like the downlink and the uplink of the
neighbouring cells Popovski & Simeone (2009), signals of multiple simultaneous source-
destination pairs Sheikholeslami et al. (2012), or signals of the invited cognitive ratio
users Stanojev & Yener (2011) and Stanojev & Yener. (2011), this type is difficult to
apply because of the change of the multiple transmission pairs.
Many applications are used in conjunction with the cooperative jamming strategy to enhance
the performance and increase the security, these include the usage of multiple antennas,
beamforming, game theory, and power allocation methods.
4. Application of cooperative jamming
4.1 Cooperative Jamming with Multiple Antennas and Beamforming
Many works apply multiple antennas method with cooperative jamming technique to enhance
the physical layer securityYang et al. (2013); Zhang et al. (2015); Li et al. (2014a); Wang
et al. (2014b); Banawan & Ulukus (2014); Xing et al. (2014); Vishwakarma & Chockalingam
(2014). The authors in Yang et al. (2013) assume a scenario that the base station has to send
multiple independent data streams to multiple legitimate users; during the transmission, many
eavesdroppers with multiple antennas have interests in the transmission stream of the base
station. The eavesdroppers may collude or not, and maximize the signal-to-interference-plus-
noise ratio (SINR) of the desired streams using the beamforming method. The cooperative
jammer will work on keeping the SINR at the eavesdroppers below a certain threshold level
to guarantee a confidential transmission between the base station and the legitimate users.
Another scenario in Li et al. (2014a) studies the Gaussian wiretap channel’s secrecy capacity
aided by an external jammer. Each of the receiver and the transmitter has a single antenna,
while the jammer and the eavesdropper are equipped with multiple antennas. The authors in
Wang et al. (2014b) reveal a scenario for secure transmission within a two-hop amplify-and-
forward relay network scheme, such that for large number of antennas at the source, the
ergodic
secrecy capacity (ESC) is independent of the number of antennas; whereas, for a large number
of antennas at the destination, the ESC is dependent on the number of antennas.
Beamforming is a very efficient method also when it is used with the cooperative jamming
technique. However, these two techniques are adopted separately in most worksWang et al.
(2013a); Tran & Kong (2014); Wang et al. (2013b); Han et al. (2015); Vishwakarma &
Chock- alingam (2014). In Wang et al. (2013a), a scheme with joint cooperative jamming
and beamforming is proposed to enhance the security of a cooperative relay network, where
part of the nodes uses a distributed beamforming mechanism while the others jam the
eavesdropper simultaneously. In Tran & Kong (2014), another scheme of using the
beamforming is proposed; by preventing the eavesdroppers from using the beamformers to
suppress the jamming signals. It uses also two orthogonal dimensions for transmitting and
receiving signals. A hybrid cooperative jamming and beamforming scheme is proposed in
Wang et al. (2013b) also; the idea is in both cooperative transmission phases, some
intermediate nodes relay the signals to the legitimate receivers by adopting the
beamforming distribution, while, simultaneously, the other nodes jam the eavesdropper,
which eventually leads in protection of the transmitted data. The authors in Han et al. (2015)
develop an optimal relay assignment algorithm to solve the secrecy capacity maximization
problem, and a smart jamming algorithm is proposed to increase the secrecy capacity of the
system.
4.2 Cooperative jamming with Power Allocation method
Since the system’s performance in cooperative jamming depends highly on the jamming
strategy as well as the power level of the jamming Park et al. (2013), three power allocation
strategies are derived in Park et al. (2013) to minimize the outage probability of the secrecy
rate, besides that, three kinds of jamming power allocation schemes are proposed according
to the available channel state information (CSI) at the destination to minimize the outage
probability. The authors in Zhang et al. (2015) propose another scenario investigates the
MISO channels with power splitting scheme used by the legitimate receiver to split the
received signal for both information decoding and energy harvesting. Another power
allocation method in Long et al.
(2014) is analysed in which the source nodes should send jamming signals as a part of their
power instead of hiring extra nodes to jam the eavesdropper. Two types of jamming signals
are analysed; a priori known jamming signals at the source nodes, and unknown jamming
signals at the source nodes. A major finding reported in this work is that, if the jamming
signals are known a priori at the source nodes, the secrecy capacity is improved significantly
when compared to the scenario in which the jamming signals are unknown at the source
nodes. In Yang et al. (2014), a linear precoding scheme is utilized at the base station, which
exploits transmit diversity by weighting the information stream, this is studied with the
cooperative jamming strategy. An optimal solution is obtained when the number of antennas
at the friendly jammer is no less than the total number at the eavesdropping antennas. The
authors in Wang et al. (2015a) propose a sequential parametric convex approximation
(SPCA) based algorithms to address the power allocation optimization and maximize the
ergodic secrecy rate (ESR) lower bound, and then it is shown that the optimized power
allocation tends to allocate more power to the jamming signals to improve the secrecy
capacity. An optimal relay selection criterion and power allocation strategy are derived in
Wang & Wang (2014) between the jamming signals and the confidential information for the
ESR maximization. Another study in Deng et al. (2015) shows that a helper node should
allocate its power as a jammer or as a helper depending on the locations of the helper and the
eavesdropper.
4.3 Jamming Policies
Several policies are proposed for relay selection Liu et al. (2015); Sun et al. (2015); Hui et
al. (2015). In Liu et al. (2015), four relay selection policies are proposed and compared,
namely random relay and random jammer, random jammer and best relay, best relay and best
jammer, and best relay and no jammer; and it characterizes the joint impact of the proposed
relay selection policies and the interference power constraint on the secrecy performance by
deriving new exact closed-form expressions for the secrecy outage probability; it is shown
then that the jammer’s absence gives rise to the outage saturation phenomenon. Two relay
and jammer selection methods are developed in Hui et al. (2015) for minimizing the secrecy
outage probability; in both these selection methods, each intermediate node knows its own
role while the
knowledge of the jammer and relay set is kept secret from all the eavesdroppers. It is shown
that maintaining the privacy of the selection result improves greatly the secrecy outage proba-
bility performance. This work assumes a decode and forward relay system, in which the
source communicates with the destination through many intermediate nodes in the presence
of several passive eavesdroppers. The intermediate nodes act as either jammers or as
conventional relays to hinder the eavesdroppers from intercepting the signal of interest. The
destination broadcasts information that allows the intermediate nodes to determine whether
they will serve as relays or jammers, but this information does not allow the eavesdroppers to
know the selection result. In Park et al. (2013), a scheme is provided which has a
destination, relay and a source; the destination starts to send jamming signals towards the
eavesdropper while the source is sending the message to the relay, and the destination then
removes the jamming noise perfectly via self-interference cancellation at the second phase.
Another scheme in Liu et al. (2013) is provided; in the first phase, the source transmits the
information bearing signal, simultaneously as it cooperates with the destination in jamming
the eavesdropper without interference at the relay. In the second phase, a relay is selected
optimally, which transmits the decoded source signal, at the same time, this relay cooperates
with the source to jam the eavesdropper without creating interference at the destination. The
authors in Lin et al. (2013a) propose a new transmission scheme, where the relaying group
and the jamming group are constructed together, this scheme enables to block the
eavesdroppers simultaneously and further increase the signal-to-noise ratio at the destination.
In Chen et al. (2013), attack strategies are investigated in a multi-relay network that consists
of both malicious and cooperative relays, where the malicious relays are given the freedom to
listen to the source in the first phase (so that they can send interfering signals in the second
phase), or to directly emit jamming signals in both phases. Subsequently, it is shown that the
malicious relays should attack in both phases rather than just listening in the first phase and
attack in the second phase. On the other hand, the opportunistic cooperative jamming and the
opportunistic relay chatting schemes are compared in Ding et al. (2011). It is shown that the
chatting scheme is better for implementing the relay nodes to jam the eavesdropper in the
both phases comparing with cooperative jamming scheme in which only the eavesdropper
jams in the first phase.
Moreover, jamming policies using game theory methods are proposed in Fakoorian &
Swindle- hurst (2013); Chen et al. (2013); Stanojev & Yener (2013); Li et al. (2014b). In
Fakoo- rian & Swindlehurst (2013), a scheme of two user multiple-input-single output
Gaussian interference channel is considered, where each transmitter aims to maximize the
difference between its secrecy rate and the other’s secrecy rate, in this scheme, the weaker
link tries to minimize the extra secrecy rate of the other transmitter, while the transmitter
with the stronger link tries to maximize it. This paper uses Nash equilibrium strategy as a
solution in its scheme. A multi-relay network is considered in Chen et al. (2013) that consists
of both malicious and cooperative relays, and applies Nash equilibrium game strategy on its
scheme, by modelling the cooperative relays set and the malicious relays set as two players in
a zero sum game with the maximum achievable rate as the utility. The authors in Stanojev &
Yener (2013) propose another game-theoretic model, Stackelberg game, with the legitimate
parties being spectrum owners acting as a game leader, and the set of the assisting jammers
which are constituting the follower. It shows that when the potential jammers’ number
increases, utility of a chosen jammer for any scheme will start to decrease as the legitimate
parties can be more aggressive when leading the game. In Li et al. (2014b), a smart jammer
can quickly learn the transmission strategies of the legitimate transmitters, then he would
adjust his strategy to damage the legitimate transmission. Meanwhile, the transmitters are
aware of the existence of the smart jammer. This anti-jamming scenario is modeled as a
Stackelberg game, where the leader is the source node and the follower is the jammer. It is
shown that the optimal power control strategies obtained from the Stackelberg equilibrium
game can decrease the damage caused by the jammer.
5. Conclusion
Unlike its conventional applications, jamming techniques are used to enhance the security of
transmission in wireless communication networks. In this paper, we have surveyed the differ-
ent challenges related to the physical layer security in wireless networks and we developed a
literature review of the different jamming techniques within the existing recent literature with
their advantages and disadvantages.
Based on this review we can conclude that there are still many issues to be resolved around
jamming techniques applications such as communication architectures for energy harvesting,
protocols, and interference management.
Acknowledgment
This work has been supported by the ETS’ research chair of physical layer security in wireless
networks.
APPENDIX IV
SECRECY ANALYSIS OF COOPERATIVE NETWORK WITH UNTRUSTWORTHY

RELAYS USING LOCATION-BASED MULTICASTING TECHNIQUE
1
Paper published in 5th International Conference on Future Internet of Things and Cloud
Workshops (FiCloudW) , November 2017.
1. Abstract
This paper studies the secrecy capacity scaling in an Amplify-and-Forward (AF) dual-phase
large network containing K relays. In our model, part of these K relays are assumed to be
potential eavesdroppers. Before transmitting the message in the first phase, the multi-
antennas source divides it to partial messages and multicasts each part to a different disjoint
sector. During the second phase, the K relays use the distributed beamforming (DBF)
technique to retransmit the message to the destination. We investigate the ergodic secrecy
capacity considering two different behaviours of the untrustworthy relays; the passive
behaviour, when the untrustworthy relays work separately from each other to intercept the
signal, and the aggressive one, when the untrusted relays collaborate to hijack the message.
As demonstrated, our location-based multicasting scenario is significantly increasing the
security compared to the recent works that employ broadcasting schemes. Additionally, it
also increases the secrecy capacity scaling remarkably. Finally, our analytical derivations
are confirmed by the simulation.
Keywords: Physical layer security, location-based multicasting, amplify and forward, dis-
tributed beamforming, secrecy capacity.
2. Introduction
106
Over the last few years, security has always been considered a critical issue in wireless net-
works’ design. The vital concept of the secrecy capacity is built on whether increasing the
legitimate channel capacity or decreasing the capacity of the illegitimate channels, which is
attainable via the usage of the dynamic nature of the wireless channels Gopala et al. (2008).
Therefore, many contributions have been recently accomplished to escalate the secrecy
capacity by associating advanced techniques in wireless communications, such as multiple
antenna schemes, game theory techniques, beamforming and power allocation methods
Atallah et al. (2015). Wireless security appears to be a crucial matter in today’s
communication systems as both the diversity and the number of users in wireless networks
keep growing. According to these security challenges, leading researchers are seeking more
information theoretical methods to accomplish almost perfect security in wireless channels.
With this approach, consider- able efforts have been made by authors in Gopala et al. (2008);
Bloch et al. (2008) to develop information-theoretical security, which raises the opportunity
to have a secure communication in an existence of eavesdroppers. The foundations of
information-theoretic security were led by Wyner (1975); Leung-Yan-Cheong & Hellman
(1978). Obviously, many facts haven’t been considered in this domain’s primal works, such
as the wireless channels that are susceptible to fading or that the communicating devices
constitute networks out of unknown topology. A few decades later, channel propagation effect
has been considered in Gopala et al. (2008); Bloch et al. (2008). In this direction, the authors
in Gopala et al. (2008) investigated the secrecy capacity of wireless fading channels
considering the channel state information (CSI). The authors in Bloch et al. (2008) found the
average secrecy capacity and the outage probability expressions of quasi-static fading
channels for both the perfect and the imperfect CSI scenarios. Considering random
topologies, the secrecy capacity has been investigated in Haenggi (2008). Following this
direction, the researchers in Koyluoglu et al. (2012) studied the secrecy capacity scaling
laws. The secrecy capacity of unicast links with the existence of multiple wiretappers was
investigated in Vuppala & Abreu (2013), where the transmission to the k-th legitimate node
was based on the order of the distance between the source and the destination. Hence- forth,
relay aided transmission has been taken into consideration as an effective way to escalate the
transmission reliability, throughput and coverage probability in the literatureLaneman et al.
(2004); Lin et al. (2014). Several strategies have been studied in literature for relay aided
transmission, particularly amplify and forward, decode and forward and demodulate and
forward. As the nature of the wireless medium previously explained, some of the relays could
possibly be eavesdroppers within the transmission area. In Kim et al. (2015), the secrecy
capacity scaling and the diversity order were calculated with the presence of potential
passive eavesdroppers with the destination-assisted jamming, in which the destination
transmits jamming messages to the relays. However, in Atallah & Kaddoum (2016), the
authors went deeper to find the capacity scaling in a worse scenario; the possibility that
potential aggressive eavesdroppers could cooperate together to intercept the received
message.
In this paper, we introduce a new system model that remarkably improves the security, espe-
cially compared to the recent works Kim et al. (2015); Atallah & Kaddoum (2016). We
propose a two-hop AF relaying model. It is also assumed that the source divides its message
to parts and sends each part to a different directional antenna in which each element covers a
disjoint area. Whereas in the second phase, using the distributed beamforming technique
(DBF), the relays retransmit the received message towards the destination. This DBF method
is proved to be very efficient compared to other methods like opportunistic relaying
technique Kim et al. (2015); Atallah & Kaddoum (2016). Again, we assume two types of
relays, trustworthy and untrustworthy. Two kinds of untrusted relays are studied; passive,
where each relay tries to intercept the message individually, and aggressive, where each
untrustworthy relay sends its received signal to a concurrent eavesdropper which in its turn
aggregates the received signals to decipher the message. The main contributions presented in
this paper are finding the ergodic secrecy capacity under location-based multicasting scenario
in two cases:
- The potential untrustworthy relays are passive where they work apart to interpret the mes-
sage.
- The potential untrustworthy relays are aggressive by collaborating between each other to
hijack the message.
Notations:
thermore, [A ]X+ and E[X max
denotes ] denote
A, 0the mean
. For expectation
a r.v. of a random
X , the notation variable (r.v.) X . Fur-
w.p.1X N c(a, b) denotes that X is
a complex Gaussian r.v. with mean a and variance b. X → denotes the convergence with
{ }
probability 1. ∼
3. SYSTEM MODEL
Figure-A IV-1 System model. The source s sends its message to the
destination d by multicasting its partial signals in the first hop, xs,l, xs,m
and xs,g, towards different sections L,M and G. In the second hop, the
relays
retransmit the signals to the destination d using beamforming technique DBF
Consider a two-hop wireless network consisting of K relays with a destination d and a multi-
antennas source s. In our network, as shown in Fig. IV-1, each relay has a single antenna
operating in a half-duplex mode. In addition, we assume that there is no direct link between
the source s and the destination d, i.e. all the transmitted information must be forwarded by
relays.
In our analysis, the channels are assumed to be reciprocal frequency-flat block-fading with
the coefficient between nodes i and j being denoted by hi, j and being modelled as a Gaussian
r.v. hi, j ∼ lnand ij , σ ij ) where (i, j ∈ {s, k, d}). The channel gains hs,k
N ( exponentially and hk,d is the
are
independent distributed r.v.’s. We assume that the noise
. variance
.2 . N 0 .2
μ 2
same in the first and the second phase, and the CSI is known by the receiving nodes. Instead
of broadcasting the message in the first phase, the source will divide the signal into multiple
parts and multicast each part to a specified sector, as demonstrated in Fig. IV-1. Each section
will be denoted by the total number of the relays that it has. Without loss of generality, we
study the case where the source divides its message into three partial messages and sends
them to three different disjoint areas, due to the use of directional antennas. This given
configuration is used to make the derivation easy to follow. Therefore, the general form of
the secrecy capacity using multicasting with many partial messages is given at the end of the
analysis part. As shown in Fig. IV-1, the first, the second and the third sections will be
denoted by L, M and G respectively. We will use the same letters L, M and G to denote the
total number of relays in its
corresponding
xs,r, when it issection. Hence,
sent to the K =R,L where
section + M +rG.
∈ We willg},
{l, m, also
R∈denote theG}
{L, M, partial
and message
1 ≤ r ≤ by
R.
The received
signal at the rth relay is expressed by
√
yr = hs,r Ps,r xs,r + ns,r , (A IV-
1) where ns,r is a complex additive white Gaussian noise at the rth relay with zero mean and
variance N0. The transmitted powers of the source are denoted by Ps,r. It is assumed that at
the
rth relay, the received signal-to-interference-plus-noise ratio SINR would be
γr = ρs,r |hs,r|2, (A IV-2)

Δ ,
where the signal to noise ratio is denoted by ρ
s,r = Ps,r N0 , r ∈ {l, m, g}. We will assume that
=
s,
r
the signal to noise ratios are equal ρ Δ ρs,r , but the extension using different ρ values is
straightforward. In the following subsections, we will find the secrecy capacity scaling for
our system in two cases; when there are untrusted relays trying to intercept their received
messages
individually, and when these untrusted relays work together to intercept the message. We will
later use LJ , M J and GJ to denote the total number of the untrustworthy aggressive relays in
each of L, M and G sectors respectively, where L > LJ , M > M J and G > GJ .
4. Secrecy Capacity for Passive Untrustworthy relays
In this case, each untrustworthy passive relay works individually to intercept the received
message without any kind of cooperation with any other relays. The received SINR at any
potential untrusted passive relay would be given by
2
γ ρ h (A IV-3)
k = s,k . .
s,k ,
Considering that each relay has just a part of the message, and that the message is already
encrypted in the upper layers before it is divided to partial messages, its interpretation is im-
possible considering the other parts are missing.
by the rth ar yr . The normalized amplifying coefficient ar for the rth relay is as
is xr = beamforming
Using the relay
distributed strategy DBF in the second phase, the retransmitted
signal follows
1
a (A IV-4)
ρs,r | + N0
r =. h | 2 .
s,r
However, the destination can receive a signal either from a trustworthy or untrustworthy relay.
The received signal at the destination, from a random relay, can be expressed as
yd = hr,dar yr + nd . (A IV-5)
where nd is a complex additive white Gaussian noise AWGN with zero mean and variance N0
at the destination. The received SINR from the rth relay becomes
ρs,r |hs,r|2ρr,d .hr,d 2

, (A IV-6)
γ =
.
2
. .
r
ρs,r |hs,r r,
hr,d 2
|+ρ d +1
for r ∈ {l, m, g}. The channel capacity from the source to the destination would be given by
G γΣ
. L γΣ . M γΣ .
1 1 1
1+ ∑
l=1 l
1+ ∑
m=1 m
1+ . ∑
(A IV-7)
g=1 g
Cd = log + log + log
2 2 2
whereas Cw is the secrecy capacity between a potential untrusted passive relay and the source,
and it is given as follows
12 , ,
k . . 2
ρ
Cw = log .1 + max . s k hs k ΣΣ . (A IV-8)
We consider the maximum SINR in Eq. (A IV-8) to evaluate the worst case in which the
eavesdropper could obtain the maximum amount of information. From Eq.(A IV-7) and Eq.
(A IV-8), the ergodic secrecy capacity could be written as
CP = E {CP}
.
= [Cd −Cw]+
EΣ
(a)
+
, (A IV-9)
≥ [E {Cd} −E {Cw}]
where CP is the instantaneous secrecy capacity.
(a) follows from Jensen’s inequality
E {max(X1, X2)} ≥ max(E {X1}, E {X2}) . (A IV-10)

For R → ∞, γr in Eq.(A IV-6) satisfies the Kolmogorov conditions i.e.
R
∑ VAR[γr]
< ∞, (A IV-11)
r=1 r2
and
μ = R (A IV-12)
E [γ ] <
∞r 1
∑ r ,
R
r=1
are true for any finite ρ Bolcskei et al. (2006) where R ∈ {L, M, G}.
Thus, we can apply the following theorem [15, Theorem 1.8.D] :

R γr E [γr] w.p.1
R
∑ R ∑ R
− → 0. (A IV-13)
r=1 r=1
w.p.1
Therefore, γr → R μr and
1
E . log .1 R γ ΣΣ
+∑ r 1 ( )
log R . (A IV-14)
2 r=1
∼
2
Fact 1: max. ρs,k hs,k Σ behaves like ρs,k log(K)+ O (loglog(K)) for K → ∞ and limited ρ
2
k . .
[16, Lemma 4].
From Fact 1 and Eq.(A IV-8), Cw will be as follows

1
Cw ∼ loglog(K) . (A IV-15)
2
The secrecy capacity in Eq.(A IV-9) can be represented by
1 1 1 1
CP ≥ log(L)+ log(M)+ log(G) − loglog(K) ,
2 2 2 2
21
CP “ log(LMG) . (A IV-16)
Since the relays are half-duplex, we use the rate-loss factor value of 1/2.
5. Aggressive Untrustworthy Relays
The aggressive untrustworthy relays are the relays that collaborate together by retransmitting
their received signals to a one wiretapper to decipher the message. The wiretapper could be
internal i.e. one of the relays, or external. In our scheme, we will assume that the wiretapper
e is external, as shown in Fig. IV-1. We will denote the total number of all the
untrusted aggressive relays in the network by U . We will assume that these untrusted
aggressive relays are distributed equally between all the sectors L, M and G. From Fig.
IV-1, LJ , M J and GJ denote the total number of the untrustworthy aggressive relays in each of
L, M and G sectors, respectively.
The retransmitted signal from each untrusted aggressive relay towards the wiretapper e will be
denoted by yrJ where rJ ∈ {l J , mJ , gJ } . Hence, the received signal at the wiretapper e from each
of the untrustworthy aggressive relays would be
ye = hrJ ,e a J yrJ + ne, (A IV-17)

r
where hrJ ,e is the channel coefficient between an untrustworthy relay and the wire-tapper, ne is
a complex AWGN with zero mean and variance N0 at the wire-tapper. The received SINR at
the wire-tapper becomes
s,rJ s,rJ 2 rJ ,e rJ ,e 2
. . ..ρ , (A IV-18)
h ρ h
2 2
γrJ = ρs,rJ .hs,rJ . + ρrJ ,e .hrJ ,e . + 1
The instantaneous channel capacity at the wiretapper will be given by

. LJ γΣ . γ Σ . γ Σ
MJ GJ
1 1+∑ 1 1+ ∑ 1 1+ ∑ . (A IV-19)
l =1 l gJ =1 g
J J
J
mJ m =1
J
Ce = log + log + log
2 2 2
Consequently, the ergodic secrecy capacity with the presence of the aggressive relays could
be written as
. Σ
CA = E {CA } = E [Cd − Ce ]+ (A IV-20)
(a)
≥ [E {Cd} −E , (A IV-21)
{Ce}]+
where (a) follows from the fact that E {max(X1, X2)} ≥ max(E {X1}, E {X2}). By applying
the steps followed for the Eq.(A IV-14) in the previous subsection, the secrecy capacity scaling
for untrusted aggressive relays will take the form
1 1 1
CA = log(L)+ log(M)+ log(G) −
2 2 2
1 1 1
. Σ . Σ . Σ
log LJ + log M J + log GJ
2 1 2 1 2
. Σ
= log (LMG) − log LJ M J GJ
2 2
12 LM G JG
LJ M
CA = log . J Σ, (A IV-22)
Without loss of generality, we will assume that
K L M G
= = = = T. (A IV-23)
U LJ M J GJ
By compensating Eq.(A IV-23) in Eq.(A IV-22), the secrecy capacity would be
1
. Σ
CA = log T 3 . (A IV-24)
2
Then, by generalizing Eq.(A IV-24) to multicast V partial messages towards V different sec-
tions instead of three, the secrecy capacity scaling will be expressed as
1 V
. Σ
CA = log TV = log(T ) , (A IV-25)
2 2
V2 KU
CA = log . Σ . (A IV-26)
6. Simulation results
Ergodic secrecy capacity [bps\Hz]
100
Chan. capacity sim; Multicasting on trusted relays

Sec. capacity sim; Multicasting on untrusted aggr. relays Chan. capacity sim; Broadcasting on trusted relay
Sec. capacity sim; Broadcasting on untrusted aggr. relays
0 50 100 150 200

Number of Relays K
Figure-A IV-2 Ergodic secrecy capacity: ρ =Δ ρs, = ρ k,

= ρ k, = 5dB,
k d e
3
.hs,k.2 = .hk,e.2 = .hk,d .2 = 1 and U = 1 K

Ergodic Secrecy Capacity [bps/Hz]
100
Multicasting 3 partial messages to 3 different sections

Multicasting 6 partial messages to 6 different sections Multicasting 9 partial messages to 9 different sections
0 50 100 150 200

Number of Relays K
Figure-A IV-3 Ergodic secrecy capacity when using Location-Besed

Multicasting technique with the presence of untrusted aggressive relays
for different values of V which is identified in Eq.(A IV-26):
ρ Δ ρs, k = ρk, d = ρk, e = 5dB, .hs, k. = .hk, e. = .hk ,d . = 1 and U = 1K
2 2 2
3
ρ ρ ρ . .2 . . 2 . .2 1
Assuming that ρ =Δ s, k = k, d = k, e = 5dB, hs, k = hk, e = hk , d = 1 and U = K, in
3
Fig.IV-2, we do the performance comparison between our scenario, where the source
multicasts each part of the message, and the other scenarios, from recent studies, in which the
source just broadcasts the signal Kim et al. (2015); Atallah & Kaddoum (2016). The results,
which are simulated in Matlab, show the improvement in the secrecy capacity of our scenario
compared to the broadcasting one. For the passive untrusted relays’ case, considering that the
message is already encrypted in the upper layers before dividing it to partial messages, the
secrecy capacity is not affected as long as the eavesdroppers cannot have an access to the
other parts of the message, which gives a lot of enhancements in the security perspective and
eliminates the need for some of the other security solutions like cooperative jamming. On the
other hand, for aggressive eavesdroppers cooperating between each other to assemble the
message’s parts,
the secrecy capacity is considerably enhanced a lot, as a result of ameliorating the channel
capacity from 1 log(K)= 1 log(L + M + G) in Kim et al. (2015),Atallah & Kaddoum (2016)
2 2
2
to 1 log(LMG) as shown in our study. In Fig.3.4, we simulate the ergodic secrecy capacity
with the presence of untrustworthy aggressive relays when V , identified in Eq.(A IV-26), takes
the values 3, 6 and 9. As shown in Fig.3.4, the more we multicast partial messages towards
different sectors, the more the security is enhanced in our network.
7. Conclusions
In this paper, we have investigated the secrecy capacity scaling in large networks within two
contrasting roles of potential eavesdroppers; the passive and the aggressive one. We showed
that using our location-based multicasting technique will not only increase the secrecy
capacity in the presence of the aggressive relays, but prohibit the individual attempts, by
passive eavesdroppers to intercept the message as well. In addition, our proposed scheme is
less energy consuming compared to some techniques in physical layer security like
cooperative jamming methods. Besides, it does not need complicated calculations or
advanced security algorithms, which opens the door for it to be applied in Internet of Things
world.
BIBLIOGRAPHY
A. El-Malek, A. H., Salhab, A. M., Zummo, S. A. & Alouini, M.-S. (2017). Power allo-
cation and cooperative jamming for enhancing physical layer security in opportunistic
relay networks in the presence of interference. Trans. Emerging Telecommun.
Technol., 28(11), e3178. doi: 10.1002/ett.3178. e3178 ett.3178.
Akhlaghpasand, H., Björnson, E. & Razavizadeh, S. M. (2019). Jamming Suppression in

Mas- sive MIMO Systems. IEEE Trans. Circuits Syst. II, Exp. Briefs, 1-1. doi:
10.1109/TC- SII.2019.2902074.
Akyildiz, I. F., Su, W., Sankarasubramaniam, Y. & Cayirci, E. (2002). Wireless sensor net-
works: a survey. Computer networks, 38(4), 393–422.
Alibeigi, M. & Taherpour, A. (2019). Optimisation of secrecy rate in cooperative device to

device communications underlaying cellular networks. IET Commun., 13(5), 512-519.
doi: 10.1049/iet-com.2018.5507.
Alouini, M. S. & Goldsmith, A. J. (1999). Capacity of Rayleigh fading channels under dif-
ferent adaptive transmission and diversity-combining techniques. 48(4), 1165-1181.
doi: 10.1109/25.775366.
Alsaba, Y., Leow, C. Y. & Abdul Rahim, S. K. (2019). Null-Steering Beamforming for
Enhancing the Physical Layer Security of Non-Orthogonal Multiple Access System.
IEEE Access, 7, 11397-11409. doi: 10.1109/ACCESS.2019.2890822.
Arafa, A., Shin, W., Vaezi, M. & Poor, H. V. (2018, Dec). Securing Downlink Non-
Orthogonal Multiple Access Systems by Trusted Relays. IEEE Global Commun. Conf.
(GLOBE- COM), pp. 1-6. doi: 10.1109/GLOCOM.2018.8648037.
Atallah, M. & Kaddoum, G. (2016). Secrecy Capacity Scaling With Untrustworthy

Aggressive Relays Cooperating With a Wire-Tapper. IEEE Wireless Commun. Lett.,
5(4), 376-379. doi: 10.1109/LWC.2016.2561285.
Atallah, M. & Kaddoum, G. (2017, Aug). Secrecy Analysis of Cooperative Network with
Untrustworthy Relays Using Location-Based Multicasting Technique. 2017 5th Intern.
Conf. on Future Internet of Things and Cloud Workshops (FiCloudW), pp. 206-210.
doi: 10.1109/FiCloudW.2017.74.
Atallah, M. & Kaddoum, G. (2019). Secrecy Analysis in Wireless Network with Pas-
sive Eavesdroppers by Using Partial Cooperation. IEEE Trans. Veh. Technol., 1-1.
doi: 10.1109/TVT.2019.2913934.
Atallah, M. & Kaddoum, G. (2019). Design and Performance Analysis of Secure Multicas-
ting Cooperative Protocol for Wireless Sensor Network Applications. IEEE Wireless
Commun. Lett.
120
Atallah, M., Kaddoum, G. & Kong, L. (2015, Oct). A Survey on Cooperative Jamming Ap-
plied to Physical Layer Security. 2015 IEEE Int. Conf. Ubiquitous Wireless Broadband
(ICUWB),, pp. 1-5. doi: 10.1109/ICUWB.2015.7324413.
Atallah, M., Alam, M. S. & Kaddoum, G. (2019). Secrecy analysis of wireless sensor
network in smart grid with destination assisted jamming. Institution of Eng. Technol., -
(0). Con- sulted at https://digital-library.theiet.org/content/journals/10.1049/iet-
com.2018.5344.
Badia, L. & Gringoli, F. (2019). A Game of One/Two Strategic Friendly Jammers Versus a
Ma- licious Strategic Node. IEEE Netw. Lett., 1(1), 6-9. doi:
10.1109/LNET.2019.2893536.
Baig, Z. A. & Amoudi, A.-R. (2013). An Analysis of Smart Grid Attacks and Countermea-
sures. JCM, 8, 473-479.
Banawan, K. & Ulukus, S. (2014, Sept.). Gaussian MIMO wiretap channel under receiver
side power constraints. Annu. Allerton Conf. Commun., Control, and Computing
(Allerton),, pp. 183-190. doi: 10.1109/ALLERTON.2014.7028454.
Barros, J. & Rodrigues, M. (2006, July). Secrecy Capacity of Wireless Channels. IEEE Int.
Symp. Inform. Theory., pp. 356-360. doi: 10.1109/ISIT.2006.261613.
Bassily, R. & Ulukus, S. (2012). Deaf Cooperation for Secrecy With Multiple An-
tennas at the Helper. IEEE Trans. Inform. Forens. Security., 7(6), 1855-1864.
doi: 10.1109/TIFS.2012.2215325.
Bassily, R. & Ulukus, S. (2013). Deaf Cooperation and Relay Selection Strategies for Secure
Communication in Multiple Relay Networks. IEEE Trans. Signal Process., 61(6),
1544- 1554. doi: 10.1109/TSP.2012.2235433.
Bloch, M., Barros, J., Rodrigues, M. & McLaughlin, S. (2008). Wireless Information-
Theoretic Security. IEEE Trans. Inform. Theory., 54(6), 2515-2534. doi:
10.1109/TIT.2008.921908.
Bloch, M. & Barros, J. (2011). Physical-Layer Security: From Information Theory to

Security Engineering. Cambridge University Press.
Bolcskei, H., Nabar, R., Oyman, O. & Paulraj, A. (2006). Capacity scaling laws
in MIMO relay networks. IEEE Trans. Wireless Commun., 5(6), 1433-1444.
doi: 10.1109/TWC.2006.1638664.
Chen, H. C., Lin, T. H., Kung, H. T., Lin, C. K. & Gwon, Y. (2012, Oct). Determining RF
angle of arrival using COTS antenna arrays: A field evaluation. MILITARY
COMMUN. CONF. MILCOM, pp. 1-6. doi: 10.1109/MILCOM.2012.6415851.
Chen, J. (2018, Oct). Secure Communication over Interference Channel: To Jam or Not
to Jam? 56th Ann. Allerton Conf. Commun. Control, and Computing, pp. 1120-1127.
doi: 10.1109/ALLERTON.2018.8635988.
Chen, M.-H., Lin, S.-C., Hong, Y.-W. & Zhou, X. (2013). On Cooperative and Malicious
Behaviors in Multirelay Fading Channels. IEEE Trans. Inform. Forensics and Security,
8(7), 1126-1139. doi: 10.1109/TIFS.2013.2262941.
Chen, X., Ng, D. W. K., Gerstacker, W. H. & Chen, H. H. (2017). A Survey on Multiple-
Antenna Techniques for Physical Layer Security. IEEE Commun. Surveys Tuts., 19(2),
1027-1053. doi: 10.1109/COMST.2016.2633387.
Cover, T. M. & Thomas, J. A. (2006). Elements of Information Theory (Wiley Series in

Telecommunications and Signal Processing). Wiley-Interscience.
Csiszar, I. & Korner, J. (1978). Broadcast channels with confidential messages. IEEE Trans.
Inform. Theory., 24(3), 339-348. doi: 10.1109/TIT.1978.1055892.
Cumanan, K., Alexandropoulos, G. C., Ding, Z. & Karagiannidis, G. K. (2017).

Secure Communications With Cooperative Jamming: Optimal Power Allocation
and Secrecy Outage Analysis. IEEE Trans. Veh. Technol., 66(8), 7495-7505.
doi: 10.1109/TVT.2017.2657629.
Dahmane, S., Kerrache, C. A., Lagraa, N. & Lorenz, P. (2017, May). WeiSTARS: A weighted
trust-aware relay selection scheme for VANET. IEEE Intern. Conf. on Commun.
(ICC), pp. 1-6. doi: 10.1109/ICC.2017.7996451.
Deng, H., Wang, H.-M., Guo, W. & Wang, W. (2015). Secrecy Transmission With a
Helper: To Relay or to Jam. IEEE Trans. Inf. Forensics and Security., 10(2), 293-307.
doi: 10.1109/TIFS.2014.2374356.
Ding, Z., Leung, K., Goeckel, D. & Towsley, D. (2011). Opportunistic Relaying for Secrecy
Communications: Cooperative Jamming vs. Relay Chatting. IEEE Trans. Wireless
Com- mun., 10(6), 1725-1729. doi: 10.1109/TWC.2011.040511.101694.
Do, Q. V., Hoan, T. N. K. & Koo, I. (2019). Optimal Power Allocation for Energy-efficient
Data Transmission Against Full-duplex Active Eavesdroppers in Wireless Sensor
Networks. IEEE Sensors J., 1-1. doi: 10.1109/JSEN.2019.2904523.
Dong, L., Yousefi’zadeh, H. & Jafarkhani, H. (2011, June). Cooperative Jamming and Power
Allocation for Wireless Relay Networks in Presence of Eavesdropper. IEEE Int. Conf.
Commun. (ICC), pp. 1-5. doi: 10.1109/icc.2011.5963094.
Dubey, A. & Mallik, R. K. (2015). PLC System Performance With AF Relaying. IEEE Trans.
Commun., 63(6), 2337-2345. doi: 10.1109/TCOMM.2015.2427171.
El Shafie, A., Mabrouk, A., Tourki, K., Al-Dhahir, N. & Hamila, R. (2017). Securing
Untrusted RF-EH Relay Networks Using Cooperative Jamming Signals. IEEE Access,
5, 24353- 24367. doi: 10.1109/ACCESS.2017.2768508.
Fakoorian, S. A. A. & Swindlehurst, A. L. (2013). Competing for Secrecy in
the MISO Interference Channel. IEEE Trans. Signal Process., 61(1), 170-181.
doi: 10.1109/TSP.2012.2223689.
Fang, X., Misra, S., Xue, G. & Yang, D. (2012). Smart Grid - The New and Im-
proved Power Grid: A Survey. IEEE Commun. Surveys Tuts., 14(4), 944-980.
doi: 10.1109/SURV.2011.101911.00087.
Felkaroski, N. & Petri, M. (2019, Feb). Secret Key Generation Based on Channel State
Information in a mmWave Communication System. SCC; 12th Int. ITG Conf. Syst.
Commun. and Coding, pp. 1-6. doi: 10.30420/454862049.
Fenton, L. (1960). The Sum of Log-Normal Probability Distributions in Scatter Transmission

Systems. IRE Trans. Commun. Syst., 8(1), 57-67.
Ghosh, M. (1996). Analysis of the effect of impulse noise on multicarrier and single carrier
QAM systems. IEEE Trans. Commun., 44(2), 145-147. doi: 10.1109/26.486604.
Goel, S. & Negi, R. (2008). Guaranteeing Secrecy using Artificial Noise. IEEE Trans.
Wireless Commun., 7(6), 2180-2189. doi: 10.1109/TWC.2008.060848.
Gopala, P. K., Lai, L. & El Gamal, H. (2008). On the Secrecy Capacity of Fading Channels.
IEEE Trans. Inform. Theory., 54(10), 4687-4698. doi: 10.1109/TIT.2008.928990.
Guillaume, R., Ludwig, S., Muller, A. & Czylwik, A. (2015, Oct). Secret key
generation from static channels with untrusted relays. IEEE 11th Int. Conf.
Wireless and Mobile Comput. Networking and Commun. (WiMob)., pp. 635-642.
doi: 10.1109/WiMOB.2015.7348022.
Gungor, V. C., Lu, B. & Hancke, G. P. (2010). Opportunities and Challenges of Wire-
less Sensor Networks in Smart Grid. IEEE Trans. Ind. Electron, 57(10), 3557-3564.
doi: 10.1109/TIE.2009.2039455.
Gungor, V. C., Sahin, D., Kocak, T., Ergut, S., Buccella, C., Cecati, C. & Hancke, G. P.
(2011). Smart Grid Technologies: Communication Technologies and Standards. IEEE
Trans. Ind. Informat., 7(4), 529-539. doi: 10.1109/TII.2011.2166794.
Guo, H., Yang, Z., Zhang, L., Zhu, J. & Zou, Y. (2017). Joint Cooperative Beamforming and
Jamming for Physical-Layer Security of Decode-and-Forward Relay Networks. IEEE
Access, 5, 19620-19630. doi: 10.1109/ACCESS.2017.2752199.
Haenggi, M. (2008). A Geometric Interpretation of Fading in Wireless Networks: Theory and

Applications. IEEE Trans. Inform. Theory, 54(12), 5500 - 5510.
Han, B., Li, J., Su, J., Guo, M. & Zhao, B. (2015). Secrecy Capacity Optimization via
Cooperative Relaying and Jamming for WANETs. IEEE Trans. Parallel and
Distributed Systems, 26(4), 1117-1128. doi: 10.1109/TPDS.2014.2316155.
He, B., Ni, Q., Chen, J., Yang, L. & Lv, L. (2019). User-Pair Selection in Multiuser Coop-
erative Networks With an Untrusted Relay. IEEE Trans. Veh. Techn., 68(1), 869-882.
doi: 10.1109/TVT.2018.2882178.
He, X. & Yener, A. (2008, Nov). Two-Hop Secure Communication Using an Untrusted
Relay: A Case for Cooperative Jamming. IEEE Global Telecommun. Conf.
(GLOBECOM)., pp. 1-5. doi: 10.1109/GLOCOM.2008.ECP.185.
He, X. & Yener, A. (2013). The Role of Feedback in Two-Way Secure Communications.
Holtzman, J. M. (1992). A simple, accurate method to calculate spread-spectrum multiple-

access error probabilities. IEEE Trans. Commun., 40(3), 461-464. doi:
10.1109/26.135712.
Hong, Y.-W. P., Lan, P.-C. & Kuo, C.-C. J. (2013). Signal Processing Approaches to Secure
Physical Layer Communications in Multi-Antenna Wireless Systems. Springer Publish-
ing Company, Incorporated.
Houjeij, A., Saad, W. & Bascar, T. (2013, June). A game-theoretic view on the physical layer
security of cognitive radio networks. IEEE Int. Conf. Commun. (ICC)., pp. 2095-2099.
doi: 10.1109/ICC.2013.6654835.
Hu, Y., Sanjab, A. & Saad, W. (2019). Dynamic Psychological Game Theory for Se-
cure Internet of Battlefield Things (IoBT) Systems. IEEE Internet of Things J., 1-1.
doi: 10.1109/JIOT.2018.2890431.
Huang, P. & Wang, X. (2013, April). Fast secret key generation in static wireless networks: A
virtual channel approach. IEEE Proc. INFOCOM., pp. 2292-2300. doi: 10.1109/INF-
COM.2013.6567033.
Hui, H., Swindlehurst, A., Li, G. & Liang, J. (2015). Secure Relay and Jammer Se-
lection for Physical Layer Security. IEEE Signal Process. Lett., 22(8), 1147-1151.
doi: 10.1109/LSP.2014.2387860.
Ibrahim, D., Hassan, E. & El-Dolil, S. (2014, Dec.). Improving physical layer security in
two-way cooperative networks with multiple eavesdroppers. Int. Conf. Informatics
and Syst. (INFOS), pp. ORDS-8-ORDS-13. doi: 10.1109/INFOS.2014.7036690.
Ibrahim, D. H., Hassan, E. S. & El-Dolil, S. A. (2015). Relay and jammer selection schemes
for improving physical layer security in two-way cooperative networks. Comput. &
Security, 50(0), 47 - 59. doi: http://dx.doi.org/10.1016/j.cose.2015.02.002.
Jameel, F., Wyne, S., Kaddoum, G. & Duong, T. Q. (2018). A Comprehensive Survey on Co-
operative Relaying and Jamming Strategies for Physical Layer Security. IEEE
Commun. Surveys Tuts., 1-1. doi: 10.1109/COMST.2018.2865607.
Jeong, C., Kim, I.-M. & Kim, D. I. (2012). Joint Secure Beamforming Design at the Source
and the Relay for an Amplify-and-Forward MIMO Untrusted Relay System. IEEE
Trans. Signal Process., 60(1), 310-325. doi: 10.1109/TSP.2011.2172433.
Kaddoum, G., Gagnon, F. & Richardson, F. (2012, Aug). Design of a secure Multi-Carrier
DCSK system. 2012 Int. Symp. Wireless Communication Syst. (ISWCS), pp. 964-968.
doi: 10.1109/ISWCS.2012.6328511.
Khisti, A. & Wornell, G. W. (2010). Secure Transmission With Multiple Antennas I: The
MISOME Wiretap Channel. IEEE Transactions on Inform. Theory., 56(7), 3088-3104.
doi: 10.1109/TIT.2010.2048445.
Kim, J.-B., Lim, J. & Cioffi, J. (2015). Capacity Scaling and Diversity Order for Secure Co-
operative Relaying With Untrustworthy Relays. IEEE Trans. Wireless Commun., 14(7),
3866-3876. doi: 10.1109/TWC.2015.2413784.
Kong, L. & Kaddoum, G. (2019). Secrecy Characteristics with Assistance of Mixture

Gamma Distribution. IEEE Wireless Commun. Lett., 1-1. doi:
10.1109/LWC.2019.2907083.
Kong, L., He, J., Kaddoum, G., Vuppala, S. & Wang, L. (2016, Sep.). Secrecy Analysis of a
MIMO Full-Duplex Active Eavesdropper with Channel Estimation Errors. IEEE 84th
Veh. Technol. Conf. (VTC-Fall), pp. 1-5. doi: 10.1109/VTCFall.2016.7881216.
Kong, L., Kaddoum, G., da Costa, D. B. & Bou-Harb, E. (2018a, June). On Secrecy Bounds
of MIMO Wiretap Channels with ZF detectors. 14th Int. Wireless Commun. Mobile
Comput. Conf. (IWCMC), pp. 724-729. doi: 10.1109/IWCMC.2018.8450477.
Kong, L., Vuppala, S. & Kaddoum, G. (2018b). Secrecy Analysis of Random MIMO
Wireless Networks Over α-μ Fading Channels. IEEE Trans. Veh. Technol., 67(12),
11654-11666. doi: 10.1109/TVT.2018.2872884.
Koyluoglu, O. O., Koksal, C. E. & Gamal, H. E. (2012). On Secrecy Capacity Scaling in

Wireless Networks. IEEE Trans. Inform. Theory, 58(5), 3000 - 3015.
Kuhestani, A., Mohammadi, A. & Masoudi, M. (2016). Joint optimal power allocation and
relay selection to establish secure transmission in uplink transmission of untrusted
relays network. IET Networks, 5(2), 30-36. doi: 10.1049/iet-net.2015.0035.
Kuhestani, A., Mohammadi, A., Wong, K., Yeoh, P. L., Moradikia, M. & Khandaker,
M. R. A. (2018a). Optimal Power Allocation by Imperfect Hardware Analysis in
Untrusted Relaying Networks. IEEE Trans. Wireless Commun., 17(7), 4302-4314.
doi: 10.1109/TWC.2018.2822286.
Kuhestani, A., Mohammadi, A. & Yeoh, P. L. (2018b). Optimal Power Allocation and
Secrecy Sum Rate in Two-Way Untrusted Relaying Networks With an External
Jammer. IEEE Trans. Commun., 66(6), 2671-2684. doi:
10.1109/TCOMM.2018.2802951.
Lai, L. & El Gamal, H. (2008). The Relay-Eavesdropper Channel: Cooperation for Secrecy.
Laneman, J. N., Tse, D. N. C. & Wornell, G. W. (2004). Cooperative Diversity in Wireless
Networks: Efficient protocols and outage behavior. IEEE Trans. Inform. Theory.,
50(12), 3062-3080.
Lee, K., Hong, J. P., Choi, H. H. & Levorato, M. (2018). Adaptive Wireless-powered
Relaying schemes with Cooperative Jamming for Two-hop Secure Communication.
IEEE Internet of Things Journal, 1-1. doi: 10.1109/JIOT.2018.2830880.
Leung-Yan-Cheong, S. & Hellman, M. (1978). The Gaussian wire-tap channel. IEEE Trans.
Inform. Theory., 24(4), 451-456. doi: 10.1109/TIT.1978.1055917.
Li, L., Chen, Z. & Fang, J. (2014a). On Secrecy Capacity of Gaussian Wiretap Chan-
nel Aided by A Cooperative Jammer. IEEE Signal Process. Lett., 21(11), 1356-1360.
doi: 10.1109/LSP.2014.2336803.
Li, W., Xin, M., Yue, M., Yinglei, T. & Yong, Z. (2013). Security-oriented transmis-
sion based on cooperative relays in cognitive radio. China, Commun., 10(8), 27-35.
doi: 10.1109/CC.2013.6633742.
Li, X., Chen, M. & Ratazzi, E. P. (2005, June). Space-time transmissions for wireless secret-
key agreement with information-theoretic secrecy. IEEE 6th Workshop on Signal Pro-
cess. Advances in Wireless Commun., pp. 811-815. doi:
10.1109/SPAWC.2005.1506252.
Li, Y., Xiao, L., Liu, J. & Tang, Y. (2014b, Nov.). Power control Stackelberg game
in cooperative anti-jamming communications. Int. Conf. Game Theory for Networks
(GAMENETS), pp. 1-6. doi: 10.1109/GAMENETS.2014.7043719.
Liang, Y., Poor, H. & Shamai, S. (2008). Secure Communication Over Fading Channels.
Lin, M., Ge, J. & Yang, Y. (2013a). An Effective Secure Transmission Scheme for AF Relay
Networks with Two-Hop Information Leakage. IEEE Commun. Lett., 17(8), 1676-
1679. doi: 10.1109/LCOMM.2013.062113.131012.
Lin, P.-H., Lai, S.-H., Lin, S.-C. & Su, H.-J. (2013b). On Secrecy Rate of the Generalized
Artificial-Noise Assisted Secure Beamforming for Wiretap Channels. IEEE J. Sel.
Areas Commun., 31(9), 1728-1740. doi: 10.1109/JSAC.2013.130907.
Lin, Z., Gao, Y., Zhang, X. & Yang, D. (2014, June). Stochastic Geometry Analysis of
Achiev- able Transmission Capacity for Relay-assisted Device-to-Device Networks.
Proc. IEEE Int. Conf. on Commun. Mobile and Wireless Networking Symp.
Liu, X., Li, B., Chen, H., Sun, Z., Liang, Y. & Zhao, C. (2019). Detecting Pilot Spoofing
Attack in MISO Systems With Trusted User. IEEE Commun. Lett., 23(2), 314-317.
doi: 10.1109/LCOMM.2018.2889491.
Liu, Y., Wang, L., Duy, T. T., Elkashlan, M. & Duong, T. (2015). Relay Selection for Security
Enhancement in Cognitive Relay Networks. IEEE Wireless Commun. Lett., 4(1), 46-
49. doi: 10.1109/LWC.2014.2365808.
Liu, Y., Li, J. & Petropulu, A. (2013). Destination Assisted Cooperative Jamming for
Wireless Physical-Layer Security. IEEE Trans. Inform. Forensics and Security., 8(4),
682-694. doi: 10.1109/TIFS.2013.2248730.
Long, H., Xiang, W., Wang, J., Zhang, Y. & Wang, W. (2013, Apr.). Coopera-
tive jamming and power allocation in three-phase two-way relaying wiretap sys-
tems. IEEE Wireless Commun. and Networking Conf. (WCNC)., pp. 4175-4179.
doi: 10.1109/WCNC.2013.6555247.
Long, H., Xiang, W., Wang, J., Zhang, Y. & Wang, W. (2014). Cooperative jamming and
power allocation with untrusty two-way relay nodes. Commun. IET, 8(13), 2290-2297.
doi: 10.1049/iet-com.2013.0580.
Lou, W. & Ren, K. (2009). Security, privacy, and accountability in wireless access networks.
IEEE Wireless Commun., 16(4), 80-87. doi: 10.1109/MWC.2009.5281259.
Luo, M. & Yin, Q. (2018, Oct). Security Transmission Designs in Two-Way Relay Wire-
less Networks. TENCON IEEE Region 10 Conf., pp. 0150-0155. doi: 10.1109/TEN-
CON.2018.8650482.
Mabrouk, A., Tourki, K., Hasna, M. O. & Hamdi, N. (2017). Performance analysis of se-
cure AF relay networks using cooperative jamming under outdated CSI. IET
Commun., 11(14), 2199-2205. doi: 10.1049/iet-com.2017.0412.
Madiseh, M. G., Neville, S. W. & McGuire, M. L. (2012). Applying Beamform-

ing to Address Temporal Correlation in Wireless Channel Characterization-Based
Secret Key Generation. IEEE Trans. Inf. Forens. Security, 7(4), 1278-1287.
doi: 10.1109/TIFS.2012.2195176.
Mavoungou, S., Kaddoum, G., Taha, M. & Matar, G. (2016). Survey on Threats and Attacks
on Mobile Networks. IEEE Access, 4, 4543-4572. doi:
10.1109/ACCESS.2016.2601009.
Middleton, D. (1977). Statistical-Physical Models of Electromagnetic Interference. IEEE

Trans. Electromagn. Compat., EMC-19(3), 106-127. doi:
10.1109/TEMC.1977.303527.
Mobini, Z., Mohammadi, M. & Tellambura, C. (2019). Wireless-Powered Full-Duplex Re-

lay and Friendly Jamming for Secure Cooperative Communications. IEEE Trans. Inf.
Forens. Security, 14(3), 621-634. doi: 10.1109/TIFS.2018.2859593.
Neagu, O. & Hamouda, W. (2016, April). Performance of smart grid communication in the
presence of impulsive noise. 2016 Intern. Conf. on Selected Topics in Mobile Wireless
Netw. (MoWNeT), pp. 1-5. doi: 10.1109/MoWNet.2016.7496614.
Neshenko, N., Bou-Harb, E., Crichigno, J., Kaddoum, G. & Ghani, N. (2019). Demys-
tifying IoT Security: An Exhaustive Survey on IoT Vulnerabilities and a First Em-
pirical Look on Internet-scale IoT Exploitations. IEEE Commun. Surveys Tuts., 1-1.
doi: 10.1109/COMST.2019.2910750.
Ng,Edward W.; Geller, M. (1969). A Table of Integrals of the Exponential Integral. J. of Res.
of the Nat. Bureau of Standards, 73B(3), 1-20. doi: 10.1109/TIFS.2018.2834301.
Oggier, F. & Hassibi, B. (2011). The Secrecy Capacity of the MIMO Wiretap Channel. IEEE
Trans. Inf. Theory., 57(8), 4961-4972. doi: 10.1109/TIT.2011.2158487.
Oohama, Y. (2007, June). Capacity Theorems for Relay Channels with Confidential Messages.
IEEE Int. Symp. Inform. Theory. ISIT., pp. 926-930. doi: 10.1109/ISIT.2007.4557113.
Park, K.-H., Wang, T. & Alouini, M.-S. (2013). On the Jamming Power Allocation for Secure
Amplify-and-Forward Relaying via Cooperative Jamming. IEEE J. Selected Areas in
Commun., 31(9), 1741-1750. doi: 10.1109/JSAC.2013.130908.
Pierrot, A. & Bloch, M. (2011). Strongly Secure Communications Over the Two-
Way Wiretap Channel. IEEE Trans. Inform. Forensics and Security, 6(3), 595-605.
doi: 10.1109/TIFS.2011.2158422.
Pittolo, A. & Tonello, A. M. (2013, March). Physical layer security in PLC networks:
Achiev- able secrecy rate and channel effects. IEEE 17th Intern. Symp. Power Line
Commun. and Its Appl., pp. 273-278. doi: 10.1109/ISPLC.2013.6525863.
Pittolo, A. & Tonello, A. M. (2014). Physical layer security in power line communication
networks: an emerging scenario, other than wireless. IET Commun., 8(8), 1239-1247.
doi: 10.1049/iet-com.2013.0472.
Popovski, P. & Simeone, O. (2009). Wireless Secrecy in Cellular Systems With

Infrastructure- Aided Cooperation. IEEE Trans. Inform. Forensics and Security.,
4(2), 242-256. doi: 10.1109/TIFS.2009.2020776.
Salem, A., Hamdi, K. A. & Alsusa, E. (2017). Physical Layer Security Over Correlated Log-
Normal Cooperative Power Line Communication Channels. IEEE Access, 5, 13909-
13921. doi: 10.1109/ACCESS.2017.2729784.
Sarr, N. B., Yazbek, A. K., Boeglen, H., Cances, J. P., Vauzelle, R. & Gagnon, F. (2017,
May). An impulsive noise resistant physical layer for smart grid communications.
2017 IEEE Intern. Conf. on Commun. (ICC), pp. 1-7. doi: 10.1109/ICC.2017.7996876.
Schneier, B. (1998). Cryptographic Design Vulnerabilities. IEEE Comput., 31(9), 29-33.
Serfling, R. (1980). Approximation Theorems of Mathematical Statistics. Wiley.
Shannon, C. (1949). Communication theory of secrecy systems. The Bell Syst. Technical J.,
28(4), 656-715. doi: 10.1002/j.1538-7305.1949.tb00928.x.
Sharif, M. & Hassibi, B. (2005). On the capacity of MIMO broadcast chan-
nels with partial side information. IEEE Trans. Inform. Theory., 51(2), 506-522.
doi: 10.1109/TIT.2004.840897.
Sheikholeslami, A., Goeckel, D., Pishro-Nik, H. & Towsley, D. (2012, Mar.). Physical layer
security from inter-session interference in large wireless networks. Proc. IEEE INFO-
COM., pp. 1179-1187. doi: 10.1109/INFCOM.2012.6195477.
Shiu, Y.-S., Chang, S. Y., Wu, H.-C., Huang, S.-H. & Chen, H.-H. (2011). Physical layer
security in wireless networks: a tutorial. IEEE Wireless Commun., 18(2), 66-74.
doi: 10.1109/MWC.2011.5751298.
Shongwey, T., Vinck, A. J. H. & Ferreira, H. C. (2014, March). On impulse noise and
its models. 18th IEEE Int. Symp. on Power Line Commun. and Its Appl., pp. 12-17.
doi: 10.1109/ISPLC.2014.6812360.
Shrestha, A. P., Riazul Islam, S. M., Dhakal, R. & Kwak, K. S. (2019, Jan). Physical
Layer Security for Cooperative Multihop Routing in Wireless Networks. IEEE 9th
Annu. Comput. Commun. Workshop Conf. (CCWC), pp. 0910-0915. doi: 10.1109/C-
CWC.2019.8666507.
Silva, N. M. O. & Cordero, C. V. (2017, Oct). Towards physical layer security sys-
tems design using game theory approaches. 2017 CHILEAN Conf. on Elect. Elec-
tron. Eng. Inform. and Commun. Technol. (CHILECON), pp. 1-6. doi:
10.1109/CHILE- CON.2017.8229597.
Simeone, O. & Popovski, P. (2008). Secure Communications via Cooperating Base Stations.
IEEE ASSP Mag. Commun. Lett., 12(3), 188-190. doi: 10.1109/LCOMM.2008.071836.
Song, C. (2018). Achievable Secrecy Rate of Artificial Fast-Fading Techniques and Secret-
Key Assisted Design for MIMO Wiretap Channels With Multiantenna Passive
Eavesdropper. IEEE Trans. Veh. Technol., 67(10), 10059-10063. doi:
10.1109/TVT.2018.2856764.
Soosahabi, R. & Naraghi-Pour, M. (2012). Scalable PHY-Layer Security for Distributed
Detection in Wireless Sensor Networks. Trans. Inf. Forens. Security, 7(4), 1118-1126.
doi: 10.1109/TIFS.2012.2194704.
Stanojev, I. & Yener., A. (2011, May). Cooperative jamming via spectrum leasing. Int. Symp.
Modeling and Optimization in Mobile, Ad Hoc and Wireless Networks (WiOpt), pp.
265- 272. doi: 10.1109/WIOPT.2011.5930026.
Stanojev, I. & Yener, A. (2011, Sept.). Recruiting multi-antenna transmitters as cooperative
jammers: An auction-theoretic approach. Annual Allerton Conf. Commun., Control,
and Computing (Allerton), pp. 1106-1112. doi: 10.1109/Allerton.2011.6120291.
Stanojev, I. & Yener, A. (2013). Improving Secrecy Rate via Spectrum Leas-
ing for Friendly Jamming. IEEE Trans. Wireless Commun., 12(1), 134-145.
doi: 10.1109/TWC.2012.120412.112001.
Sun, L., Zhang, T., Li, Y. & Niu, H. (2012). Performance Study of Two-Hop Amplify-and-
Forward Systems With Untrustworthy Relay Nodes. IEEE Trans. Vehicular
Technology., 61(8), 3801-3807. doi: 10.1109/TVT.2012.2207438.
Sun, L., Ren, P., Du, Q., Wang, Y. & Gao, Z. (2015). Security-Aware Relaying Scheme
for Cooperative Networks With Untrusted Relay Nodes. IEEE Commun. Lett., 19(3),
463-466. doi: 10.1109/LCOMM.2014.2385095.
Tekin, E. & Yener, A. (2008). The General Gaussian Multiple-Access and Two-Way Wiretap
Channels: Achievable Rates and Cooperative Jamming. IEEE Trans. Inform. Theory.,
54(6), 2735-2751. doi: 10.1109/TIT.2008.921680.
et al, G. M. (2011). Impacts of impulsive noise from partial discharges on wireless systems
performance: application to MIMO precoders. EURASIP J. Wireless Commun. Netw.,
2011(1), 186. doi: 10.1186/1687-1499-2011-186.
Tolossa, Y. J., Vuppala, S., Kaddoum, G. & Abreu, G. (2018). On the Uplink Secrecy Ca-
pacity Analysis in D2D-Enabled Cellular Network. IEEE Syst. J., 12(3), 2297-2307.
doi: 10.1109/JSYST.2017.2700438.
Tran, D., Tran, H., Ha, D. & Kaddoum, G. (2019). Secure Transmit Antenna Selection Pro-
tocol for MIMO NOMA Networks Over Nakagami-m Channels. IEEE Syst. J., 1-12.
doi: 10.1109/JSYST.2019.2900090.
Tran, T. T. & Kong, H. Y. (2014). CSI-Secured Orthogonal Jamming Method for Wire-
less Physical Layer Security. IEEE Commun. Lett., 18(5), 841-844. doi: 10.1109/L-
COMM.2014.040214.140109.
Vishwakarma, S. & Chockalingam, A. (2014, Feb.). MIMO decode-and-forward relay beam-

forming for secrecy with cooperative jamming. Nat. Conf. Commun. (NCC), pp. 1-6.
doi: 10.1109/NCC.2014.6811256.
Vuppala, S. & Abreu, G. (2013). Unicasting on the Secrecy Graph. IEEE Trans. Inf. Forens.
Security, 8(9), 1469 - 1481.
Vuppala, S., Tolossa, Y. J., Kaddoum, G. & Abreu, G. (2018). On the Physical Layer Security
Analysis of Hybrid Millimeter Wave Networks. IEEE Trans. Commun., 66(3), 1139-
1152. doi: 10.1109/TCOMM.2017.2776944.
Wang, C. & Wang, H. (2015). Robust Joint Beamforming and Jamming for Secure AF
Networks: Low-Complexity Design. IEEE Trans. Veh. Techn., 64(5), 2192-2198.
doi: 10.1109/TVT.2014.2334640.
Wang, C. & Wang, H.-M. (2014, June). Joint relay selection and artificial jamming power
allocation for secure DF relay networks. IEEE Int. Conf. Commun. Workshops (ICC),
pp. 819-824. doi: 10.1109/ICCW.2014.6881301.
Wang, C., Wang, H.-M. & Xia, X.-G. (2015a). Hybrid Opportunistic Relaying and Jamming
With Power Allocation for Secure Cooperative Networks. IEEE Trans. Wireless Com-
mun., 14(2), 589-605. doi: 10.1109/TWC.2014.2354635.
Wang, H. & Xia, X. (2015). Enhancing wireless secrecy via cooperation: signal design and
optimization. IEEE Commun. Mag., 53(12), 47-53. doi:
10.1109/MCOM.2015.7355565.
Wang, H., Liu, F. & Yang, M. (2015b). Joint Cooperative Beamforming, Jamming, and
Power Allocation to Secure AF Relay Systems. IEEE Trans. Veh. Technol., 64(10),
4893-4898. doi: 10.1109/TVT.2014.2370754.
Wang, H.-M., Luo, M., Xia, X.-G. & Yin, Q. (2013a). Joint Cooperative Beam-
forming and Jamming to Secure AF Relay Systems With Individual Power Con-
straint and No Eavesdropper’s CSI. IEEE Signal Process. Lett., 20(1), 39-42.
doi: 10.1109/LSP.2012.2227725.
Wang, H.-M., Luo, M., Yin, Q. & Xia, X.-G. (2013b). Hybrid Cooperative Beamforming
and Jamming for Physical-Layer Security of Two-Way Relay Networks. IEEE Trans.
Inform. Forensics and Security., 8(12), 2007-2020. doi: 10.1109/TIFS.2013.2287046.
Wang, H.-M., Zheng, T. & Mu, P. (2014a, June). Secure MISO wiretap channels with multi-
antenna passive eavesdropper via artificial fast fading. IEEE Int. Conf. Commun.
(ICC)., pp. 5396-5401. doi: 10.1109/ICC.2014.6884179.
Wang, H.-M., Zheng, T. & Xia, X.-G. (2015c). Secure MISO Wiretap Channels With Multi-
antenna Passive Eavesdropper: Artificial Noise vs. Artificial Fast Fading. IEEE Trans.
Wireless Commun., 14(1), 94-106. doi: 10.1109/TWC.2014.2332164.
Wang, L., Elkashlan, M., Huang, J., Tran, N. & Duong, T. (2014b). Secure Transmission with
Optimal Power Allocation in Untrusted Relay Networks. IEEE Wireless Commun.
Lett., 3(3), 289-292. doi: 10.1109/WCL.2014.031114.140018.
Wang, T. & Yang, Y. (2013, April). Analysis on perfect location spoofing attacks us-
ing beamforming. IEEE Proc. INFOCOM., pp. 2778-2786. doi: 10.1109/INF-
COM.2013.6567087.
Wang, T. & Yang, Y. (2012, Oct). Enhancing wireless communication privacy with artifi-
cial fading. IEEE 9th Int. Conf. Mobile Adhoc and Sensor Syst. (MASS)., pp. 173-181.
doi: 10.1109/MASS.2012.6502515.
Waqas, M., Ahmed, M., Li, Y., Jin, D. & Chen, S. (2018). Social-Aware Secret Key Genera-
tion for Secure Device-to-Device Communication via Trusted and Non-Trusted
Relays. IEEE Trans. Wireless Commun., 17(6), 3918-3930. doi:
10.1109/TWC.2018.2817607.
Wen, Y., Huo, Y., Ma, L., Jing, T. & Gao, Q. (2019). A Scheme for Trustworthy Friendly
Jammer Selection in Cooperative Cognitive Radio Networks. IEEE Trans. Veh.
Technol., 1-1. doi: 10.1109/TVT.2019.2895639.
Wyner, A. D. (1975). The Wire-Tap Channel. Bell System Technical Journal, 54(8), 1355–
1387. doi: 10.1002/j.1538-7305.1975.tb02040.x.
Xing, H., Chu, Z., Ding, Z. & Nallanathan, A. (2014, Dec.). Harvest-and-jam: Improving
security for wireless energy harvesting cooperative networks. IEEE Global Commun.
Conf. (GLOBECOM)., pp. 3145-3150. doi: 10.1109/GLOCOM.2014.7037289.
Yang, J., Kim, I.-M. & Kim, D. I. (2013). Optimal Cooperative Jamming for Multiuser
Broadcast Channel with Multiple Eavesdroppers. IEEE Trans. Wireless Commun.,
12(6), 2840-2852. doi: 10.1109/TWC.2013.040413.120972.
Yang, J., Kim, I.-M. & Kim, D. I. (2014). Joint Design of Optimal Cooperative Jamming
and Power Allocation for Linear Precoding. IEEE Trans. Commun., 62(9), 3285-3298.
doi: 10.1109/TCOMM.2014.2345659.
Yener, A. & He, X. (2010). Cooperation With an Untrusted Relay: A Secrecy Perspective.
Zeng, M., Nguyen, P., Dobre, O. & Poor, H. V. (2019). Securing Downlink Massive MIMO
NOMA Networks with Artificial Noise. IEEE J. Sel. Topics. Signal Process., 1-1.
doi: 10.1109/JSTSP.2019.2901170.
Zhang, J., Duong, T., Marshall, A. & Woods, R. (2016a). Key Generation from Wireless
Channels: A Review. IEEE Access., PP(99), 1-1. doi:
10.1109/ACCESS.2016.2521718.
Zhang, J., Firooz, M. H., Patwari, N. & Kasera, S. K. (2008). Advancing Wireless Link
Signatures for Location Distinction. Proc. 14th ACM Int. Conf. Mobile Comput. and
Networking, (MobiCom ’08), 26–37. doi: 10.1145/1409944.1409949.
Zhang, M., Liu, Y. & Zhang, R. (2016b). Artificial Noise Aided Secrecy Information and
Power Transfer in OFDMA Systems. IEEE Trans. Wireless Commun., PP(99), 1-1.
doi: 10.1109/TWC.2016.2516528.
Zhang, Q., Huang, X., Li, Q. & Qin, J. (2015). Cooperative Jamming Aided Robust Secure
Transmission for Wireless Information and Power Transfer in MISO Channels. IEEE
Trans. Commun., 63(3), 906-915. doi: 10.1109/TCOMM.2015.2405063.
Zhang, R., Song, L., Han, Z. & Jiao, B. (2011, June). Distributed Coalition Formation of
Relay and Friendly Jammers for Secure Cooperative Networks. IEEE Int. Conf.
Commun. (ICC)., pp. 1-6. doi: 10.1109/icc.2011.5962513.
Zhang, S., Jin, L., Lou, Y., Huang, K. & Zhong, Z. (2018, Oct). High-Rate Secret Key Gener-
ation Method Using Two-Way Random Signal. Int. Conf. Cyber-Enabled Distrib.
Com- put. and Knowl. Discovery (CyberC), pp. 32-324. doi:
10.1109/CyberC.2018.00017.
Zhu, Y., Suo, D. & Gao, Z. (2010, Aug). Secure Cooperative Spectrum Trading in Cognitive
Radio Networks: A Reversed Stackelberg Approach. Int. Conf. Multimedia Commun.
(Mediacom)., pp. 202-205. doi: 10.1109/MEDIACOM.2010.33.

ATALLAH Michael

Transféré par

Droits d'auteur :

Formats disponibles

ATALLAH Michael

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

ATALLAH Michael

Transféré par

Droits d'auteur :

Formats disponibles

Physical Layer Security Solutions Against Passive and Colluding

Eavesdroppers in Large Wireless Networks and Impulsive Noise

MANUSCRIPT-BASED THESIS PRESENTED TO ÉCOLE DE

MONTREAL, AUGUST 8, 2019

ÉCOLE DE TECHNOLOGIE SUPÉRIEURE

Michael Atallah, 2019

THIS THESIS HAS BEEN EVALUATED

BY THE FOLLOWING BOARD OF EXAMINERS

M. Georges Kaddoum, Thesis Supervisor

M. Julien Gascon-Samson , President of the Board of Examiners

M. Chamseddine Talhi, Member of the jury

M. Yousef R. Shayan, External Independent Examiner

THIS THESIS WAS PRESENTED AND DEFENDED

IN THE PRESENCE OF A BOARD OF EXAMINERS AND THE PUBLIC

ON JULY 17, 2019

AT ÉCOLE DE TECHNOLOGIE SUPÉRIEURE

Dans le deuxième chapitre, nous examinons les performances de capacité de conﬁdentialité

Mots-clés: Couche physique, brouillage, transmission sectorielle, bruit impulsif.

In the second chapter, we investigate the secrecy capacity performance of an amplify-and-

Keywords: Physical Layer, Jamming, Sectoral Transmission, Impulsive Noise.

CHAPTER 1 LITERATURE REVIEW................................................................................5

CHAPTER 2 SECRECY CAPACITY SCALING WITH

CHAPTER 3 SECRECY ANALYSIS IN WIRELESS NETWORK WITH

CHAPTER 4 DESIGN AND PERFORMANCE ANALYSIS OF SECURE

CHAPTER 5 SECRECY ANALYSIS OF WIRELESS SENSOR NETWORK

CONCLUSION AND RECOMMENDATIONS.....................................................................83

Figure 1.1 Wireless wiretap system model........................................................................6

Figure 1.3 The spoofer is transmitting a deceiving signal to a legitimate

Figure 1.4 Representation of a general MIMO wiretap channel......................................12

Figure 1.5 Representation of trusted (distinct relay and eavesdropper) relay

Figure 1.6 Representation of untrusted (co-located relay and eavesdropper)

Figure 1.7 Eavesdroppers’ passive behavior....................................................................17

Figure 1.8 Eavesdroppers’ colluding behavior................................................................18

Figure 1.9 Representation of a network with a jammer...................................................19

this ﬁgure, T = 9, K = 3 and G = 3..................................................................44

σk = 0.95 and μs = μk = 1................................................................................66

σs = σk = 1.1 and μs = μk = 0.69.....................................................................67

AWGN Additive White Gaussian Noise

CDF Cumulative Distribution Function

CRN Cognitive Radio Network

CSI Channel State Information

DBF Distributed Beamforming

ESR Ergodic Secrecy Rate

FDJ Full Duplex Jammer

HDJ Half Duplex Jammer

MC-DCSK Multi-Carrier Differential Chaos Shift Keying

MIMO Multiple-Input Multiple-Output

MISO Multiple-Input Single-Output

NOMA Non-Orthogonal Multiple Access

PDF Probability Density Function

PLS Physical Layer Security

RSS Received Signal Strength

SINR Signal-to-Interference-plus-Noise Ratio

SNR Signal-to-Noise Ratio

SOP Secrecy Outage Probability

SPCA Sequential Parametric Convex Approximation

Contributions and Outline

Chapter 4 proposes a novel protocol that implements the location-based multicasting

– J5: G. Kaddoum, H. Tran, L. Kong and M. Atallah, "Design of Simultaneous Wireless