IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 14, NO.

11, NOVEMBER 2015 6425

RF-Based Energy Harvesting in Decode-and-Forward

Relaying Systems: Ergodic and Outage Capacities
Yanju Gu, Student Member, IEEE, and Sonia Aïssa, Senior Member, IEEE

Abstract—Radio-frequency energy harvesting constitutes an ef- feasible for such huge number of small devices, like sensors,
fective way to prolong the lifetime of wireless networks, wean implantable medical devices, etc. Although far-field microwave
communication devices off the battery and power line, benefit power transfer is a strong candidate to replace cables in long-
the energy saving and lower the carbon footprint of wireless
communications. In this paper, an interference aided energy har- distance power transfer, additional power beacons need to be
vesting scheme is proposed for cooperative relaying systems, where settled and deployed, which is not a ready work for nowadays
energy-constrained relays harvest energy from the received infor- communication systems [5]–[7].
mation signal and co-channel interference signals, and then use Energy harvesting has become an appealing solution to such
that harvested energy to forward the correctly decoded signal to problems [8], [9]. Energy from solar, vibration, thermoelectric
the destination. The time-switching scheme (TS), in which the
receiver switches between decoding information and harvesting effects, and so forth [10], can be harvested and converted to
energy, as well as the power-splitting scheme (PS), where a portion electrical energy to support these energy-constrained communi-
of the received power is used for energy harvesting and the re- cation devices [11]. A promising harvesting technology is to use
maining power is utilized for information processing, are adopted the radio frequency (RF) energy, since ambient RF signals, e.g.,
separately. Applying the proposed energy harvesting approach from TV broadcast and cellular communications, are widely
to a decode-and-forward relaying system with the three-terminal
model, the analytical expressions of the ergodic capacity and the available in urban areas (day and night, indoors and outdoors)
outage capacity are derived, and the corresponding achievable [12]. In this technique, the ambient RF radiation is captured
throughputs are determined. Comparative results are provided by the receive antennas of wireless devices and converted into
and show that PS is superior to TS at high signal-to-noise ratio direct current voltage through appropriate circuits [13], [14]. A
(SNR) in terms of throughput, while at low SNR, TS outperforms safe way has been advanced to wirelessly power chips in human
PS. Furthermore, considering different interference power distri-
butions with equal aggregate interference power at the relay, the body by using such method [15].
corresponding system capacity relationship, i.e., the ordering of As the signal carries information as well as energy at the
capacities, is obtained. same time, simultaneous wireless information and power trans-
Index Terms—Co-channel interference, decode-and-forward, fer has been studied recently, where the receiver is assumed to
energy harvesting, ergodic capacity, outage capacity, relaying, be able to decode the information and harvest energy from the
throughput. same signal [16], [17]. However, due to practical circuit limita-
tions, it is difficult to harvest the energy and decode the infor-
mation at the same time. There are two schemes for harvesting
I. I NTRODUCTION
energy and decoding information separately [18]–[21], one is

I N modern society, wireless communication devices are om-

nipresent and have become numerous. They are intensively
involved in different applications such as video and audio
the time-switching scheme (TS) in which the receiver switches
over time between decoding information and harvesting energy;
and the other is the power-splitting scheme (PS) in which a por-
information transmission [1], monitoring in modern healthcare tion of the received power is used for energy harvesting and the
systems [2] and safety message exchange in vehicular net- remaining power is utilized for the information processing. From
works [3]. On one hand, the energy consumption is tremen- the perspective of receiver’s complexity, TS is superior to PS in
dous [4], which makes energy saving for communication a that commercially available circuits that are separately designed
critical problem to be solved. On the other hand, recharging for information decoding and energy harvesting can be used.
by traditional wiring method or battery replacement is not Simultaneous information decoding and energy harvesting
has applications and advantages in wireless systems in gen-
eral, whether in point-to-point communication or when nodes
Manuscript received September 23, 2014; revised March 19, 2015 and cooperate together in delivering the source signal to its final
May 24, 2015; accepted June 25, 2015. Date of publication July 8, 2015; date
of current version November 9, 2015. This work was supported by a Discovery destination, [21], [22]. Indeed, in cooperative networks, by
Grant from the Natural Sciences and Engineering Research Council (NSERC) deploying relays between the source and the destination, the
of Canada. The associate editor coordinating the review of this paper and cover range and capacity of the communication system can be
approving it for publication was E. A. Jorswieck.
The authors are with the Institut National de la Recherche Scientifique enhanced. However, the relays may have limited battery and
(INRS-EMT), University of Quebec, Montreal, QC H5A 1K6, Canada (e-mail: wired charging may be difficult to be implemented when and
[email protected]; [email protected]). where needed. To prolong the lifetime of relaying systems,
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. wireless energy harvesting at the relays becomes a necessity
Digital Object Identifier 10.1109/TWC.2015.2453418 [20], [23]–[25].
1536-1276 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

Authorized licensed use limited to: University of L'Aquila. Downloaded on May 13,2020 at 11:13:37 UTC from IEEE Xplore. Restrictions apply.
6426 IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 14, NO. 11, NOVEMBER 2015

Since the radio signal propagates freely over space, a receiver

would receive the desired signal with a superposition of un-
wanted signals, namely interferences, which in turn results in
low capacity between the transmitter and the receiver. Interfer-
ence is the primary bottleneck on the data rate capacity of most
wireless networks. How to decrease or avoid interference and
increase the signal-to-interference-plus-noise ratio (SINR) has
always been a big concern in research and industry. Techniques
such as frequency reuse [26], multi-cell coordination [27] and
interference alignment [28], [29] have been proposed for in-
terference cancellation. While interference decreases the com-
munication system capacity, from the energy point-of-view the
interference signal provides additional energy for the harvesting
system. Therefore, investigating the role that the interference
plays in energy-harvesting based communication system is of
major importance, though still missing.
In this paper, a decode-and-forward (DF) relaying system
where the relays need to replenish energy from the received RF
signals, is considered. For the limitation of hardware, harvest-
use strategy in which no device equipment is dedicated to store
the harvested energy, is adopted [11]. As opposed to traditional
relaying where co-channel interference (CCI) within the same Fig. 1. (a) Energy-harvesting based relaying system, where the energy-
bandwidth as the transmitted signal deteriorates the system constrained relay harvests energy from the signal and the co-channel inter-
performance [30], [31], and have to be eliminated by applying ferences. (b) Time-switching protocol for energy harvesting and information
processing at the relay. (c) Power-splitting protocol for energy harvesting and
interference alignment approach or by decoding the interfering information processing at the relay.
signals when they are strong, in this work, CCI signals are
utilized as a new source of power for relay recharging. Specifi-
cally, the relays harvest energy from both the information signal results which corroborate the analytical results are provided
and the CCI signals, and then use that energy to decode the in Section V. Finally, the paper’s conclusion is presented in
source signal and forward it to the destination node. In this way, Section VI.
the interference acts as useful power in the energy harvesting
phase and as noise in the information decoding phase. Initial
results for the ergodic capacity of a DF relaying system with II. E NERGY-H ARVESTING BASED R ELAYING
TS protocol appear in [23].
A. System and Channel Models
To provide a thorough study and guidelines for practical
applications, both TS and PS energy harvesting schemes are We consider a cooperative DF relaying system, where the
investigated. The analysis of the system performance is chal- source S communicates with the destination D through the
lenging due to the random feature of the transmission power at help of an energy-constrained intermediate relaying node R,
the relay in the proposed energy harvesting system. First, the as shown in Fig. 1(a). Each node is equipped with a single
ergodic capacity, which is a fundamental performance indica- antenna and operates in the half-duplex mode in which the
tor for delay-insensitive services, when the codeword length node cannot simultaneously transmit and receive signals in
can be sufficiently long to span over all the fading blocks, the same frequency band. Both, the first hop (source-to-relay)
is investigated. Moreover, for real-time applications, a more and the second hop (relay-to-destination), experience indepen-
appropriate performance metric, the outage capacity, defined as dent Rayleigh fading with the complex channel fading gains
the maximum constant rate that can be maintained over fading given by h ∼ CN(0, h ) and g ∼ CN(0, g ), respectively. The
blocks with a given outage probability, is studied. Analytical channels follow the block-fading model in which the channel
expressions for both the ergodic capacity and the outage capac- remains constant during the transmission of a block and varies
ity are derived and the corresponding achievable throughputs independently from one block to another. The channel state
are obtained. In addition, the impacts of the interference power information is only available at the receiver.
distribution on the ergodic capacity and the outage capacity Wireless communication networks are generally subjected to
as well as the corresponding achievable throughputs of the CCI due to the aggressive frequency reuse for a more efficient
proposed energy harvesting system are also studied based on resource utilization. In this vein, we assume that there are M
the majorization theory. CCI signals affecting the relay. The CCI signals are assumed
The rest of this paper is organized as follows. The sys- independent but not identically distributed. Specifically, the
tem model and energy harvesting schemes are described in channel fading gain between the ith interferer and the relay node
Section II. Considering time-switching and power-splitting R, denoted βi , is modeled as βi ∼ CN(0, βi ). Hereafter, the
protocols, the ergodic capacity and outage capacity are ana- desired channels and the interference channels are assumed to
lyzed in Section III and Section IV, respectively. Simulation be independent from each other.

Authorized licensed use limited to: University of L'Aquila. Downloaded on May 13,2020 at 11:13:37 UTC from IEEE Xplore. Restrictions apply.
GU AND AÏSSA: RF-BASED ENERGY HARVESTING IN DECODE-AND-FORWARD RELAYING SYSTEMS 6427

B. Wireless Energy Harvesting at the Relay the information to the destination. Therefore, from (3), the
transmission power of the relay is readily given by
In the network under study, the relay is considered to be
constrained in terms of energy. That is, it may have limited EH
PR =
battery reserves and needs to rely on some external charging (1 − α)T/2
mechanism in order to remain active in the network, and assist 2αησR2
the communication process between the source, S, and the = (γh + IR ). (4)
1−α
destination, D, as required. In the proposed approaches, the
received interference and information signals at the relay are Then, the received signal at the destination node D is given by
exploited to replenish energy for the relay. Both TS and PS 
yRD = PR gsR + nD , (5)
architectures for harvesting energy are studied.
1) Time-Switching Scheme: The time-switching based pro- where sR is the signal transmitted from the relay and nD is the
tocol is adopted at the relay node as illustrated in Fig. 1(b), AWGN noise at the destination, with zero mean and variance
where T is the block time in which a certain block of infor- σD2 . From (5), the received signal-to-noise ratio (SNR) at the
mation is transmitted from the source node to the destination destination node is obtained as
node and α, with 0 ≤ α ≤ 1, is the fraction of the block time in
PR |g|2
which the relay harvests energy from the received interference γRD =
signal and information signal. The remaining block time is di- σD2
vided into two equal parts, namely (1 − α)T/2, for information 2αη σR2 2
= |g| (γh + IR ), (6)
transmission from the source to the relay and from the relay 1 − α σD2
to the destination, respectively. Since there is no energy buffer  
to store the harvested energy (Harvest-Use) [11], [32], all the W
energy collected during the harvesting phase is consumed by where the defined random variable W follows the same distri-
the relay. bution as of |g|2.
In the first-hop phase, source S transmits signal s with power 2) Power-Splitting Scheme: In this case, the protocol
PS to the relay R. Accordingly, the received signal at the relay adopted at the relay node is as illustrated in Fig. 1(c), where
is given by P is the power of the received signal and θ , with 0 ≤ θ ≤ 1, is
 
M  the fraction of power that the relay harvests from the received
ySR = PS hs + Pi βi si + nR , (1) interference and information signal. The remaining power is
i=1 (1 − θ )P, which is used for information detection. In this
paper, we consider a pessimistic case in which power splitting
where si and Pi denote the signal and its corresponding power, only reduces the signal power, but not the noise power, which
from the ith interferer, and nR is the additive white Gaussian can provide a lower-bound performance measure for relaying
noise (AWGN) at the relay with zero mean and variance σR2 . networks in practice.
Accordingly, the received SINR at the relay is given by Accordingly, after power-splitting, the received signal at the
PS |h|2 relay for information detection is given by
γSR = 
σR2 + M i=1 Pi |βi |
2  
M 
ySR = (1 − θ )PS hs + (1 − θ )Pi βi si + nR . (7)
γh
= , (2) i=1
1 + IR
 Then, the received SINR at the relay is obtained as
where γh  PS2 |h|2 and IR  M Pi
i=1 2 |βi | . The received data
2
σR σR (1 − θ )PS |h|2
is correctly decoded if the instantaneous received SINR γSR at γSR = 
the relay is higher than the pre-defined threshold γth . σR2 + M i=1 (1 − θ )Pi |βi |
2
γh
When the relay is active, it harvests energy from the received = , (8)
information signal and the interference signal for a duration of 1 + IR
αT at each block, and thus, the harvested energy is obtained as 
where γh  (1−θ)P S
|h|2 and IR  M i=1
(1−θ)Pi
|βi |2 . Note that
  σR σR
2 2

M γh and IR here have distinct denotations from those in (2) for

EH = η PS |h| + 2
Pi |βi | αT,
2
(3) the TS protocol. We use the same symbols γh and IR in order to
i=1 unify the analysis in the following sections.
where η is the energy conversion efficiency coefficient, with value Different from TS, for PS, the relay harvests energy from the
varying from 0 to 1 depending upon the harvesting circuitry. received information and interference signal for a duration of
Since the processing power required by the transmit/receive T/2 at each block, and thus, the harvested energy at the relay is
circuitry at the relay is generally negligible compared to the obtained as
power used for signal transmission [18], [19], here we suppose  
 M
2 2 T
that all the energy harvested from the received signals (the EH = ηθ PS |h| + Pi |βi | . (9)
source’s and the CCI’s) is consumed by the relay for forwarding 2
i=1

Authorized licensed use limited to: University of L'Aquila. Downloaded on May 13,2020 at 11:13:37 UTC from IEEE Xplore. Restrictions apply.
6428 IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 14, NO. 11, NOVEMBER 2015

Suppose that all the harvested energy is consumed by the relay τi (A) is the multiplicity of μi , and χi,j (A) is the (i, j)th
for forwarding the information to the destination node in the characteristic coefficient of A [33]. Note that when the inter-
second-hop phase. From (9), the transmission power of the fering signals are statistically independent and identically dis-
relay node is readily given by tributed (i.i.d.), i.e., μi = μ for i = 1, 2, . . . , M, then υ(A) = 1,
EH τ1 (A) = M and IR is a sum of M i.i.d. exponential random
PR = variables with the central chi-squared
 distribution given by
T/2 μ−M γ
fIR (γ ) = (M−1)! γ M−1 exp − μ .
ηθ σR2 The CDF of γSR is then obtained as
= (γh + IR ). (10)
1−θ   
γ (1 + IR )
Then, the received SNR at the destination node is expressed as FγSR (γ ) = EIR 1 − exp − (14)
γ̄h
υ(A) τi (A)
γRD =
PR |g|2 γ 
= 1 − exp − χi,j (A)
σD2 γ̄h
i=1 j=1
ηθ σR2 2 −j
μi
= |g| (γh + IR ). (11) × 1+ γ , (15)
(1 − θ ) σD2 γ̄h
 
W where E[·] denotes the statistical expectation operator. In the
case when the interfering signals are i.i.d., the CDF of γSR
By introducing the random variables γh , IR and W, we unify −M 
the derivations of the distribution of the end-to-end SNR and the reduces to FγSR (γ ) = 1 − 1 + γ̄μh γ exp − γ̄γh .
capacity metrics for these two schemes (TS and PS), as detailed Similar to the derivation of (15), the PDF of γRD , which
in the following sections. involves products of two random variables, is determined as

 τ
υ(B) i (B)
j
III. E RGODIC C APACITY χi,j (B) γ 2
FγRD (γ ) = 1 − 2
In this section, the exact closed-form cumulative distribution (j − 1)! γ̄g μi
i=1 j=1

function (CDF) of the end-to-end SNR is derived. Then, the er- γ
godic capacity and the corresponding achievable throughput are × Kj 2 , (16)
γ̄g μi
investigated for the energy harvesting DF relaying system with
2αη σR θη σR
2 2
time-switching or power-splitting. The impact of the interfer- where γ̄g =
ence power distribution on the ergodic capacity and achievable 1−α σ 2 g for TS and γ̄g = 1−θ σ 2 g for PS,
D D
throughput is also analyzed, based on the majorization theory. B = diag(μ1 , μ2 , . . . , μM+1 ) with μM+1 = γ̄h , υ(B) denotes
the number of distinct diagonal elements of B, μ1 > μ2 >
A. End-to-End SNR . . . > μυ(B) are the distinct diagonal elements in decreasing
order, τi (B) is the multiplicity of μi , χi,j (B) is the (i, j)th
All channels, i.e., h, {βi }M
i=1 and g are supposed to be subject characteristic coefficient of B, and Kj (·) stands for the jth-order
to independent Rayleigh fading. Then, the received SNR at the modified Bessel function of the second kind [34].
first hop, γh , is of exponential distribution with the probability
density function (PDF) given by B. Ergodic Capacity and Achievable Throughput
1 x Ergodic capacity, in the unit of bit/s/Hz, quantifies the ul-
fγh (x) = exp − , x ≥ 0, (12)
γ̄h γ̄h timate reliable communication limit of the fading channel. It
PS (1−θ)PS is only achievable with infinite coding delay. Ergodic capacity
where γ̄h , equal to  for TS and to h for PS, is the
σR2 h σR2 can be obtained by averaging the instantaneous capacity over
average SNR from the source to the relay in a given time slot. all fading states. In the DF-cooperative communication system
The quantity IR is the sum of M statistically independent under study, the instantaneous capacity is determined by the
and not necessarily identically distributed (i.n.i.d.) exponential minimum one of each individual link, i.e., the first- and second-
random variables, each with mean μi = P2i βi for the TS- hop links. Therefore, the ergodic capacity is expressed as
σR
(1−θ)Pi   
based scheme and μi = βi for the PS-based one. Thus, 1 1
σR2 Cerg = E min log2 (1 + γSR ), log2 (1 + γRD ) (17)
the PDF of IR can be explicitly obtained as  2 2 
1
 τ
υ(A) i (A) −j
μi y =E log2 (1 + min{γSR , γRD }) (18)
fIR (y) = χi,j (A) yj−1 exp − , y ≥ 0, 2
 ∞
(j − 1)! μi 1
i=1 j=1 = log2 (1 + γ )fγmin (γ )dγ , (19)
(13) 2 0
where matrix A = diag(μ1 , μ2 , . . . , μM ), υ(A) denotes the where fγmin (γ ) stands for the PDF of the random variable
number of distinct diagonal elements of A, μ1 > μ2 > . . . > min{γSR , γRD }. The factor 1/2 in (17) is introduced by the
μυ(A) are the distinct diagonal elements in decreasing order, fact that two transmission phases are involved in the system.

Authorized licensed use limited to: University of L'Aquila. Downloaded on May 13,2020 at 11:13:37 UTC from IEEE Xplore. Restrictions apply.
GU AND AÏSSA: RF-BASED ENERGY HARVESTING IN DECODE-AND-FORWARD RELAYING SYSTEMS 6429

Expression (18) follows from the strictly monotonically in- Lemma 1: Suppose the function f : R+ → R+ is concave.
creasing property of the logarithm function for non-negative Then,
real numbers.   n 

Using the integration-by-parts method, (19) can be rewritten as G(μ) = Eω1 ,...,ωn f μk ωk (26)
1  ∞ k=1
Cerg = log2 (1 + γ ) Fγmin (γ ) − 1 0
2  ∞ is Schur-concave. Assume f is convex. Then the function G in
1 1  
(26) is Schur-convex [35].
− Fγmin (γ ) − 1 dγ (20)
2 ln 2 0 1 + γ Theorem 1: The ergodic capacity, Cerg , and the achievable
1 ∞ 1  
= 1 − Fγmin (γ ) dγ , (21) throughput, Terg , of the interference aided energy-harvesting
2 ln 2 0 1 + γ DF relaying system are Schur-convex with respect to μ, where
μ = (μ1 , μ2 , . . . , μM ) with μ1 ≥ μ 2 ≥ · · · ≥ μM ≥ 0.
where in (20) the operator {f (x)}ba  f (b) − f (a) and Fγmin (γ ) Proof: We can see that IR = M i=1 μi ωi , where ω1 , . . . ,
denotes the CDF of the random variable min{γSR , γRD } and is ωn are i.i.d. standard exponentially distributed with unit mean

given by and M i=1 μ i = E{I R }. According to (14),
  
Fγmin (γ ) = FγSR (γ ) + FγRD (γ ) − Pr{γSR ≤ γ , γRD ≤ γ } (22) γ (1 + IR )
FγSR (μ) = EIR 1 − exp −
= Pr{γSR ≤ γ } + Pr{γSR > γ , γRD ≤ γ }. (23) γ̄h
= EIR {f (IR )} . (27)
If γ is set to be a pre-defined threshold, (23) is the expression of  
Since the second derivative f (IR ) = − γ 2 exp − γ (1+I R)
2
the outage probability at the destination, the detailed derivation γ̄h ≤ 0,
γ̄h
of which is illustrated in the following section. then f (IR ) is a concave function. Thus, according to Lemma 1,
The achievable throughput at the destination relates only to FγSR (μ) is Schur-concave with respect to μ. Similarly, having
the effective information transmission time and is given by
  
1 γ
(1 − α)T FγRD (μ) = EW,IR 1 − exp − − IR
Terg = Cerg γ̄h W
T
= (1 − α)Cerg , (24) = EW,IR {g(IR )} , (28)
 γ 
for the system with TS protocol, and by and since g (IR ) = − 1
exp − γ̄1h W − IR ≤ 0, the function
γ̄h2
FγRD (μ) conditioned on W is Schur-concave. That is, for any
Terg = Cerg , (25)
two vectors μ1 μ2 , FγRD (μ1 |W) ≤ FγRD (μ2 |W). Averaging
over W, we have FγRD (μ1 ) ≤ FγRD (μ2 ). Both f (IR ) and g(IR )
for the PS based scheme.
are concave functions and using the same arguments,
Different from the conventional relaying system with no
  
rechargeable nodes, from (24) and (25) it is clear that in the γ (1 + IR )
interference aided energy harvesting system, the achievable Fγmin (μ) = EIR 1 − exp −
  γ̄h 
throughput depends not only on PS , σR2 , and σD2 , but also on γ (1 + IR )
α or θ , η, and Pi . + EW,IR exp −
 γ̄h 
1 γ
C. Impact of Interference Power Distribution − exp − − IR
γ̄h W
In order to provide an analysis of the impact of the inter- 1
= exp − EIR {f (IR )} + EW,IR {f (IR )}
ference power distribution on the energy harvesting system γ̄g
performance when the total received interference power is the + EW,IR {g(IR )} (29)
same, in this section, the Schur-convex property of ergodic
capacity and throughput is investigated. is also Schur-concave with respect to μ. Since 1 − Fγmin (μ1 ) ≥
For two vectors x and y(∈ Rn ) with descending ordered com- 1 − Fγmin (μ2 ) ≥ 0, integration with respect to γ gives
ponents x1 ≥ x2 ≥ · · · ≥ xn ≥ 0 and y1 ≥ y2 ≥ · · · ≥ yn ≥ 0, Cerg (μ1 ) ≥ Cerg (μ2 ). Therefore, the ergodic capacity, Cerg , and
respectively, one can say that the vector x majorizes the vector accordingly the achievable throughput, Terg , are Schur-convex
m 
m
with respect to μ. 
y and writes x y if xk ≥ yk for m = 1, . . . , n − 1, and
k=1 k=1 According to the Schur-convex property of ergodic ca-

n 
n
pacity, under different interference power distributions, the
xk = yk . A real-valued function defined on A ⊂ Rn is
k=1 k=1 corresponding system capacity relationship, i.e., the ordering of
said to be Schur-convex on A if x y on A ⇒ (x) ≥ (y). capacities, can be obtained. For example, our results imply that
Assume that ω1 , . . . , ωn are i.i.d. random variables according the worst scenario for the capacity performance occurs when
to a given PDF. Furthermore, assume vector μ to have non- the received interfering signals are of equal strength at the relay,
negative entries that are ordered in non-increasing order μ1 ≥ whereas the best case happens when there is only one interferer
μ2 ≥ · · · ≥ μn ≥ 0. affecting the relay.

Authorized licensed use limited to: University of L'Aquila. Downloaded on May 13,2020 at 11:13:37 UTC from IEEE Xplore. Restrictions apply.
6430 IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 14, NO. 11, NOVEMBER 2015

IV. O UTAGE C APACITY Theorem 2: Define Z  (γh + IR )1C , then the PDF of Z is
given by
In this section, the exact closed-form expressions of the out-
age probability, outage capacity and the achievable throughput  τ
υ(A) i (A) −j
1 z μi
are derived for the dual-hop energy harvesting DF relaying fZ (z) = 1Z exp − χi,j(A) 1 −
γ̄h γ̄h γ̄h
system. The impact of the interference power distribution on the ⎡ i=1 j=1
⎤
outage capacity and the achievable throughput is also analyzed,
z − γth 
j−1
aki z − γth k
based on the majorization theory. × ⎣1 − exp −ai ⎦ (32)
1 + γth k! 1 + γth
k=0

A. Outage Probability
where ai  μ1i − γ̄1h and 1Z is the indicator random variable
As an important performance measure of wireless systems, for the set Z = {z > γth }, i.e., 1Z = 1 if z > γth , otherwise,
outage probability is defined as the probability that the in- 1Z = 0.
stantaneous output SNR falls below a pre-defined threshold Proof: The CDF of the random variable Z = (γh + IR )1C
γth . This SNR threshold guarantees the minimum quality-of- is given by
service requirement of the destination users. Mathematically
speaking, Pout (γth) = Pr{γ ≤ γth }. In the DF relaying system FZ (z) = Pr{Z ≤ z}
 
under study, if the received SINR γSR at the relay is below γth ,
then the data received over that fading block cannot be decoded = fγh ,IR (x, y)dxdy, (33)
x,y∈S
correctly with probability approaching 1, and thus, the receiver # $
at the destination declares an outage since the data will not be where the set S = x + y ≤ z, 1+y x
> γth , x ≥ 0, y ≥ 0 . After
transmitted to the destination. Therefore, the outage probability
some set manipulations, we have S = ∅ if and only if z > γth.
at the destination is composed of two parts, that is,
Since γh and IR are independent, we get the joint distribution
fγh ,IR (x, y) = fγh (x)fIR (y). Then, after some straightforward al-
Pout(γth ) = Pr{γSR ≤ γth } + Pr{γSR > γth}
gebraic derivations, we obtain
× Pr{γRD ≤ γth |γSR > γth }
   z−γth  z−y
γh 1+γth
= Pr W(γh + IR ) ≤ γth , > γth FZ (z) = 1Z fγh (x)fIR (y)dxdy. (34)
1 + IR 0 (1+y)γth
 
γh
+ Pr ≤ γth (30) Now, substituting (12) and (13) into (34) and integrating with
1 + IR respect to x and y yields the CDF of Z. Then the PDF of Z
= Pr {W(γh + IR )1C ≤ γth } , (31) follows directly from differentiating FZ (z) with respect to z.
Here [34, Eq.(3.351.1)] was used to reach (32). 
where 1C is the indicator random variable for the set C = Next, we evaluate the outage probability at the destination by
{γSR > γth }, i.e., 1C = 1 if γSR > γth , otherwise, 1C = 0. using the above Theorem 2.
Note that, in contrast to traditional DF relaying system with Theorem 3: The outage probability at the destination node
no rechargeable nodes, the transmission power PR at the relay of the interference aided energy harvesting DF relaying system
in the energy harvesting system is a random variable, which is given by (35), shown at the bottom of the page, where bi =
ai
γ̄h + 1+γth and (a, x; b) is the generalized
depends on the replenished energy from the interference and 1
incomplete Gamma
information signal. Therefore, the distribution of the received %∞
function defined by (a, x; b)  x ta−1 exp(−t − bt−1 )dt.
SNR at the destination is determined not only by the distribu-
Proof: We have
tion of the relay-to-destination channel power gain |g|2 , but also
by the distribution of the information and interference signal Pout (γth ) = Pr{WZ ≤ γth }
power, i.e., γh and IR . On the other hand, in the counterpart sys-  
γth
tem of conventional DF relaying, when the relay can decode the = EZ 1 − exp −
γ̄g Z
information correctly, its transmission power PR is a constant,  ∞
γth
and thus, the received SNR at the destination only depends on =1− exp − fZ (z)dz. (36)
the relay-to-destination channel power gain |g|2 . 0 γ̄ gz

⎧
 τ
υ(A) i (A) −j ⎨
μi γth γth 1 ai γth
Pout (γth ) = 1 − χi,j (A) 1 −  1, ; − exp
γ̄h ⎩ γ̄h γ̄h γ̄g γ̄h 1 + γth
i=1 j=1
⎫

j−1
1 −ai γth k 
k
k b−1 bi γth ⎬
i
×    m + 1, bi γth ; (35)
k! 1 + γth m −bi γth m γ̄g ⎭
k=0 m=0

Authorized licensed use limited to: University of L'Aquila. Downloaded on May 13,2020 at 11:13:37 UTC from IEEE Xplore. Restrictions apply.
GU AND AÏSSA: RF-BASED ENERGY HARVESTING IN DECODE-AND-FORWARD RELAYING SYSTEMS 6431

According to Theorem 2, by substituting (32) into (36), we relates only to the effective information transmission time and
obtain (37), shown at the bottom of the page. Next, we focus is then given by
on the two integrations in (37). For the first integration, from
the definition of the generalized incomplete Gamma function, Tout = (1 − α)Cout , (42)
we have
 ∞ for the system employing time switching, and by
1 γth z γth γth
exp − − dz =  1, ; . (38)
γth γ̄h γ̄g z γ̄h γ¯h γ̄h γ̄g Tout = Cout , (43)

For the second integration, exploiting the Taylor series ex- for the system implementing power splitting.
pansion of (z − γk )k with respect to z and the identity of the
generalized incomplete Gamma function lead to (35). 
C. Impact of Interference Power Distribution
Simplified expressions for the outage probability as 1) the
interferences are i.i.d. or 2) the number of interferers equals Theorem 4: The outage capacity, Cout , and the achievable
one, are derived and given by throughput, Tout , of the interference aided energy-harvesting
 DF relaying system is Schur-convex with respect to μ, where
(1) μ −M γth γth 1
Pout(γth ) = 1 − 1 −  1, ; − μ = (μ1 , μ2 , . . . , μM ) with μ1 ≥ μ2 ≥ · · · ≥ μM ≥ 0.
γ̄h γ̄h γ̄h γ̄g γ̄h Proof: According to (30) and using the same arguments
aγth 
M−1
1 −aγth k 
k as in the proof of Theorem 1, we can see that the outage
× exp capacity Cout and accordingly the achievable throughput Tout
1 + γth k! 1 + γth
k=0 m=0 are also Schur-convex with respect to μ. 

k b−1 bγth Note that Theorem 1 and Theorem 4 provide engineering
 m + 1, bγth ; (39)
m (−bγth)m γ̄g insights for design of energy harvesting relay system. With
regard to application, the proposed energy harvesting relay
and system can be seen as building block of a larger cellular
network. For instance, consider full frequency reuse for all
(2) γ̄h γth γth b−1 base stations, which is studied extensively recently for multi-
Pout(γth ) = 1 −  1, ; +
γ̄h − μ γ̄h γ̄h γ̄g γ̄h − μ cell cooperation, and where a base station serves farther users
aγth bγth through the help of intermediate relaying nodes. From a system
× exp  1, bγth; (40) design point-of-view, how to choose the relay location to obtain
1 + γth γ̄g
the largest capacity is a meaningful and challenging problem.
respectively, where a  1
μ − 1
γ̄h and b  1
γ̄h + a
1+γth .
Relays that are positioned at different geometric locations may
suffer the same total received interference power (at the same
contour) from neighboring base stations, but with different
B. Outage Capacity and Achievable Throughput power distributions. Based on the analysis provided in this
Outage capacity, in the unit of bit/s/Hz, is defined as the paper, the best relay positioning can be identified. Definitely,
maximum constant rate that can be maintained over fading the detailed application depends on the specific problem, which
blocks with a specified outage probability. It is used for slowly is beyond the scope of the paper, and can be considered in future
varying channels, where the instantaneous SNR γ is assumed extensions of this work.
to be constant for a large number of symbols. In the DF-
cooperative communication system under study, the outage
V. N UMERICAL R ESULTS AND D ISCUSSIONS
capacity in the unit of bit/s/Hz is expressed as
In this section, numerical examples are presented and cor-
1 
Cout = 1 − Pout (γth ) log2 (1 + γth ). (41) roborated by simulation results to examine the throughput, Terg
2 and Tout , of the DF cooperative communication system, where
The factor 1/2 accounts for the fact that two transmission phases the energy-constrained relay harvests energy from the received
are involved in the communication between the source S and information signal and the CCI signals. Hereafter, and unless
the destination D. The achievable throughput at the destination stated otherwise, the number of CCI signals at the relay, M, is

⎡
 τ
υ(A) i (A) −j  ∞
μi ⎣ 1 γth z
Pout(γth ) = 1 − χi,j (A) 1 − exp − − dz
γ̄h γth γ̄h γ̄g z γ̄h
i=1 j=1
⎤
j−1 k 
 ∞ k
ai 1 γth z z − γth z − γth
− exp − − exp −ai dz⎦ (37)
k! γth γ̄h γ̄g z γ̄h 1 + γth 1 + γth
k=0

Authorized licensed use limited to: University of L'Aquila. Downloaded on May 13,2020 at 11:13:37 UTC from IEEE Xplore. Restrictions apply.
6432 IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 14, NO. 11, NOVEMBER 2015

Fig. 2. Throughput Terg and Tout versus the energy harvesting ratio α for Fig. 3. Throughput Terg and Tout versus the energy harvesting ratio θ for
different values of average SIR received at the relay, where the first-hop average different values of average SIR received at the relay, where the first-hop average
SNR is 20 dB. SNR is 20 dB.

set to 2 with normalized μ̂ = E{IμR } = (0.6, 0.4). The threshold

γth is set to 8 dB and the energy conversion efficiency η is set
to 1. To better evaluate the effects of the interferences on the
system’s throughput, we define MPS h as the average signal-
i=1 Pi βi
PS h
to-interference ratio (SIR) at the relaying node and as the
σR2
first-hop average SNR.
For the system with TS protocol, Fig. 2 shows the throughput
Terg and Tout versus the energy harvesting ratio α for different
values of average SIR received at the relay, where the first-
hop average SNR is 20 dB. It is observed that the analytical
results of (24) and (42) match well the simulation results. As the
energy harvesting ratio α increases from 0 to 1, the throughput
of the system increases at first until α reaches the optimal
value where the throughput gets its maximum, and thereafter Fig. 4. Optimal throughput Terg versus the first-hop average SNR at different
values of average SIR, for both power-splitting (PS) and time-switching (TS).
decreases from the maximum to zero. The concave feature of
the curves is due to the fact that the energy harvested for the
second-hop transmission increases with increasing α, which system. Meanwhile, as θ increases, more power is harvested for
effectively decreases the outage and enhances the capacity of information transmission and less power is left for information
the second hop and, accordingly, improves the throughput of decoding which deteriorates the throughput of the system and,
the system.Meanwhile, as α increases, more data are wasted thus, the throughput reaches a maximum and then drops down.
on energy harvesting and less information is decoded for infor- In both plots (Fig. 2 and Fig. 3), it is seen that the throughput
mation transmission which heavily reduces the throughput of Tout is less than the throughput Terg due to the requirement of
the system, therefore, the throughput reaches a maximum and the outage capacity that a fixed date rate is maintained in all
then drops down. As SIR increases, the optimal throughput and non-outage channel states.
the optimal α both increase. This means that when the received Next, we compare the throughput performances of the energy
average SNR at the relay is fixed, an increase in the power harvesting systems with TS and PS protocols, respectively, to
of the CCI signals can deteriorate the system performance, facilitate the choice of these two schemes for designing energy
but effectively reduces the optimal α required to achieve the harvesting system. Figs. 4 and 5 illustrate the optimal Terg and
optimal throughput. the optimal Tout versus the first-hop average SNR, respectively,
For comparison purposes, Fig. 3 depicts the throughput Terg for these two protocols given different values for the average
and Tout versus the energy harvesting ratio θ for the system SIR at the relay. It is observed that the PS protocol is superior to
with PS protocol under the same simulation settings. It is the TS protocol at high SNR, in terms of optimal Terg and Tout.
observed that the analytical results of (25) and (43) match At relatively low SNR, on the other hand, the TS-based scheme
perfectly the simulation results. The concave feature of the outperforms the PS one in terms of optimal Tout , but with little
curves is due to the fact that the energy harvested for the difference in optimal Terg . This can be explained as follows.
second-hop transmission increases with increasing θ , which At high SNR, power-splitting with optimal ratio θ (θ is around
effectively decreases the outage and enhances the capacity of half) would not decrease the received SNR significantly so that
the second hop and, accordingly, improves the throughput of the the information could still be correctly decoded at the relaying

Authorized licensed use limited to: University of L'Aquila. Downloaded on May 13,2020 at 11:13:37 UTC from IEEE Xplore. Restrictions apply.
GU AND AÏSSA: RF-BASED ENERGY HARVESTING IN DECODE-AND-FORWARD RELAYING SYSTEMS 6433

Fig. 5. Optimal throughput Tout versus the first-hop average SNR at different Fig. 7. Throughput of the PS based system, Terg and Tout , versus the first-hop
values of average SIR, for both PS and TS. average SNR under different CCI power distributions.

VI. C ONCLUSION
In this paper, an RF-based energy harvesting relaying system
was proposed, where the energy-constrained relay harvests
energy from the superposition of received information signal
and co-channel interference (CCI) signals, and then uses that
harvested energy to forward the correctly decoded signal to the
destination. The time-switching (TS) and the power-splitting
(PS) protocols were adopted, and their ensuing performance
was compared. Different from traditional decode-and-forward
relaying system with no rechargeable nodes, the transmission
power of the energy constrained relay is not a constant anymore
but a random variable depending on the variation of available
energy harvested from the received information and CCI signals
at the relay. Analytical expressions for the ergodic capacity as
well as for the outage capacity were derived to determine the
Fig. 6. Throughput of the TS based system, Terg and Tout , versus the first-hop
average SNR under different CCI power distributions. corresponding system achievable throughputs. The PS scheme
was demonstrated to be superior to TS at high SNR in terms
of the achievable throughput from ergodic or outage capacity,
node, but for the time-switching scheme, there always exists an while at relatively low SNR, TS outperforms PS in terms of the
information loss at the energy harvesting phase. Similarly, at achievable throughput from outage capacity. Furthermore, con-
low SNR, power splitting with optimal ratio θ would lead to sidering different interference power distributions with equal
more decoding errors at the relaying node. aggregate interference power at the relay, the corresponding
The impact of the interference power distribution on the system capacity relationship, i.e., the ordering of capacities,
throughput performance is shown in Fig. 6 for the system was obtained. The results reveal that the worst scenario for
with TS protocol and in Fig. 7 for the system with PS pro- the capacity performance occurs when the received interfering
tocol. The energy harvesting ratios α and θ are set to 0.2 signals are of equal strength at the relay, whereas the best case
and 0.6, respectively. The SIR at the relay is 10 dB. The occurs when there is only one interferer affecting the relay.
total interference power is the same at each SNR but with
different normalized power distribution: μ̂1 = (1, 0, 0, 0, 0),
R EFERENCES
μ̂2 = (0.6, 0.4, 0, 0, 0) and μ̂3 = (0.2, 0.2, 0.2, 0.2, 0.2). Ac-
cording to the definition of majorization, we have μ1 μ2 [1] S. Misra, M. Reisslein, and G. Xue, “A survey of multimedia streaming
in wireless sensor networks,” IEEE Commun. Surveys Tuts., vol. 10,
μ3 and thus, Terg (μ1 ) ≥ Terg (μ2 ) ≥ Terg (μ3 ) and Tout (μ1 ) ≥ no. 4, pp. 18–39, 1st Quart. 2008.
Tout (μ2 ) ≥ Tout (μ3 ), since the throughput is Schur-convex with [2] F. Hu, Y. Xiao, and Q. Hao, “Congestion-aware, loss-resilient bio-
respect to μ as proven by Theorems 1 and 4. This is clearly monitoring sensor networking for mobile health applications,” IEEE J.
Sel. Areas Commun., vol. 27, no. 4, pp. 450–465, May 2009.
shown by the simulation results, and it implies that the worst [3] J. B. Kenney, “Dedicated Short-Range Communications (DSRC) stan-
scenario for the throughput performance occurs when the inter- dards in the united states,” Proc. IEEE, vol. 99, no. 7, pp. 1612–1182,
fering signals are of equal received power at the relay, whereas Jul. 2011.
[4] Z. Hasan, H. Boostanimehr, and V. Bhargava, “Green cellular networks:
the best case happens when there is only one interferer affecting A survey, some research issues and challenges,” IEEE Commun. Surveys
the relay. Tuts., vol. 13, no. 4, pp. 524–540, 4th Quart. 2011.

Authorized licensed use limited to: University of L'Aquila. Downloaded on May 13,2020 at 11:13:37 UTC from IEEE Xplore. Restrictions apply.
6434 IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 14, NO. 11, NOVEMBER 2015

[5] K. Huang and V. Lau, “Enabling wireless power transfer in cellular net- [31] Y. Gu, S. Ikki, and S. Aissa, “Opportunistic cooperative communication
works: Architecture, modeling and deployment,” IEEE Trans. Wireless in the presence of co-channel interferences and outdated channel informa-
Commun., vol. 13, no. 2, pp. 902–912, Feb. 2014. tion,” IEEE Commun. Lett., vol. 17, no. 10, pp. 1948–1951, Oct. 2013.
[6] M. Xia and S. Aissa, “Effect of opportunistic scheduling on the effi- [32] O. Ozel and S. Ulukus, “AWGN channel under time-varying amplitude
ciency of wireless power transfer,” in Proc. IEEE Global Commun. Conf., constraints with causal information at the transmitter,” in Proc. 45th
Dec. 2014, pp. 4288–4293. Asilomar Conf. Signals, Syst. Comput., Nov. 2011, pp. 373–377.
[7] M. Xia and S. Aissa, “On the efficiency of far-field wireless power [33] H. Shin and M. Win, “MIMO diversity in the presence of double scatter-
transfer,” IEEE Trans. Signal Process., vol. 63, no. 11, pp. 2835–2847, ing,” IEEE Trans. Inf. Theory, vol. 54, no. 7, pp. 2976–2996, Jul. 2008.
Jun. 2015. [34] I. S. Gradshteyn and I. M. Ryzhik, Table of Integrals, Series and Products,
[8] L. Mateu and F. Moll, “Review of energy harvesting techniques and 7th ed. New York, NY, USA: Academic Press, 2007.
applications for microelectronics,” in Proc. SPIE Circuits Syst. II, 2005, [35] E. Jorswieck and H. Boche, “Majorization and matrix-monotone functions
pp. 359–373. in wireless communications,” Found. Trends Commun. Inf. Theory, vol. 3,
[9] S. Sudevalayam and P. Kulkarni, “Energy harvesting sensor nodes: no. 6, pp. 553–701, 2007.
Survey and implications,” IEEE Commun. Surveys Tuts., vol. 13, no. 3,
pp. 443–461, 3rd Quart. 2011.
[10] V. Raghunathan, S. Ganeriwal, and M. Srivastava, “Emerging techniques Yanju Gu (S’13) received the B.Sc. (honors) and
for long lived wireless sensor networks,” IEEE Commun. Mag., vol. 44, M.E. (honors) degrees in electrical engineering from
no. 4, pp. 108–114, Apr. 2006. the University of Electronic Science and Technology
[11] R. Ramachandran, V. Sharma, and P. Viswanath, “Capacity of Gaussian of China, Chengdu, China. She is currently working
channels with energy harvesting and processing cost,” IEEE Trans. Inf. towards the Ph.D. degree at the Institut National
Theory, vol. 60, no. 5, pp. 2563–2575, May 2014. de la Recherche Scientifique-Energy, Materials and
[12] V. Liu et al., “Ambient backscatter: Wireless communication out of thin Telecommunications Center (INRS-EMT), Univer-
air,” in Proc. ACM SIGCOMM, Aug. 2013, pp. 39–50. sity of Quebec, Montreal, QC, Canada.
[13] T. Paing, J. Shin, R. Zane, and Z. Popovic, “Resistor emulation approach Her research interests are in the area of green
to low-power RF energy harvesting,” IEEE Trans. Power Electron., communication, energy harvesting system design
vol. 23, no. 3, pp. 1494–1501, May 2008. and analysis for wireless communications. She is a
[14] R. Rajesh, V. Sharma, and P. Viswanath, “Information capacity of energy recipient of Doctoral Research Scholarship from the Fonds de Recherche du
harvesting sensor nodes,” in Proc. IEEE Int. Symp. Inf. Theory, Jul. 2011, Québec Nature et Technologies (FRQNT)-Scholarship Program for Foreign
pp. 2363–2367. Students, 2015.
[15] J. S. Hoa et al., “Wireless power transfer to deep-tissue microimplants,”
Proc. Nat. Acad. Sci., vol. 111, no. 22, pp. 7974–7979, Jun. 2014.
[16] L. Varshney, “Transporting information and energy simultaneously,” in
Proc. IEEE Int. Symp. Inf. Theory, Jul. 2008, pp. 1612–1616. Sonia Aïssa (S’93–M’00–SM’03) received the
[17] P. Grover and A. Sahai, “Shannon meets Tesla: Wireless information Ph.D. degree in electrical and computer engineering
and power transfer,” in Proc. IEEE Int. Symp. Inf. Theory, Jun. 2010, from McGill University, Montreal, QC, Canada, in
pp. 2363–2367. 1998. Since then, she has been with the Institut
[18] B. Medepally and N. Mehta, “Voluntary energy harvesting relays National de la Recherche Scientifique-Energy, Mate-
and selection in cooperative wireless networks,” IEEE Trans. Wireless rials and Telecommunications Center (INRS-EMT),
Commun., vol. 9, no. 11, pp. 3543–3553, Nov. 2010. University of Quebec, Montreal, QC, Canada, where
[19] X. Zhou, R. Zhang, and C. K. Ho, “Wireless information and power she is a Full Professor.
transfer: Architecture design and rate-energy tradeoff,” in Proc. IEEE From 1996 to 1997, she was a Researcher with
Global Commun. Conf., Dec. 2012, pp. 3982–3987. the Department of Electronics and Communications
[20] A. Nasir, X. Zhou, S. Durrani, and R. Kennedy, “Relaying protocols of Kyoto University, and with the Wireless Systems
for wireless energy harvesting and information processing,” IEEE Trans. Laboratories of NTT, Japan. From 1998 to 2000, she was a Research Associate
Wireless Commun., vol. 12, no. 7, pp. 3622–3636, Jul. 2013. at INRS-EMT. In 2000–2002, while she was an Assistant Professor, she
[21] L. Liu, R. Zhang, and K.-C. Chua, “Wireless information transfer with was a Principal Investigator in the major program of personal and mobile
opportunistic energy harvesting,” IEEE Trans. Wireless Commun., vol. 12, communications of the Canadian Institute for Telecommunications Research,
no. 1, pp. 288–300, Jan. 2013. leading research in radio resource management for wireless networks. From
[22] Z. Ding, S. Perlaza, I. Esnaola, and H. Poor, “Power allocation strate- 2004 to 2007, she was an Adjunct Professor with Concordia University,
gies in energy harvesting wireless cooperative networks,” IEEE Trans. Montreal. In 2006, she was Visiting Invited Professor with the Graduate School
Wireless Commun., vol. 13, no. 2, pp. 846–860, Feb. 2014. of Informatics, Kyoto University, Japan. Her research interests include the
[23] Y. Gu and S. Aissa, “Interference aided energy harvesting in decode-and- modeling, design and performance analysis of wireless communication systems
forward relaying systems,” in Proc. IEEE Int. Conf. Commun., Jun. 2014, and networks.
pp. 5378–5382. Dr. Aïssa is the Founding Chair of the IEEE Women in Engineering Affinity
[24] Z. Ding, I. Krikidis, B. Sharif, and H. Poor, “Wireless information Group in Montreal, 2004–2007; acted as TPC Leading Chair or Cochair of
and power transfer in cooperative networks with spatially random re- the Wireless Communications Symposium at IEEE ICC ’06 ’09 ’11 ’12;
lays,” IEEE Trans. Wireless Commun., vol. 13, no. 8, pp. 4440–4453, PHY/MAC Program Cochair of the 2007 IEEE WCNC; TPC Cochair of the
Aug. 2014. 2013 IEEE VTC-spring; and TPC Symposia Chair of the 2014 IEEE Globe-
[25] H. Chen, Y. Li, Y. Jiang, Y. Ma, and B. Vucetic, “Distributed power com. Her main editorial activities include: Editor, IEEE T RANSACTIONS ON
splitting for swipt in relay interference channels using game theory,” IEEE W IRELESS C OMMUNICATIONS, 2004–2012; Associate Editor, IEEE Com-
Trans. Wireless Commun., vol. 14, no. 1, pp. 410–420, Jan. 2015. munications Magazine, 2004–2009; Technical Editor, IEEE Wireless Com-
[26] H. Zhu, “On frequency reuse in cooperative distributed antenna systems,” munications Magazine, 2006–2010; and Associate Editor, Wiley Security and
IEEE Commun. Mag., vol. 50, no. 4, pp. 85–89, Apr. 2012. Communication Networks Journal, 2007–2012. She currently serves as Area
[27] D. Gesbert et al., “Multi-Cell MIMO cooperative networks: A new look at Editor for the IEEE T RANSACTIONS ON W IRELESS C OMMUNICATIONS ,
interference,” IEEE J. Sel. Areas Commun., vol. 28, no. 9, pp. 1380–1408, and Technical Editor for the IEEE Communications Magazine. Awards to her
Dec. 2010. credit include the NSERC University Faculty Award in 1999; the Quebec
[28] V. Cadambe and S. Jafar, “Interference alignment and degrees of freedom Government FQRNT Strategic Faculty Fellowship in 2001–2006; the INRS-
of the K-user interference channel,” IEEE Trans. Inf. Theory, vol. 54, EMT Performance Award multiple times since 2004, for outstanding achieve-
no. 8, pp. 3425–3441, Aug. 2008. ments in research, teaching and service; and the Technical Community Service
[29] Y. Yang, S. Aissa, A. Eltawil, and K. Salama, “An interference cancella- Award from the FRQNT Centre for Advanced Systems and Technologies in
tion strategy for broadcast in hierarchical cell structure,” in Proc. IEEE Communications, 2007. She is co-recipient of five IEEE Best Paper Awards
Global Commun. Conf., Dec. 2014, pp. 1792–1797. and of the 2012 IEICE Best Paper Award; and recipient of NSERC Discovery
[30] M. Xia and S. Aissa, “Cooperative AF relaying in spectrum-sharing Accelerator Supplement Award. She is a Distinguished Lecturer of the IEEE
systems: Performance analysis under average interference power con- Communications Society (ComSoc) and an Elected Member of the ComSoc
straints and Nakagami-m fading,” IEEE Trans. Commun., vol. 60, no. 6, Board of Governors. Professor Aïssa is a Fellow of the Canadian Academy of
pp. 1523–1533, Jun. 2012. Engineering.

Authorized licensed use limited to: University of L'Aquila. Downloaded on May 13,2020 at 11:13:37 UTC from IEEE Xplore. Restrictions apply.