Commnet19v2

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Physical Layer Security of an Amplify-and-Forward

Energy Harvesting-based mixed RF/UOW System


Elmehdi Illi1 , Faissal El Bouanani1 , and Fouad Ayoub3
1
ENSIAS, Mohammed V University in Rabat, Morocco
3
CRMEF Kenitra, Morocco
Emails: {elmehdi.illi, f.elbouanani}@um5s.net.ma, ayoub@crmefk.ma

Abstract—This paper investigates the secrecy outage perfor- electromagnetic wave for conveying both information sig-
mance of an energy harvesting-based dual-hop amplify-and- nal as well as energy one, exploited to power the node’s
forward (AF) mixed radio-frequency/underwater optical wireless battery. Interestingly, two practical system designs, namely
communication (RF/UOWC) system. A single-antenna source
node (S) is considered, communicating with one legitimate time switching (TS), and power splitting (PS), allows the
destination node (D) with the aid of a multi-antenna AF relay (R) simultaneous processing of information and energy signals.
device. In this setup, the relay node receives the incoming signal Underwater optical wireless communication (UOWC) tech-
from S via an RF link, which is subject to Rayleigh fading, then nology has attracted remarkable interest as a promising tech-
performs maximal-ratio-combining (MRC) followed by a fixed- nology for high-speed underwater transmissions [8]. It may
gain amplification, before transmitting it to the destination via
a UOWC link, subject to mixture Exponential-Gamma fading. provide a considerably higher data rate (i.e., tens of Gbps)
Assuming the presence of a malicious eavesdropper attempting at midway propagation distance [9], in addition to its low
to intercept the S-R hop, a closed-form expression for the secrecy latency, highly secure communication links, as well as energy
outage probability is retrieved. The derived results provide useful efficiency [10]. However, in spite of all of these promising
insights into the influence of key system parameters on the secrecy features, light propagation in the marine environment is signif-
outage performance. Our analytical results are corroborated
through computer simulations, which verifies their validity. icantly corrupted by the turbulence phenomenon [11], caused
by pressure and temperature inhomogeneities encountered in
I. I NTRODUCTION the underwater medium as well as by ocean currents [10],
[9]. Interestingly, very few works reported on the modelling
In the last decade, with the emergence of the internet of of UOWC channel’s turbulence impairment. Specifically, the
things (IoT) as well as machine-to-machine (M2M) paradigms, authors in [10] proposed the mixture Exponential-Gamma
future wireless networks are targetting new critical services UOWC turbulence-induced fading model, where the vast ma-
such as remote healthcare, surveillance, transportation, etc [1]. jority of turbulence conditions in the underwater medium
In this regard, energy efficiency has been recently among (weak to strong), over both fresh and salt waters.
hottest spots in the design of future cellular networks such It is recalled that due to the broadcast nature of the RF links,
as 5G [2], as well as in wireless sensors network (WSN), the security level of next-generation wireless communication
where power consumption is becoming a major concern in systems has been of paramount importance recently because of
addition to other performance metrics [2], [3]. Indeed, the its vulnerability to intrusion threats of potential eavesdroppers
connected devices in M2M network or WSN are low-cost, [12]. In fact, higher layers consider the security paradigm as
with limited energy resources (e.g., battery), and are typically an application of cryptographic and authentication protocols.
deployed in difficult-to-access scenarios , e.g., structural health Nevertheless, these security protocols are performed assuming
monitoring, mine tunnels, that sense and collect information ideal security at the physical layer [13]. Interestingly, the
from the surrounding area [4], [5], which makes the battery notion of physical layer security, introduced pioneeringly by
recharging or replacement impractical in most of these cases. Wyner in [14], aims at establishing perfectly secure communi-
Recently, energy harvesting has been widely advocated as a cation by exploiting the random characteristics of the wireless
promising solution to overcome the finite-energy dilemma in channel alongside with channel coding [13].
contemporary wireless networks, more particularly in energy- Due to the incurred limitations of OWC by the turbulence,
constrained networks [6]. In practice, energy can be harvested relay-based mixed dual-hop RF/OWC systems have been
from environmental sources such as solar or thermal, or from broadly inspected in the literature as an effective solution to
other wireless radio-frequency (RF) based energy sources [3], mitigate optical channel’s limitations. In this mixed architec-
[7]. In M2M and WSN contexts, the communicating nodes ture, the source transmits the information signal through an RF
harvests energy from either the access points or continuously- link to the relay node. This latter converts the received infor-
powered fixed nodes/control stations [2]. Simultaneous wire- mation signal into an optical light wave and conveys it through
less information and power transfer (SWIPT) technique has either a free-space optical (FSO), visible light communication
attracted significant attention, since it uses the same emitted (VLC), or UOW link to the destination node. The relaying can
be performed with either amplify-and-forward (AF) or decode- to the battery-equipped relay (R), while in the remaining part
and-forward (DF) protocols [15]. In this context, several works of this time slot (i.e., (1 − ε) T0 ), the source node radiates an
such as [16], [17], [18] dealt with the performance analysis energy signal to the relay for energy harvesting purpose. The
of a dual-hop mixed RF/OWC system, where the RF link was relay employs Nr -branch MRC diversity technique for both
subject to Nakagami-m, while the OWC was either a FSO signals, namely the information and the energy one. Thereafter,
or UOW link, subject to Gamma-Gamma, Málaga, or mixture it converts the received electromagnetic wave into an electrical
Exponential-Gamma turbulence fading models. Interestingly, power, stores it on its finite-capacity battery, and exploits it
from a PHY security perspective, several contributions such to forward the information signal, after amplifying it by a
in [9], [18], [19], [20] analyzed the secrecy performance of fixed gain, to the legitimate destination (D) through an UOW
mixed set-ups of either RF/FSO or RF/UOWC systems. link. The end-to-end communication is performed under the
Motivated by the above discussion, we aim herein to malicious attempt of a distant eavesdropper (E) to overhear
determine the secrecy outage probability performance of an the RF communication side.
energy harvesting-based RF/UOWC system with a multiple-
antenna AF relay in the presence of a single wiretapper.
In the considered network, the source node, consisting of A. First time slot: RF links
a distant terrestrial control station, communicates with an 1) Information signal transmission and interception:
underwater destination node (e.g., submarine, sensor) through Within this time slot, S sends the information signal xs to
a moving relay node. The relay performs energy harvesting R during the first εT0 seconds. By its turn, E attempts to
and information forwarding to the destination node via an overhear the broadcasted information signal from S during
UOW link, after amplifying it with a fixed gain. It is worthy this period.
to mention that the considered setup is of practical use in
The instantaneous SNR at each relay branches (Ri )i=1,...,Nr
relay-based wireless sensors networks (WSN) in commercial
as well as at E can be generalized as
as well as military applications. To the best of our knowledge,
the secrecy performance of dual-hop energy harvesting-based PS 2
mixed RF-UOWC systems has not been explored yet in the γSX = |hX | , X = {Ri , E}, (1)
dδX N0X
literature.
The main contributions of this paper are summarized below: with PS denotes the transmit power, dX is the S-X distance,
• The statistical properties of the end-to-end signal-to-noise δ is the free space path-loss exponent, hX , refers to the fading
ratio (SNR) of the legitimate link, as well as of the overall coefficienti of the S-X linkshwith average fading powers ΩX =
wiretap channel SNR are derived, by considering time-
h i
2 2
E |hX | , and N0X = E |nX | stands for power spectral
switching (TS) energy harvesting at the relay.
density (PSD) of the additive white Gaussian noise (AWGN)
• The secrecy outage probability (SOP) performance of
process. It is noteworthy to mention that dRi = dR .
the analyzed network is derived, formulated in terms
of the hypergeometric incomplete Fox’s H-function, and In the sequel, Nakagami-m fading model is taken into
involving the system parameters. Both legitimate and account for representing the channel gain distribution for both
wiretap RF links are subject to Nakagami-m fading. S-Ri and S-E links. Consequently, the received SNR for
Particularly, independent and not identically distributed the S-Ri and S-E links are Gamma-distributed, and their
(i.n.i.d) fading amplitudes are considered for the main respective PDF/CDF expressions are given as [21]
link, while the UOWC channel is modeled by the mixture m
z mv −1
  
Exponential-Gamma fading. mX mX
fγSX (z) = exp − z , (2)
The remainder of this paper is organized as follows: Section γX Γ (mX ) γX
 
II introduces the system and channel models, while in Section γinc mX , m
γX z
X

III, statistical properties of the considered communication FγSX (z) = , (3)


Γ (mX )
system are retrieved. In Section IV, an analytical expression
for the SOP is derived, whereas Section V is dedicated
with Γ (.) and γinc (., .) being the Gamma and the lower-
to illustrative numerical examples, followed by discussions.
incomplete Gamma functions, respectively [22, Eqs. (8.310.1),
Finally, Section VI concludes the paper.
(8.350.1)], and the average SNRs can be expressed from (1)
II. S YSTEM AND CHANNEL MODELS as
A mixed dual-hop AF energy harvesting-based communi- PS
γX = δ ΩX . (4)
cation system operating under both RF and UOWC links is dX N0X
considered. The communication protocol is divided into two
time slots, namely T0 and T1 . The source node (S) (e.g., Interestingly, we consider in this work Pthe i.n.i.d fading
Nr
boat, ground control station, etc), assumed to be continuously amplitudes. The combined SNR is γout = i=1 γi , with γi
powered through wirelines, transmits during a ratio of the first being the SNR at the ith branch. On the other hand, as it is
time slot (i.e., εT0 , with 0 < ε < 1), the information signal complicated enough to derive the closed-form PDF of the sum
of i.n.i.d Gamma random variates, an accurate approximation with
for the PDF/CDF, given in [23] will be considered mI T1 dδR
 mI mI −1   Ψ= Nr
. (13)
mI z z θεT0 Ps
P
Ωi
fγSR (z) ≈ exp −mI , (5) i=1
γI Γ (mI ) γI
 
γinc mI , mI γz B. Second time slot: UOW link
I
FγSR (z) ≈ , (6)
Γ (mI )
The R-D hop is modeled by a mixture Exponential-Gamma
Nr
PS P E2 [Φ] model, where the PDF of the received light irradiance IRD is
where γ I = dδR N0R
ΩR i , m I = E[Φ2 ]−E2 [Φ] ,Φ =
i=1 expressed as [10]
Nr
P 2
|hRi | ,
 
I
i=1 ω

I
 exp − β
α−1
n X v1 vNr −2      fIRD (I) = λ exp − λ + (1 − ω)I β α Γ (α)
, I > 0.
X X n v 1 v Nr −2
E [Φn ] = ... ... (14)
v1 v2 vNr −1 with ω ∈ [0, 1] denotes the mixture weight factor, λ > 0
v1 =0 v2 =0 vNr −1 =0
2vNr −1 stands for the mean of the Exponential distribution, and α >
h i h i h i
2(n−v1 ) 2(v −v )
× E |hR1 | E |hR2 | 1 2 ...E hRNr ,
0 and β > 0 depict Gamma distribution’s shape and scale
(7) parameters, respectively.
Γ mRi + n2
  
ΩR i 2
n
Considering coherent detection, the R-D SNR is related to
n
E [|hRi | ] = . (8) the irradiance as follows [16]
Γ (mRi ) mRi
and E[.] refers to the expectation operator. ηPR IRD
γRD = , (15)
2) Energy Harvesting: The source node exploits the re- N0D
where 0 < η < 1 indicates the photodetector efficiency, and
maining ratio of T0 (i.e., εT0 ) to radiate energy signals to R,
N0D denotes the noise’s PSD at D’s detector.
which combines the received energies through MRC combiner.
Consequently, the harvested energy during the time slot of III. S TATISTICAL PROPERTIES
duration εT0 is expressed as
In this section, the CDF expression in closed-form for
2
θεT0 PS hSR the end-to-end SNR of the analyzed RF/UOWC system is
ER = , (9) presented.
dδR
Since the relay node performs AF relaying, the total re-
with hSR = [hR1 , ..., hRNr ] denotes the fading amplitudes of ceived SNR at D is expressed as [15]
the S-Ri links, θ stands for the conversion efficiency of the γSR γRD
relay’s energy harvester, and ||.|| refers to the Frobenius norm. γeq = , (16)
γRD + C
The relay R will exploit the abovementioned harvested with C being a fixed-gain relaying constant.
energy for forwarding the signal to D during the time slot
T1 . Actually, assuming a finite battery storage model, two
cases might be distinguished in this scenario, namely: (i) the Proposition 1. The CDF expression of the end-to-end SNR
harvested energy exceeds the battery storage, (ii) the collected for the considered system is expressed as
energy is less than the battery capacity. In the former case, the
    
(1) BR (2) BR
relay will forward to D using the whole battery energy, while Fγeq (z) = Fγeq (z) FPr +Fγeq (z) 1 − FPr ,
T1 T1
in the latter, only the available harvested energy at R will be (17)
exploited. That is, the relay’s transmit power is formulated as where

Fγ(1)

PE ; if ER < BR (z) = 1 − ωE(z, 1, λ) − (1 − ω) E(z, α, β), (18)
PR = , (10) eq
PB ; if ER ≥ BR
with
where PE = ETR1 , and PB = BTR1 , with BR denoting the relay
battery capacity.  x m mX I −1 X
j  
Similarly to (5) and (6), the PDF/CDF of the power PE can 1 ∆Ψz − γI z j
E(z, x, y) = e I
be approximated, using (9) and (10), as Γ (x) Γ (mI ) y k
j=0 k=0
y mI −1  j−k
fPR (y) ≈ ΨmI exp (−Ψy) , (11) mI
γI z
 
Γ (mI ) ∆Ψz −; −
× G3,0
0,3 .
γinc (mI , Ψy) j! y 0, mI − x, k − x; −
FPR (y) ≈ , (12) (19)
Γ (mI )
and and
 m E  l−j
 j−k " mE   mI −1 l mI
m
m I −1 X
l mI # γE − γE 1
−1
X X γI
γI z
(1−ω) T (x, y) = e E 2RS
Γ(α) W
(α, β)
mI
− z
X
Fγ(2) (z) = 1−e γI
, Γ (x) Γ (mE )
l=0 j=0
j!(l − j)!
eq
j=0 k=0
k!(j − k)! +ωW (1, λ)
mXE −1
 m −1−n
2 RS − 1
E  
(20) mE − 1

−; −
 × Υ1−n−l+j
with W (x, y) = 2,0
G0,2 (y) , 2 RS mE n
n=0
x, k; − 
j − l − n, 1, Υ 2R − 1 ; −
 

(ai )i=1:p

2,1 ∆
Gm,n z , p ≥ n, and q ≥ m, refers to the × M1,2 ,
p,q
(bk )k=1:q PB yΥ (x, 1, 1), (j, 1, 1) ; −
Meijer’s G-function [24, Eqs. (1.111), (1.112)], (y) = P∆ z, (24)
By
∆ = κCm
 
γI
I
, κ = E [I] /µr , E [I] = ωλ + (1 − ω)αβ denoting mI mE m,n (ai , Ai , αi )i=1:p
with Υ = γ + γ 2Rs , and Mp,q x
the average value of the received light irradiance, and I E (bk , Bk , βk )k=1:q
µr = ηE[I]
N0D .
is the upper incomplete Fox’s H-function defined as [25].
Proof: The proof is provided in Appendix A. Proof: See Appendix B.
Remark 1. It is worthy to mention that at RS = 0, the SOP
formula corresponds to the intercept probability (IP) metric.
IV. S ECRECY ANALYSIS
B. Diversity Order
The closed-form SOP expression of the considered PS
At high S-R average SNR values (i.e., ξ = N0R → ∞), the
RF/UOWC system is derived in the following subsection. SOP formula can be expanded as
−Gd
Psop ∼ (Gc ξ) , (25)
A. Secrecy outage probability
which can be formulated similarly as
The SOP is defined as the probability that the secrecy log (Psop ) ∼ −Gd (log (Gc ) + log (ξ)) (26)
capacity, being the difference between the main link capacity
(i.e., S-R-D link) and he wiretap link one, drops below a given with Gc and Gd referring to the coding gain and the diversity
threshold rate RS . In this case, the eavesdropper believably order, respectively.
wiretaps the confidential message. Mathematically, SOP is be So as to evaluate this latter, we consider the equation (26)
expressed as [18] evaluated at two high average SNR values, namely ξ1 and ξ2 .
That is, the diversity order Gd is obtained as
Z ∞
log (Psop2 ) − log (Psop1 )
Fγeq 2R (z + 1) − 1 fγSE (z)dz.

Psop = (21) ∼ Gd , (27)
0 log (ξ1 ) − log (ξ2 )
which provides useful insights on the impact of the involved
Proposition 2. The SOP of the considered communication average SNRs on the system performance.
system can be expressed as
V. N UMERICAL RESULTS
 
BR The derived analytical results, which are endorsed by re-
Psop = FPr (1 − ωS (1, λ) − (1 − ω) S (α, β))
T1 spective Monte Carlo simulations, are employed to analyze
 
BR
 the secrecy performance of the considered energy harvesting-
+ 1 − FPr (1 − ωT (1, λ) − (1 − ω) T (α, β)) , based mixed RF/UOWC system. Without loss of generality, the
T1
(22) system and channel parameters are set to, mE = 2, ΩE = 2
for the S-E link. The number of antennas at the relay is
where chosen as Nr = {2, 3, 4}, mRi = {1, 2, 3, 3.5, 1}, ΩRi =
{1.5, 2.5, 2, 3} are the fading parameter and average power
 of the S-R hop, α = 6.7615, β = 0.3059, λ = 0.1992,
 x  m E
κmI ΨC mE m I −1 Xj 
γI y γE X j ω = 0.5717, η = 0.7, µr = 5 dB stand for the R-D link
S(x, y) =  
Γ (x) Γ (mI ) mE 1
−1 k parameters. Futhermore, the harvester conversion efficiency
Γ (mE ) e γE R
2 S j=0 k=0
 j−k was fixed to θ = 0.7, the transmit power PS = 10 W, the
mI mXE −1  RS
mE −1−n time slots T = T = 1s, the free space path-loss coefficient
2 −1 0 1

γI mE − 1
× was fixed to δ = 2, and the battery size BR = 500 mAh ×
j! n Υ1+x+j−k+n 2RS mE
n=0
 V. In addition, the threshold rate is fixed to RS = 0 (intercept
−x − j + k − n, 1, Υ 2RS − 1 ; − probability case).
 
3,1 ∆Ψ
× M1,3 ,
y (0, 1, 1), (mI − x, 1, 1), (k − x, 1) ; − In fig. 1, the IP is depicted as a function of the transmit
(23) power to noise ratio NP0R S
in dB, for various values of relay
10 0

10 -1

10 -2

PIP
10 -3

10 -4

10 -5

Fig. 1: IP performance versus PS /N0R , for various values of 10 -6

antennas Nr , NP0E
30
S
= 3 dB, dE = 20 m, and ε = 0.7. 25
50
20 40
15 30
20
PS dR [m] 10
antennas Nr = {2, 3, 4}, and considering N0E = 3 dB, 5
10 PS
N0R [dB]
0
dR = 10 m, and dE = 20 m. One can remark evidently that
the analytical results plotted from (22) by setting RS = 0,
match tightly with its Monte Carlo counterpart, which shows
the accuracy of the adopted approximation of the PDF (22) Fig. 2: IP performance versus PS /N0R , and the distance dR ,
used in the SOP derivation. Furthermore, it is remarkably with NP0E
S
= 3 dB, dE = 20 m, and ε = 0.7.
noticed that the secrecy performance of the system gets better
by increasing the ratio NP0R S
. That is, the greater the harvested
power or the lower is the relay’s receiver noise power, the more
secure the system is. In addition, the IP drops significantly by
increasing the number of receive antennas at the relay.
In fig. 2 shows the IP performance in three dimensions
(3D) versus the S-R distance dR as well as NP0R S
. One can
ascertain from this figure that the IP metric is affected by
the distance dR . In fact, the farther the source and the relay
node are, the lower the received signal power due to path-loss,
and consequently, the greater is the probability to intercept
10 0
the information signal by the wiretapper E. For instance, we
notice that at lower NP0R
S
(e.g., lower transmit power), the path-
loss affects significantly the IP performance, that reaches about 10 -2

1 for NP0R
S
= 0 dB and dR = 30 m.
PIP

Fig. 3 depicts the secrecy performance versus the harvester 10 -4


conversion efficiency ε and the transmit power to noise ratio
PS
N0R . It can be seen that the IP metric of the system increases
10 -6
slightly as a function ε, in particular for lower NP0R S
values 1 50
PS
(e.g., N0R = 0 dB). On the other hand, the decrease of the 0.8 40

IP versus ε for higher values NP0R S


is significant. That is, the 0.6 30

0.4 20 PS
greater the time ratio dedicated to energy harvesting, the more α N0R [dB]
0.2 10
secure the overall system is. 0 0
In fig. 4 shows the IP performance in three dimensions
(3D) versus the NP0R S
as well as NP0E S
, for fixed relay and
eavesdropper distances to the source. It is evidently seen that
Fig. 3: IP performance versus the harvester efficiency ε and
the IP is significantly affected by the S-R and S-E links signal
PS /N0R , with NP0E
S
= 3 dB, and dE = 20 m.
to noise ratio. In particular, we can remark that for lower NP0E S

PS
values, the IP drops notably by increasing the N0R . On the
other hand, at higher S-E average SNR values, even increasing
closed-form expression for the secrecy outage probability was
derived, expressed by either the incomplete Fox’s H-function,
based on which the impact of the system parameters on the
overall secrecy performance was inspected. It was shown that
the system’s security improves significantly by increasing the
number of antennas, the legitimate link parameters (i.e., aver-
10 0
age S and R transmit powers (i.e., average S-R SNRs/transmit
power) as well as decreasing the S-R distance. Furthermore,
10 -2 it has been shown also that the harvester efficiency impacts
significantly the secrecy performance, particularly for higher
PIP

transmit power values. Also, it is shown that the secrecy


10 -4
performance of the system deteriorates in spite of varying the
S-R average SNR while keeping the S-E one in a high value.
10 -6 50
50
Also, the achieved diversity order equals 1, independently from
40
40 the number of antennas at the relay.
30
30
20
20
PS
PS
[dB] [dB]
N0R 10 10 N0E VII. A PPENDIX A: P ROOF OF P ROPOSITION 1
0 0
Given the end-to-end SNR expression in (16), its CDF can
be expressed as [18]
Fig. 4: IP performance versus PS /N0R and PS /N0E , with Z ∞   
dR = 10 m, dE = 20 m, and ε = 0.7. C
Fγeq (z) = FγSR z 1 + fγRD (x)dx. (28)
0 x

In order to derive the CDF expression of the total end-to-


end SNR, we have to derive first the PDF of the R-D hop
in the presence of energy harvesting. Two scenarios can be
distinguished as:
A. First case: ER < BR

In this scenario, we assume that the harvested energy at the


relay is less than the relay’s battery capacity, that is ER < BR .
2
θεT0 PS ||hSR ||
That is, PR = PE = T1 dδR
.
Indeed, one can notice that the SNR in (15) is the product
of two random variates, namely PR and W = ηI N0D . For the
RD

latter, its PDF is given as follows [9]


Fig. 5: Diversity order of the considered system for various
values of Nr . fW (z) = ωU (1, λ) + (1 − ω)U (α, β) , z > 0, (29)
where
 x  
the S-R can not improve the security level of the system, 1 κ x−1 κ
U (x, y) = z exp − z , (30)
where the IP tends to 1. Γ (x) y y
Fig. 5 depicts the system’s diversity order, evaluated numer- and µr = Nη0 E [I] denotes the average value of W .
ically through (27), by varying ξ1 from 0 to 90 dB, while we Relying on the PDF of W given in (29), the PDF of γRD
fixed ξ2 = 100 dB. It can be remarkably seen that the achieved can be expressed as
diversity order of the overall communication system Gd = 1, Z ∞
independently from other system parameters (e.g., number of (1) 1 z
fγRD (z) = fPR fW (x) dx. (31)
antennas, fading parameter,...). That is, these channel parame- 0 x x
ters do not influence on the secrecy performance of the system
at higher NP0R
S
values. By involving the expressions (29) and (11) in the equation
VI. C ONCLUSION above, and using some algebraic manipulations, the PDF of
γRD is attained as
The secrecy outage probability of a dual-hop energy
harvesting-based AF mixed RF/UOW system was inspected. A fγ(1)
RD
(z) = ωA1 (z, 1, λ) + (1 − ω)A1 (z, α, β),
where the constant PB . Based on (29), the PDF of γRD herein is
x given as
z x−1 2,0
  
1 κΨ κΨz −; −
A(z, x, y) = G ,
Γ (x) y Γ (mI ) 0,2 y 0, mI − x; − f (2) (z) = ω U (1, λ) + (1 − ω) U (α, β) , z > 0. (38)
γRD
(32) PBx PBx
with Ψ given in (13). where U (x, y) is defined in (30). Given the above result, the
Consequently, the CDF of the end-to-end SNR assuming end-to-end SNR of the considered system for the case of a
that ER < BR is fixed transmit power PB is given in [9, Eq. (13)] as shown in
(20). Hence, the total CDF of the end-to-end SNR considering
Z ∞   
the two investigated cases can be expressed as
C
Fγ(1) (z) = FγSR z 1 + fγ(1) (x)dx. (33)    
eq
0 x RD
(1) BR (2) BR
Fγeq (z) = Fγeq (z) Pr Pr < +Fγeq (z) Pr Pr ≥ .
T1 T1
Now, by involving (6) and (32) into the above equation, and (39)
using the finite sum representation of the lower-incomplete Consequently, by involving (18), (20), and (12) into the
Gamma function in [22, Eq. (8.352.1)] yields above equation, the CDF of the end-to-end SNR considering
  j
 both cases is expressed as given in (17), which concludes the
m I −1
mI C
Z ∞ m
γI z 1 + x result of Proposition 1.
e γI ( x )
− I z 1+ C
X
Fγ(1) (z) = 1− fγ(1) (x)dx,
eq
0 j=0
j! RD
VIII. A PPENDIX B: P ROOF OF P ROPOSITION 2
| {z }By plugging the expressions (17) and (2), with X = E, into
I(z)
(34) (21), the SOP can be expressed as follows
with Z ∞
I(z)=ωE(z, 1, λ) + (1 − ω)E(z, α, β), (35) Fγeq 2RS (1 + z) − 1 fγSE (z)dz

Psop =
0
and = S +T; (40)
κΨ u ∞
 Z  Z ∞
mI
v − z (1+ C
x)
BR
E(z, u, v) = Fγ(1) 2RS (1 + z) − 1 fγSE (z)dz,

e γI
S = FPr (41)
Γ (mI ) Γ (u) 0 T1 eq

  j 0 
m I −1
mI C
X γI z 1 + x
xu−1 E(2RS (1 + z) − 1, 1, λ)
  
×   R∞
j! BR  
= FPr 1 − 0 +E(2RS (1 + z) − 1, α, β) ,
j=0
 
  T1 
κΨx −; − ×fγSE (z)dz

× G2,0 dx. (36)
 
0,2 0, mI − u; −
v
| {z }
J
  j and
Relying on the binomial expansion of m I C
γI z 1 + x , 
BR
Z ∞
Fγ(2) 2RS (1 + z) − 1 fγSE (z)dz,

with the help the identity (07.34.03.0228.01) of [26] as well T = FPr
T1 0
eq

as some algebraic manipulations, one obtains (42)


By involving (2) into (41), and performing the following
Ψκ u Z ∞ mXI −1 X j 
 
E(z, u, v) = v
m
− γI z
e I
j change of variables t = 2RS (1 + z) − 1, the first term of the
Γ (mI ) Γ (u) 0 k integral J above can be expressed as
j=0 k=0
 j−k
mI m E
γI z
  
mI C −; − mE ∞ mI −1 X
j
G1,0
Z
× z ωmI ΨCκ γE −
mI
t
X
j! 0,1
γI x k; − J1 = te γI

  λ2RS γ I Γ (mI ) Γ (mE ) 2RS −1 j=0 k=0


2,0 κΨx −; −
× G0,2 dx. (37)  j−k
v 0, mI − u; −   mI t  mE −1 m 
t+1

j γ I − γ E t+1
RS −1
× −1 e E 2
Armed by the identities (07.34.16.0002.01) and k j! 2RS
(07.34.21.0011.01) of [26], (19) is attained. In a similar
 
κmI ΨCt −; −
manner, the remaining term can be calculated. × G3,0
0,3 dt. (43)
λγ I 0, mI − 1; −
By using the identities (06.06.02.0001.01) and
B. Second case: ER ≥ BR
(07.34.02.0001.01) of [26] alongside with some algebraic
In this case, the harvested power is assumed to exceed the manipulations, we obtain the term ωS (1, λ) given in (22).
battery size BR . That is, the transmit power used by the relay In a similar manner, the remaining terms, namely S (α, β) ,
R in the second time slot is PR = PB = BTR1 . Consequently, T (1, λ) , T (α, β) can be computed with the same identities,
the SNR γRD is the random variable W = ηI RD
N0D scaled by which concludes the result provided in Proposition 2.
R EFERENCES [22] I. S. Gradshteyn and I. M. Ryzhik, Table of Integrals, Series, and
Products: Seventh Edition. Burlington, MA: Elsevier, 2007.
[1] Y. Zhang, R. Yu, M. Nekovee, Y. Liu, S. Xie, and S. Gjessing, “Cognitive [23] V. Bankey, P. K. Upadhyay, D. B. Da Costa, P. S. Bithas, A. G. Kanatas,
machine-to-machine communications: visions and potentials for the and U. S. Dias, “Performance analysis of multi-antenna multiuser hybrid
smart grid,” IEEE Network, vol. 26, no. 3, pp. 6–13, May 2012. satellite-terrestrial relay systems for mobile services delivery,” IEEE
[2] L. Jiang, H. Tian, Z. Xing, K. Wang, K. Zhang, S. Maharjan, S. Gjessing, Access, vol. 6, pp. 24 729–24 745, 2018.
and Y. Zhang, “Social-aware energy harvesting device-to-device com- [24] A. Mathai, R. K. Saxena, and H. J. Haubol, The H-Function Theory and
munications in 5g networks,” IEEE Wireless Communications, vol. 23, Applications. New York: Springer, 2010.
no. 4, pp. 20–27, August 2016. [25] F. Yilmaz and M. S. Alouini, “Product of shifted exponential variates
[3] P. Kamalinejad, C. Mahapatra, Z. Sheng, S. Mirabbasi, V. C. M. Leung, and outage capacity of multicarrier systems,” in 2009 European Wireless
and Y. L. Guan, “Wireless energy harvesting for the internet of things,” Conference, May 2009, pp. 282–286.
IEEE Communications Magazine, vol. 53, no. 6, pp. 102–108, June [26] I. W. Research, Mathematica Edition: version 11.3. Champaign,
2015. Illinois: Wolfram Research, Inc., 2018.
[4] G. Pan, H. Lei, Y. Yuan, and Z. Ding, “Performance analysis and
optimization for swipt wireless sensor networks,” IEEE Transactions
on Communications, vol. 65, no. 5, pp. 2291–2302, May 2017.
[5] K. Z. Panatik, K. Kamardin, S. A. Shariff, S. S. Yuhaniz, N. A. Ahmad,
O. M. Yusop, and S. Ismail, “Energy harvesting in wireless sensor
networks: A survey,” in 2016 IEEE 3rd International Symposium on
Telecommunication Technologies (ISTT), Nov 2016, pp. 53–58.
[6] S. Sudevalayam and P. Kulkarni, “Energy harvesting sensor nodes:
Survey and implications,” IEEE Communications Surveys Tutorials,
vol. 13, no. 3, pp. 443–461, Third 2011.
[7] J. Ye, H. Lei, Y. Liu, G. Pan, D. B. da Costa, Q. Ni, and Z. Ding,
“Cooperative communications with wireless energy harvesting over
nakagami- m fading channels,” IEEE Transactions on Communications,
vol. 65, no. 12, pp. 5149–5164, Dec 2017.
[8] Z. Zeng, S. Fu, H. Zhang, Y. Dong, and J. Cheng, “A Survey of
Underwater Optical Wireless Communications,” IEEE Commun. Surveys
Tuts., vol. 19, no. 1, pp. 204–238, First quarter 2017.
[9] E. Illi, F. El Bouanani, D. B. da Costa, F. Ayoub, and U. S. Dias,
“Dual-Hop Mixed RF-UOW Communication System: A PHY Security
Analysis,” IEEE Access, vol. 6, pp. 55 345–55 360, 2018.
[10] E. Zedini et al., “A New Simple Model for Underwater Wireless Optical
Channels in the Presence of Air Bubbles,” in GLOBECOM 2017 - 2017
IEEE Global Communications Conference, Dec 2017, pp. 1–6.
[11] M. V. Jamali et al., “Statistical distribution of intensity fluctuations for
underwater wireless optical channels in the presence of air bubbles,”
in 2016 Iran Workshop on Communication and Information Theory
(IWCIT), May 2016, pp. 1–6.
[12] Y. R. Ortega, P. K. Upadhyay, D. B. da Costa, P. S. Bithas, A. G. Kanatas,
U. S. Dias, and R. T. de Sousa Junior, “Joint effect of jamming and noise
in wiretap channels with multiple antennas,” in 2017 13th International
Wireless Communications and Mobile Computing Conference (IWCMC),
June 2017, pp. 1344–1349.
[13] M. Bloch, J. Barros, M. R. D. Rodrigues, and S. W. McLaughlin,
“Wireless Information-Theoretic Security,” IEEE Trans. Inf. Theory,
vol. 54, no. 6, pp. 2515–2534, June 2008.
[14] A. D. Wyner, “The wire-tap channel,” The Bell System Technical
Journal, vol. 54, no. 8, pp. 1355–1387, Oct 1975.
[15] M. O. Hasna and M. S. Alouini, “A performance study of dual-hop
transmissions with fixed gain relays,” IEEE Trans. Wireless Commun.,
vol. 3, no. 6, pp. 1963–1968, Nov 2004.
[16] I. S. Ansari, “On the Performance of Free-Space Optical Systems over
Generalized Atmospheric Turbulence Channels with Pointing Errors,”
Ph.D. dissertation, King Abdullah University of Science and Technol-
ogy (KAUST), Thuwal, Makkah Province, Kingdom of Saudi Arabia,
February 2015.
[17] E. Zedini, I. S. Ansari, and M. S. Alouini, “Performance Analysis of
Mixed Nakagami- m and Gamma-Gamma Dual-Hop FSO Transmission
Systems,” IEEE Photon. J., vol. 7, no. 1, pp. 1–20, Feb 2015.
[18] E. Illi, F. El Bouanani, D. B. da Costa, F. Ayoub, and U. S. Dias, “On
the Secrecy Performance of Mixed RF/UOW Communication System,”
in GLOBECOM’18 - IEEE Global Communications Conference, Dec
2018, pp. 1–6.
[19] H. Lei, Z. Dai, I. S. Ansari, K. H. Park, G. Pan, and M. S. Alouini,
“On Secrecy Performance of Mixed RF-FSO Systems,” IEEE Photon.
J., vol. 9, no. 4, pp. 1–14, Aug 2017.
[20] A. H. A. El-Malek, A. M. Salhab, S. A. Zummo, and M. S. Alouini,
“Enhancing Physical Layer Security of Multiuser SIMO Mixed RF/FSO
Relay Networks with Multi-Eavesdroppers,” in 2016 IEEE Globecom
Workshops (GC Wkshps), Dec 2016, pp. 1–7.
[21] M.-K. Simon and M.-S. Alouini, Digital Communication Over Fading
Channels. New York: John Wiley and Sons, 2005.

You might also like