560 IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 66, NO.

2, FEBRUARY 2018

Exploiting Full/Half-Duplex User Relaying

in NOMA Systems
Xinwei Yue, Student Member, IEEE, Yuanwei Liu, Member, IEEE, Shaoli Kang,
Arumugam Nallanathan, Fellow, IEEE, and Zhiguo Ding, Senior Member, IEEE
Abstract— In this paper, a novel cooperative non-orthogonal spectrum is of great significance in new radio (NR) usage
multiple access (NOMA) system is proposed, where one near scenarios [2]. To achieve higher spectral efficiency of the
user is employed as decode-and-forward relaying switching fifth generation (5G) mobile communication network, non-
between full-duplex (FD) and half-duplex (HD) mode to help
a far user. Two representative cooperative relaying scenarios are orthogonal multiple access (NOMA) has received a great deal
investigated insightfully. The first scenario is that no direct link of attention [3]. Recently, several NOMA schemes have
exists between the base station (BS) and far user. The second been researched in detail, such as power domain NOMA
scenario is that the direct link exists between the BS and far user. (PD-NOMA) [4], sparse code multiple access (SCMA) [5],
To characterize the performance of potential gains brought by pattern division multiple access (PDMA) [6], multiuser sharing
the FD NOMA in two considered scenarios, three performance
metrics outage probability, ergodic rate, and energy efficiency access (MUSA) [7], etc. Generally speaking, NOMA schemes
are discussed. More particularly, we derive new closed-form can be classified into two categories, namely power-domain
expressions for both exact and asymptotic outage probabilities NOMA1 and code-domain NOMA. Downlink multiuser super-
as well as delay-limited throughput for two NOMA users. Based position transmission (DL MUST), the special case of NOMA,
on the derived results, the diversity orders achieved by users has been studied for 3rd generation partnership project (3GPP)
are obtained. We confirm that the use of direct link overcomes
zero diversity order of far NOMA user inherent to FD relaying. in [8]. The pivotal characteristic of NOMA is to allow
In addition, we derive new closed-form expressions for asymptotic multiple users to share the same physical resource (i.e.,
ergodic rates. Based on these, the high signal-to-noise ratio (SNR) time/fequency/code) via different power levels. At the receiver
slopes of two users for FD NOMA are obtained. Simulation side, the successive interference cancellation (SIC) is carried
results demonstrate that: 1) the FD NOMA is superior to the out [9].
HD NOMA in terms of outage probability and ergodic sum rate
in the low SNR region; and 2) in delay-limited transmission mode, So far, point-to-point NOMA has been studied
the FD NOMA has higher energy efficiency than the HD NOMA extensively in [10]–[14]. To evaluate the performance
in the low SNR region; However, in delay-tolerant transmission of uplink NOMA systems, the authors in [10] proposed the
mode, the system energy efficiency of the HD NOMA exceeds the uplink NOMA transmission scheme to achieve higher system
FD NOMA in the high SNR region. rate. The expressions of outage probability and achievable
Index Terms— Decode-and-forward, full-duplex, half-duplex, sum rates for uplink NOMA were derived with a novel
non-orthogonal multiple access, user relaying. uplink control protocol in [11]. Regarding downlink NOMA
I. I NTRODUCTION scenarios, authors in [12] analyzed the outage behavior and
ergodic rates of NOMA networks, where multiple NOMA
W ITH the rapid increasing demand of wireless net-
works, the requirements for efficiently exploiting the users are spatial randomly deployed in a disc. In [13],
the cognitive radio inspired NOMA concept was proposed,
Manuscript received June 16, 2017; accepted August 28, 2017. Date of in which the influence of user pairing with the fixed power
publication September 6, 2017; date of current version February 14, 2018. This allocation in NOMA systems was discussed. As the interplay
work was supported by National High Technology Research and Development
Program of China (863 Program, 2015AA01A709). The work of Z. Ding between NOMA and cognitive radio is bidirectional, NOMA
was supported by the UK EPSRC under grant number EP/L025272/1 and was also applied to cognitive radio networks in [14]. More
by H2020-MSCA-RISE-2015 under grant number 690750. This work was particularly, the analytical expressions of outage probability
presented at the IEEE ICC 2017 [1]. The associate editor coordinating
the review of this paper and approving it for publication was X. Yuan. was derived and diversity orders were characterized. Apart
(Corresponding author: Yuanwei Liu.) from the above works, a new opportunistic NOMA scheme
X. Yue is with the School of Electronic and Information Engineer- was proposed in [15] to improve the efficiency of SIC.
ing, Beihang University, Beijing 100191, China (e-mail: xinwei_yue@
buaa.edu.cn). In [16], the flexible power allocation mode was researched
Y. Liu and A. Nallanathan are with the School of Electronic Engineering in terms of outage probability for hybrid NOMA systems.
and Computer Science, Queen Mary University of London, London E14 NS, The quantum-assisted multiple users transmission mode for
U.K. (e-mail: yuanwei.liu@qmul.ac.uk; a.nallanathan@qmul.ac.uk).
S. Kang is with the School of Electronic and Information Engineering, NOMA was proposed in [17], which utilizes the minimum
Beihang University, Beijing 100191, China, and also with the State Key bit error ratio criterion to optimize the predefined transmitted
Laboratory of Wireless Mobile Communications, China Academy of Telecom- information. The author of [18] has studied the linear MUST
munications Technology, Beijing 100094, China (e-mail: kangshaoli@catt.cn).
Z. Ding is with the Department of Electrical Engineering, Princeton Univer- scheme for NOMA to maximize sum rate of the entire
sity, Princeton, NJ 08540 USA, and also with the School of Computing and network. With the emphasis on physical layer security,
Communications, Lancaster University, Lancaster LA1 4YW, U.K. (e-mail: in [19], authors have adopted two effective approaches,
z.ding@lancaster.ac.uk).
Color versions of one or more of the figures in this paper are available
1 In this paper, we focus on power-domain NOMA and use NOMA to
online at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TCOMM.2017.2749400 represent PD-NOMA.
0090-6778 © 2017 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: ULAKBIM UASL - Yakin Dogu Universitesi. Downloaded on July 17,2024 at 09:09:22 UTC from IEEE Xplore. Restrictions apply.
YUE et al.: EXPLOITING FULL/HALF-DUPLEX USER RELAYING IN NOMA SYSTEMS 561

namely protected zone and artificial noise for enhancing the NOMA with a direct link was only considered with HD opera-
secrecy performance of NOMA networks with the aid of tion mode. In [34], the authors investigated the performance of
stochastic geometry. FD device-to-device based cooperative NOMA. However, only
Cooperative communication is a particularly effective the outage performance of far user was analyzed. To the best
approach by providing the higher diversity as well as extending of our knowledge, there is no existing work investigating the
the coverage of networks [20]. Current NOMA research contri- impact of the direct link for FD user relaying on the network
butions in terms of cooperative communication mainly include performance, which motivates us to develop this treatise. Also,
two aspects. The first aspect is the application of NOMA into there is lack of systematic performance evaluation metrics
cooperative networks [21]–[24]. The coordinated two-point i.e., considering ergodic rate and energy efficiency in terms of
system with superposition coding (SC) was researched in the FD/HD NOMA systems. Different from [25], [34], we present
downlink communication in [21]. The authors in [22], [23] a comprehensive investigation on adopting near user as a
investigated outage probability and system capacity of decode- FD/HD relaying to improve the reliability of far user. More
and-forward (DF) relaying for NOMA. In [24], the outage specifically, we attempt to explore the potential ability of user
behavior of amplify-and-forward (AF) relaying with NOMA relaying in NOMA networks with identifying the following
has been discussed over Nakagami-m fading channels. The key impact factors.
second aspect is cooperative NOMA, which was first pro- • Will FD NOMA relaying bring performance gains com-
posed in [25]. The key idea of cooperative NOMA is to pared to HD NOMA relaying? If yes, what is the
regard the near NOMA user as a DF user relaying to help condition?
far NOMA user. On the standpoint of considering energy • What is the impact of direct link on the considered
efficiency issues, simultaneous wireless information and power system? Will it significantly improve the network per-
transfer (SWIPT) was employed at the near NOMA user, formance in terms of outage probability and throughput?
which was regarded as DF relaying in [26]. • Will NOMA relaying bring performance gains com-
Although cooperative NOMA is capable of enhancing the pared to conventional orthogonal multiple access (OMA)
performance gains for far user, it results in additional band- relaying?
width costs for the system. To avoid this issue, one promising • In delay-limited/tolerant transmission modes, what are the
solution is to adopt the full-duplex (FD) relay technology. relationships between energy efficiency (EE) and HD/FD
FD relay receives and transmits simultaneously in the same NOMA systems?
frequency band, which is the reason why it has attracted signif-
icant interest to realize more spectrally efficient systems [27].
B. Contributions
In a general case, due to the imperfect isolation or cancellation
process, FD operation may suffer from residual loop self- In this paper, we propose a comprehensive NOMA
interference (LI) which is modeled as a fading channel. With user relaying system, where near user can switch between
the development of signal processing and antenna technolo- FD and HD mode according to the channel conditions.
gies, relaying with FD operation is feasible [28]. Recently, We also consider the setting of two scenarios in which
FD relay technologies have been proposed as a promising the direct link exists or not between the BS and far
technique for 5G networks in [29]. Two main types of FD user. Based on our proposed NOMA user relaying systems,
relay techniques, namely FD AF relaying and FD DF relaying, the primary contributions of this paper are summarised as
have been discussed in [30]–[32]. The expressions for outage follows:
probability of FD AF relaying were provided in [30], which 1) Without direct link: We derive the closed-form expres-
considers the processing delay of relaying in practical scenar- sions of outage probability for the near user and far user,
ios. In [31], the performance of FD AF relaying in terms of respectively. For obtaining more insights, we further
outage probability was investigated considering the direct link. derive the asymptotic outage probability of two users
The authors in [32] characterized the outage performance of and obtain diversity orders at high SNR. We demonstrate
FD DF relaying. It is demonstrated that the optimal duplex that FD NOMA converges to an error floor and results
mode can be selected according to the outage probability. in a zero diversity order. We show that FD NOMA is
Furthermore, the operations of randomly switching between superior to HD NOMA in terms of outage probability in
FD and HD mode were considered for enhancing spectral the low SNR region rather than in the high SNR region.
efficiency in [33]. In addition, we also obtain the diversity orders of two
users for HD NOMA. Furthermore, we analyze the sys-
A. Motivations and Related Works tem throughput in delay-limited transmission according
While the aforementioned research contributions have laid to the derived outage probability.
a solid foundation with providing a good understanding of 2) Without direct link: We study the ergodic rate of
cooperative NOMA and FD relay technology, the treatises two users for FD/HD NOMA. To gain better insights,
for investigating the potential benefits by integrating these we derive the asymptotic ergodic rates of two users and
two promising technologies are still in their infancy. Some obtain the high SNR slopes. We demonstrate that the
related cooperative NOMA studies have been investigated ergodic rate of far user converges to a throughput ceiling
in [25], [34]. In [25], it is demonstrated that the maximum for FD/HD NOMA in the high SNR region. Moreover,
diversity order can be obtained for all users, but cooperative we also demonstrate that FD NOMA outperforms HD

Authorized licensed use limited to: ULAKBIM UASL - Yakin Dogu Universitesi. Downloaded on July 17,2024 at 09:09:22 UTC from IEEE Xplore. Restrictions apply.
562 IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 66, NO. 2, FEBRUARY 2018

NOMA in terms of ergodic sum rate in the low SNR

region.
3) With direct link: We first derive the closed-form expres-
sion in terms of outage probability for far user. In order
to get the corresponding diversity order, we also derive
the approximated outage probability of far user. We find
that the reliability of far user is improved with the help
of direct link. We confirm that the use of direct link
overcomes the zero diversity order of far user inherent to
conventional FD relaying. For the near user, the diversity
order is the same as that of FD relaying. Additionally, Fig. 1. A downlink FD/HD cooperative NOMA system model.
we conclude that the superiority of FD NOMA is no
longer apparent with the values of LI increasing. independent non-selective block Rayleigh fading and are dis-
4) With direct link: We analyze the ergodic rate of far user turbed by additive white Gaussian noise with mean power N0 .
for FD/HD NOMA. For this scenario, it is the fact that h 1 , h 2 , and h 0 are denoted as the complex channel coefficient
the ergodic rate of near user is invariant which is not of B S → D1 , D1 → D2 , and B S → D2 links, respectively.
affected by the direct link. Similarly, we also derive the The channel power gains |h 1 |2 , |h 2 |2 and |h 0 |2 are assumed
approximated expressions for ergodic rate and obtain the to be exponentially distributed random variables (RVs) with
high SNR slopes. We demonstrate that the use of direct the parameters i , i ∈ {1, 2, 0}, respectively. When D1
link is incapable of assisting far user to obtain additional operates in FD mode, we assume that an imperfect self-
high SNR slope. interference cancellation scheme2 is executed at D1 such as
5) Energy efficiency: We derive expressions in terms of in [31] and [35]. The LI is modeled as a Rayleigh fading
energy efficiency for FD/HD NOMA. We conclude that channel with coefficient h L I , and  L I is the corresponding
FD NOMA without/with direct link have a higher energy average power. To analyze HD NOMA, we introduce the
efficiency corresponding to HD NOMA in the low SNR switching operation factor detailed in the following.
region for delay-limited transmission mode. However, During the k-th time slot, according to [12], D1 receives the
in delay-tolerant transmission mode, the system energy superposed signal and loop interference signal simultaneously.
efficiency of HD NOMA exceeds FD NOMA with- The observation at D1 is given by
out/with direct link.  
y D1 [k] = h 1 ( a1 Ps x 1 [k] + a2 Ps x 2 [k])

+ h L I  Pr x L I [k − τ ] + n D1 [k], (1)
C. Organization and Notation
where  is the switching operation factor between FD and
The rest of the paper is organized as follows. In Section II, HD mode.  = 1 and  = 0 denote D1 working in FD and
the system model of user relaying for FD NOMA is set up. HD mode, respectively. Based on the practical application sce-
In Section III, the analytical expressions for outage probability, narios, we can select the different operation mode. x L I [k − τ ]
diversity order and throughput of FD/HD user relaying are denotes loop interference signal and τ denotes the processing
derived and analyzed. In Section IV, the performance of user delay at D1 with an integer τ ≥ 1. More particularly,
relaying for FD/HD NOMA are evaluated in terms of ergodic we assume that the time k satisfies the relationship k ≥ τ .
rate. Section V considers the system energy efficiency for Ps and Pr are the normalized transmission powers at the BS
FD/HD NOMA systems. Analytical results and simulations and D1 , respectively. x 1 and x 2 are the signals for D1 and D2 ,
are presented in Section VI. Section VII concludes the paper. respectively. a1 and a2 are the corresponding power allocation
The main notations of this paper is shown as follows: E{·} coefficients. To stipulate better fairness between the users,
denotes expectation operation; f X (·) and FX (·) denote the we assume that a2 > a1 with a1 +a2 = 1. The SIC3 [36] can be
probability density function (PDF) and the cumulative distri- invoked by D1 for first detecting D2 having a larger transmit
bution function (CDF) of a random variable X; ∝ represents power, which has less inference signal. Then the signal of
“be proportional to”. D2 can be detected from the superposed signal. Therefore,
the received signal-to-interference-plus-noise ratio (SINR) at
II. S YSTEM M ODEL D1 to detect D2 ’s message x 2 is given by

We consider a FD cooperative NOMA system consisting of |h 1 |2 a2 ρ

γ D2 →D1 = , (2)
one source, i.e, the BS, that intends to communicate with far |h 1 |2 a1 ρ +  |h L I |2 ρ + 1
user D2 via the assistance of near user D1 illustrated in Fig. 1.
2 LI refers to the signals that are transmitted by a FD relaying and
D1 is regarded as user relaying and DF protocol is employed looped back to the receiver simultaneously. Through radio frequency (RF)
to decode and forward the information to D2 . To enable FD cancellation, antenna cancellation and signal process technologies, etc, those
communication, D1 is equipped with one transmit antenna and LI can be suppressed to a lower level. However, LI still remains in the receiver
due to imperfect self-interference cancellation, when decoding the desired
one receive antenna, while the BS and D2 are single-antenna signal.
nodes. Note that D1 can switch operation between FD and 3 It is assumed that perfect SIC is employed at D , our future work will
1
HD mode. All wireless links in network are assumed to be relax this ideal assumption.

Authorized licensed use limited to: ULAKBIM UASL - Yakin Dogu Universitesi. Downloaded on July 17,2024 at 09:09:22 UTC from IEEE Xplore. Restrictions apply.
YUE et al.: EXPLOITING FULL/HALF-DUPLEX USER RELAYING IN NOMA SYSTEMS 563

where ρ = NPs0 is the transmit signal-to-noise radio (SNR). A. User Relaying Without Direct Link
Note that x 1 and x 2 are supposed to be normalized unity power In this subsection, the first scenario is investigated in terms
signals, i.e, E{x 12 } = E{x 22 } = 1. of outage probability.
After SIC, the received SINR at D1 to detect its own 1) Outage Probability of D1 : According to NOMA pro-
message x 1 is given by tocol, the complementary events of outage at D1 can
|h 1 |2 a1 ρ be explained as: D1 can detect x 2 as well as its own
γ D1 = . (3) message x 1 . From the above description, the outage probability
 |h L I |2 ρ + 1 of D1 is expressed as
In the FD mode, the √ received signal √ at D2 is writ-  
ten√ as y D2 [k] = h 0 ( a1 Ps x 1 [k] + a2 Ps x 2 [k]) + PDF1D = 1 − Pr γ D2 →D1 > γtFh 2D , γ D1 > γtFh 1D , (9)
h 2 Pr x 2 [k − τ ] + n D2 [k]. However, the observation √ at D2
where  = 1. γtFh 1D = 2 R1 − 1 with R1 being the target rate at
for
√ the direct link is written as y 1,D [k] = h 0 ( a 1 Ps x 1 [k] +
2
a2 Ps x 2 [k]) + n D2 [k]. Due to the existence of residue inter- D1 to detect x 1 and γtFh 2D = 2 R2 − 1 with R2 being the target
ference (RI) from relaying link, the received SINR at D2 to rate at D1 to detect x 2 .
detect x 2 for direct link is given by The following theorem provides the outage probability of
D1 for FD NOMA.
RI |h 0 |2 a2 ρ Theorem 1: The closed-form expression for the outage
γ1,D = , (4)
2
|h 0 a1 ρ + κ|h 2 |2 ρ + 1
|2 probability of D1 is given by
where κ denotes the impact levels of RI. Since DF relaying 1 θ
− 1
PDF1D = 1 − e 1 , (10)
protocol is invoked in D1 , we assume that D1 can decode and 1 + ρ θ1  L I
forward the signal x 2 to D2 successfully for relaying link from γthF D
D1 to D2 . As a consequence, the observation at D2 for relaying where  = 1. θ1 = max (τ1 , β1 ), τ1 =  2  and
√ ρ a2 −a1 γthF D
link is written as y2,D2 [k] = h 2 Pr x 2 [k − τ ] + n D2 [k]. γ FD
2

Similarly, considering the impact of RI from direct link, β1 = ath1 ρ1 . Note (10) is derived on the condition of a2 >
the received SINR at D2 to x 2 for relaying link is given by a1 γtFh 2D .
Proof: By definition, J1 denotes the complementary event
|h 2 |2 ρ
γ2,D
RI
= . (5) at D1 and is calculated as
2
κ|h 0 |2 ρ + 1    
J1 = Pr |h 1 |2 ≥  |h L I |2 ρ + 1 θ1
As stated in [31] and [33], the relaying link correspond-  ∞ ∞
ing to direct link from BS to D2 has small time delay
= f |h L I |2 (x) f|h 1 |2 (y)d x d y
for any transmitted signals. In other words, there is some 0 (xρ+1)θ1
temporal separation between the signal from D1 and BS. 1 θ
− 1
To derive the theoretical results for practical NOMA systems, = e 1 . (11)
1 + ρ θ1  L I
we assume that these signals from D1 and BS are fully
resolvable by D2 [34]. Hence, we provide the upper bounds Substituting (11) into (9), (10) can be obtained and the proof
of (4) and (5) in the following parts, which are the received is completed.
SINRs at D2 to detect x 2 for direct link and relaying link, i.e. Corollary 1: Based on (10), the outage probability of D1
for HD NOMA with  = 0 is given by
|h 0 |2 a2 ρ θ
γ1,D2 = , (6) PDH1D = 1 − e
− 2
1 , (12)
|h 0 a1 ρ + 1
|2
and where γtHh 1D = 22R1 −1 and γtHh 2D = 22R2 −1 denote the target
SNRs at D1 to detect x 1 and x 2 with HD mode, respectively.
γ2,D2 = |h 2 |2 ρ, (7) γthH D γthH D
θ2 = max (τ2 , β2 ), β2 = α1 ρ
1
and τ2 = 2
with
ρ (a2 −a1 γthH D )
2
respectively. At this moment, the signals from the relaying a2 > a1 γtHh 2D .
link and direct link are combined by maximal ratio combin- 2) Outage Probability of D2 : The outage events of D2 can
ing (MRC) at D2 . So the received SINR after MRC at D2 is be explained as below. The first is that D1 cannot detect x 2 .
given by The second is that D2 cannot detect its own message x 2 on
|h 0 |2 a2 ρ the conditions that D1 can detect x 2 successfully. Based on
γ DM2 RC = |h 2 |2 ρ + . (8) these, the outage probability of D2 is expressed as
|h 0 |2 a1 ρ + 1  
PDF2D,nodir = Pr γ D2 →D1 < γtFh 2D
III. O UTAGE P ROBABILITY  
+ Pr γ2,D2 < γtFh 2D , γ D2 →D1 > γtFh 2D , (13)
When the target rate of users is determined by its quality
of service (QoS), the outage probability is an important where  = 1.
metric for performance evaluation. We will evaluate the outage The following theorem provides the outage probability of
performance in two representative scenarios in the following. D2 for FD NOMA.

Authorized licensed use limited to: ULAKBIM UASL - Yakin Dogu Universitesi. Downloaded on July 17,2024 at 09:09:22 UTC from IEEE Xplore. Restrictions apply.
564 IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 66, NO. 2, FEBRUARY 2018

Theorem 2: The closed-form expression for the outage d) D2 for HD NOMA case: Based on (17), the asymptotic
probability of D2 without direct link is given by outage probability of D2 for HD NOMA is given by
 
τ1
FD
γth γtHh 2D τ2 1
− 1 + ρ2
2
PDH2D,∞
1 ,nodir = + ∝ . (22)
PDF2D,nodir =1− e , (14) ρ2 1 ρ
1 + ρ τ1 LI
Substituting (22) into (18), we can obtain d DH2D,nodir = 1.
where  = 1.
Proof: By definition, J2 and J3 denote the first and second Remark 3: As can be observed that PDF1D,∞ and PDF2D,∞ ,nodir
outage events, respectively. The process calculated is given by are a constant independent of ρ, respectively. Substituting (19)
   and (21) into (18), we see that there are the error floors for
J2 = Pr |h 1 |2 < τ1  |h L I |2 ρ + 1 outage probability of two users.
 ∞  τ1 ( yρ+1) 4) Throughput Analysis: In this subsection, the delay-
= f |h 1 |2 (x) f |h L I |2 (y) d x d y limited transmission mode [26], [37] is considered for FD/HD
0 0 NOMA.
1 τ
− 1 a) FD NOMA case: In this mode, the BS transmits
= 1− e 1 . (15)
1 + ρ τ1 LI information at a constant rate R, which is subject to the
Applying some algebraic manipulations, J3 is given by: effect of outage probability due to wireless fading channels.
 FD 
γth
The system throughput of FD NOMA without direct link is
1 τ
− 1 − ρ2 given by
J3 = e 1 1−e 2 . (16)    
1 + ρ τ1 LI FD
Rl_nodir = 1 − PDF1D R1 + 1 − PDF2D,nodir R2 , (23)
Combining (15) and (16), (14) can be obtained and the proof
is completed. where PDF1D and PDF2D,nodir are given in (10) and (14),
Corollary 2: Based on (14), the outage probability of D2 respectively.
without direct link for HD NOMA with  = 0 is given by b) HD NOMA case: Similar to (23), the system through-
put of HD NOMA without direct link is given by
τ
HD
γth    
− 2 − 2
HD
PDH2D,nodir =1−e 1 ρ2 . (17) Rl_nodir = 1 − PDH1D R1 + 1 − PDH2D,nodir R2 , (24)
3) Diversity Analysis: To get more insights, the asymptotic where PDH1D and PDH2D,nodir are given in (12) and (17),
diversity analysis is provided in terms of outage probabil- respectively.
ity investigated in high SNR region. The diversity order is
defined as B. User Relaying With Direct Link
 In this subsection, we explore a more challenging scenario,
log PD∞ (ρ)
d = − lim . (18) where the direct link between the BS and D2 is used to convey
ρ→∞ log ρ information and system reliability can be improved. However,
a) D1 for FD NOMA case: Based on analytical result the outage probability of D1 will not be affected by the direct
in (10), when ρ → ∞, the asymptotic outage probability of link. As such, we only show outage probability of D2 in the
D1 for FD NOMA with e−x ≈ 1 − x is given by following.
1 1) Outage Probability of D2 : For the second scenario,
PDF1D,∞ = 1 − . (19) the outage events of D2 for FD NOMA is described as
1 + ρθ1  L I
below. One of the events is when x 2 can be detected at D1 ,
Substituting (19) into (18), we can obtain d DF1D = 0. but the received SINR after MRC at D2 in one slot is less
Remark 1: The diversity order of D1 is zero, which is the than its target SNR. Another event is that neither D1 nor
same as the conventional FD relaying. D2 can detect x 2 . Therefore, the outage probability of D2 is
b) D1 for HD NOMA case: Based on analytical result expressed as
in (12), the asymptotic outage probability of D1 for HD  
PDF2D,R
,dir
I
= Pr γ RI
1,D2 + γ RI
< γ FD
, γ D2 →D1 > γ FD
NOMA is given by  2,D2 t h2 t h2

FD RI FD
θ2 1 + Pr γ D2 →D1 < γt h 2 , γ1,D2 < γt h 2 . (25)
PDH1D,∞ = ∝ . (20)
1 ρ Unfortunately, the closed-form expression of (25) for D2 can
Substituting (20) into (18), we can obtain d DH1D = 1. not be derived successfully. However, it can be evaluated by
c) D2 for FD NOMA case: Based on (14), the asymptotic using numerical simulations. To further obtain a theoretical
outage probability of D2 for FD NOMA is given by result for D2 , exploiting the upper bounds of received SINRs
derived in (6) and (7), the outage probability of D2 is
1 2 ρ − 1 γtFh 2D − τ1 ρ2 expressed as
PDF2D,∞
,nodir = 1 − . (21)  
2 ρ (1 + τ1 ρLI )
PDF2D,dir = Pr γ DM2 RC < γtFh 2D , γ D2 →D1 > γtFh 2D
 
Substituting (21) into (18), we can obtain d DF2D,nodir = 0. + Pr γ D2 →D1 < γtFh 2D , γ1,D2 < γtFh 2D , (26)
Remark 2: The diversity order of D2 is zero, which is the
same as D1 in FD NOMA. where  = 1.

Authorized licensed use limited to: ULAKBIM UASL - Yakin Dogu Universitesi. Downloaded on July 17,2024 at 09:09:22 UTC from IEEE Xplore. Restrictions apply.
YUE et al.: EXPLOITING FULL/HALF-DUPLEX USER RELAYING IN NOMA SYSTEMS 565

The following theorem provides the outage probability of given by

D2 for FD NOMA.    
Theorem 3: The closed-form expression for the outage FD
Rl_dir = 1 − PDF1D R1 + 1 − PDF2D,dir R2 , (29)
probability of D2 with direct link is given by
where PDF1D and PDF2D,dir can be obtained from (10) and (27),
PDF2D,dir respectively.
∞
τ
− 1 (−1)n eϕ (−1)2n+1 φ1 n+1 b) HD NOMA case: Similar to (29), the system through-
= 1−e 0 − put of HD NOMA with direct link is given by
n!φ2 n+1 (n + 1)!
n=0
    
HD
n
(1+a1ρτ1 )n+1 eψ ψ k −eφ1 φ1 k Rl_dir = 1 − PDH1D R1 + 1 − PDH2D,dir R2 , (30)
× (Ei (ψ)−Ei (φ1 )) +
(n + 1) n · · · (n + 1 − k)
τ1 
k=0
τ1   τ 
where PDH1D and PDH2D,dir can be obtained from (12)
− − − 1
× χe 1 + 1 − χe 1 1 − e 0 , (27) and [25, eq. (11)].

γ FD
where ϕ = 1
− ρ
ρa1 0
th 2
2
− φ1 , φ1 = a1−a
ρ2 , φ2 = a1 ρ0 , ψ =
2 IV. E RGODIC R ATE
−a2 1
a1 ρ2 (1+a1 ρτ1 ) and χ = 1 +τ1 ρLI . Ei (·) is the exponential When user’s rates are determined by their channel con-
integral function [38, eq. (8.211.1)]. ditions, the ergodic sum rate is an important metric for
Proof: See Appendix A. performance evaluation. Hence the performance of FD/HD
2) Diversity Analysis: In this subsection, the diversity order user relaying are characterized in terms of ergodic sum rates
of D2 with direct link for FD NOMA is analyzed in the in the following.
following.
a) D2 for FD NOMA case: For D2 with direct link,
it is challenging to obtain diversity order from (27). We can A. User Relaying Without Direct Link
use Gaussian-Chebyshev quadrature to find an approximation 1) Ergodic Rate of D1 : On the condition that D1 can
 x 2 , the achievable rate of D1 can be written as R D1 =
from (26) and the approximated expression of outage proba- detect
bility for D2 at high SNR is given by log 1 + γ D1 . The ergodic rate of D1 for FD NOMA can be
F D,appro
obtained in the following theorem.
PD2 ,dir Theorem 4: The closed expression of ergodic rate for D1
⎡ ⎛  ⎞ without direct link for FD NOMA is given by
τ1  2 τ 1 + 2 0 τ 1 a 2 − a 1 γ FD
t h2
=⎣ − ⎝1 − ⎠ a1 1
0 20 2
FD
RD =
1
ln 2 ( L I − a1 1 )
    
  1 −1 1 −1
τ1 π
N
(sn + 1) τ1 a2 sn τ1 × e a 1 ρ 1 Ei −e ρ L I
Ei . (31)
× 1+ − a1 ρ1 ρ L I
2N0 2 ((sn + 1) τ1 a1 ρ + 2) 20
n=1
Proof: See Appendix B.
⎤
   As such, we can derive the ergodic rate of D1 for HD
1 1 τ1
× 1 − sn2 ⎦ + 1− , NOMA in the following corollary.
(1 +τ1 ρLI ) (1 +τ1 ρLI ) 0 Corollary 3: The ergodic rate of D1 for HD NOMA is given
by
(28)
1  
−e a1 ρ1 −1
where N is a parameter
 to ensure a complexity-accuracy HD
RD = Ei . (32)
tradeoff, sn = cos 2n−1 π . Substituting (28) into (18), we can
1
2 ln 2 a1 ρ1
2N
obtain d DF2D,dir = 1. 2) Ergodic Rate of D2 : Since x 2 should be detected at
Remark 4: From above explanation, the observation is that D2 as well as at D1 for SIC, the achievable rate of D2
the direct link (B S → D2 ) to convey information is an
effective way to overcome the problem of zero diversity order
without
  link for FD NOMA is written as R D2 =
direct
log 1 + min γ D2 →D1 , γ2,D2 . The corresponding ergodic
for D2 . rate is given by
b) D2 for HD NOMA case: The outage performance of  ∞
D2 for HD NOMA has been investigated in [25] and we can FD 1 1 − FX 1 (x 1 )
R D2 ,nodir = d x1, (33)
obtain d DH2D,dir = 2. ln 2 0 1 + x1
3) Throughput Analysis: Based on the derived results of  
where X 1 = min |h |2 |h 1 | a|h2 ρ |2
2
outage probability above, we obtain the throughput expres- , |h 2 |2 ρ with  = 1.
1 a1 ρ+ L I ρ+1
sions for FD/HD NOMA in delay-limited transmission mode Obviously, it is difficult to obtain the CDF of X 1 . However,
as below. in order to derive an accurate closed-form expression for the
a) FD NOMA case: As suggested in Section III-A4, ergodic rate applicable to high SNR region, the following
the system throughput of FD NOMA with direct link is theorem provides the high SNR approximation.

Authorized licensed use limited to: ULAKBIM UASL - Yakin Dogu Universitesi. Downloaded on July 17,2024 at 09:09:22 UTC from IEEE Xplore. Restrictions apply.
566 IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 66, NO. 2, FEBRUARY 2018

Theorem 5: The asymptotic expression for ergodic rate of d) D2 for HD NOMA case: Such as (IV-A3.c), substitut-
D2 without direct link for FD NOMA in the high SNR region ing (36) into (37), we can obtain S DH2D,nodir = 0.
is given by Remark 5: Based on above analysis, the ergodic rate of D2
F D,∞
converges to a throughput ceiling in the high SNR region for
RD 2 ,nodir FD/HD NOMA without direct link.
     Combing (34) and (38), the asymptotic expression for
1 1 −1 −1
= e ρ2 Ei − Ei ergodic sum rate of FD NOMA without direct link is expressed
ln 2 ρa1 2 ρ2
as
  a2  1   
1 e ρ2 ξ a2 ξ + a1 a2 1 F D,∞
× − Ei − Rsum,nodir
a2 1 − ξ ξ ρa1 ξ 2  
    a1 1 1
a2 1 a1 a2 1 + a2 1 ξ
2 = 1+
− Ei − , (34) ln 2 ( L I − a1 1 ) a ρ
ρ2 ξ a2 1 − ξ      1 1    
1 1 1
× ln +C − 1+ ln +C
where ξ = ( L I − a1 1 ). a1 ρ1 ρ L I ρ L I
      
Proof: See Appendix C. 1 1 −1 −1 1
For  = 0, the ergodic rate of D2 without direct link for + ρ
e 2 Ei − Ei
ln 2 ρa1 2 ρ2 a2 1 − ξ
HD NOMA is given by a2  1     
 a2 −ρ y
− y e ρ2 ξ −a2 1 −a2 ξ − a1 a2 1
1 a1 e (a2 −ya1 )1 ρ2 + Ei − Ei
HD
RD = d y. (35) ξ ρ2 ξ ρa1 ξ 2
2 ,nodir
2 ln 2 1+ y  
0
a1 a2 1 + a2 1 ξ
2
As can be seen from the above expression, (35) does not × . (40)
a2 1 − ξ
have a closed-form solution. Corollary 4 gives the high SNR
approximation. Similarly, combing (36) and (39), the asymptotic expression
Corollary 4: The asymptotic expression for ergodic rate of for ergodic sum rate of HD NOMA without direct link is
D2 without direct link for HD NOMA in the high SNR region expressed as
is given by     
H D,∞ −1 1 1
1      Rsum,nodir = 1− ln +C
H D,∞ e ρ2 −1 −1 2 ln 2 a1 ρ1 a1 ρ1
RD 2 ,nodir
= Ei − Ei . (36) 1     
2 ln 2 ρa1 2 ρ2 e ρ2 −1 −1
+ Ei − Ei . (41)
Proof: See Appendix D. 2 ln 2 ρa1 2 ρ2
3) Slope Analysis: In this subsection, the high SNR slope
4) Throughput Analysis: In this subsection, the throughput
is evaluated, which is the key parameter determining ergodic
in delay-tolerant transmission for FD/HD NOMA are pre-
rate in high SNR region. The high SNR slope is defined as
sented, respectively.
R∞D (ρ) a) FD NOMA case: In this mode, the throughput is
S = lim . (37) determined by evaluating the ergodic rate. Using (31) and (33),
ρ→∞ log (ρ)
the system throughput of FD NOMA without direct link is
a) D1 for FD NOMA case: Based on (31), when given by
ρ → ∞, by using Ei (−x) ≈ ln (x)+C [38, eq. (8.212.1)] and
e−x ≈ 1 − x, where C is the Euler constant, the asymptotic RtF_nodir
D FD
= RD FD
+ RD . (42)
1 2 ,nodir
ergodic rate of D1 for FD NOMA is given by
    b) HD NOMA case: Similar to (42), using (32) and (35),
F D,∞ a1 1 1 1 the system throughput of HD NOMA without direct link is
R D1 = 1+ ln
ln 2 ( L I − a1 1 ) a ρ a ρ given by
    1 1  1 1
1 1
+C − 1+ ln + C . (38) RtH_nodir
D
= RD
HD
+ RD
HD
. (43)
ρ L I ρ L I 1 2 ,nodir

F D = 0.
Substituting (38) into (37), we can obtain S D 1
b) D1 for HD NOMA case: Based on (32), the asymptotic B. User Relaying With Direct Link
ergodic rate of D1 for HD NOMA in the high SNR region is In this subsection, we investigate the ergodic rate of D2 for
given by FD/HD NOMA with direct link.
     1) Ergodic Rate of D2 : Assume that the signal x 2 from
H D,∞ −1 1 1
RD = 1 − ln + C . (39) relaying and direct link can be detected at D2 as well
1 2 ln 2 a1 ρ1 a1 ρ1
as at D1 for SIC. Moreover, considering the effect of RI
Substituting (39) into (37), we can obtain S DH1D = 12 . 
between these two links, the achievable rate of D2 is written
c) D2 for FD NOMA case: Based on above analysis, as R DRI
2 ,dir
= log 1 + min γ D 2 →D 1 , γ RI + γ RI
1,D2 2,D2 . For
FD
substituting (34) into (37), we can obtain S D 2 ,nodir
= 0. the sake of simplicity, the achievable rate for D2 can be

Authorized licensed use limited to: ULAKBIM UASL - Yakin Dogu Universitesi. Downloaded on July 17,2024 at 09:09:22 UTC from IEEE Xplore. Restrictions apply.
YUE et al.: EXPLOITING FULL/HALF-DUPLEX USER RELAYING IN NOMA SYSTEMS 567

  
further written as R D2 ,dir = log 1 + min γ D2 →D1 , γ DM2 RC . TABLE I
Hence, the ergodic rate of D2 for FD NOMA is given by D IVERSITY O RDER AND H IGH SNR S LOPE FOR FD/HD NOMA S YSTEMS
 ∞
FD 1 1 − FX 2 (x 2 )
R D2 ,dir = d x2, (44)
ln 2 0 1 + x2
 
|h 1 |2 a2 ρ |h 0 |2 a2 ρ
where X 2 = min |h |2 a ρ+ |h |2 ρ+1
, |h 2 | 2
ρ + |h |2 a ρ+1
1 1 LI 0 1
with  = 1. It is also difficult to obtain the CDF of X 2
and (44) has not closed-form expression. The following theo-
rem provides the high SNR approximation.
Theorem 6: The asymptotic expression for ergodic rate of
D2 with direct link for FD NOMA in the high SNR region is
link is expressed as
given by     
   H D,∞ −1 1 1
1 a2 1 1 Rsum,dir = 1− ln +C
F D,∞
= + − ln 2 ln 2 a1 ρ1 a1 ρ1
RD 2 ,dir
ln 1  
ln 2 a1 a2 1 − ξ ξ 1 a2
   + log 1 +
2 a1
. (49)
ξ a2 1 ξ + a1 a2 21
× 1+ . (45)
a1 1 a2 1 − ξ 2) Slope Analysis: Based on the derived asymptotic ergodic
rates, the high SNR slopes of D2 with direct link are charac-
Proof: See Appendix E. terized in the following.
For  = 0, the ergodic rate of D2 for HD NOMA with a) D2 for FD NOMA case: Substituting (45) into (37),
direct link is given by we can obtain S DFD
2 ,dir
= 0.
HD
b) D2 for HD NOMA case: Such as (IV-B2.a), substitut-
RD 2 ,dir ing (47) into (37), we can obtain S DH2D,dir = 0.
y (0 +1 )
 a2 −ρ
(a2 −a1 y )0 1
Remark 6: Based on above derived results, the ergodic rate
a1 e
= dy of D2 also converge to a throughput ceiling in the high SNR
0 1+ y region with direct link for FD/HD NOMA. The user of direct
γth (xa1 ρ+1)−xa2 ρ
 a2  y − x − 2
ρ (xa1 ρ+1)2
− ρ a −ay y 
( 2 1 ) 1
link is incapable of assisting D2 to obtain the additional high
a1 ρ (a2 −a1 y ) e 0
+ d x d y. SNR slope.
0 0 (1 + y) 0 3) Throughput Analysis:
(46) a) FD NOMA case: As suggested in Section IV-A4,
in delay-tolerant transmission mode, using (31) and (44),
To obtain the closed-form expression of ergodic rate for D2 ,
the system throughput for FD NOMA with direct link is
the complicated integrals are required to be computed.
given by
Corollary 5 gives an efficient high SNR approximation.
Corollary 5: The asymptotic expression for ergodic rate of RtF_dir
D
= RD
FD
1
+ RD
FD
2 ,dir
. (50)
D2 with direct link for HD NOMA in the high SNR region is
given by b) HD NOMA case: Similar to (50), using (32) and (46),
  the system throughput for HD NOMA with direct link is
H D,∞ 1 a2 given by
R D2 ,dir = log 1 + . (47)
2 a1 RtH_dir
D
= RD
HD
+ RD
HD
. (51)
1 2 ,dir
Combing (38) and (45), the asymptotic expression for the As shown in TABLE I, the diversity orders and high SNR
ergodic sum rate of FD NOMA with direct link is expressed as slopes of two users for FD/HD NOMA are summarized to
F D,∞
Rsum,dir illustrate the comparison between them. In TABLE I, we use
    “D” and “S” to represent the diversity order and high SNR
a1 1 1 1
= 1+ ln slope, respectively.
ln 2 ( L I − a1 1 ) a ρ a1 ρ1
    1 1  
1 1 1 V. E NERGY E FFICIENCY
+C − 1+ ln +C +
ρ L I ρ L I ln 2 Based on throughput analysis, we aim to provide the system
    
a2 1 1 ξ energy efficiency (EE) considering user relaying for FD/HD
× ln 1 + − ln 1 + NOMA systems.
a1 a2 1 − ξ ξ a1 1
  The definition of energy efficiency is given by
a2 1 ξ + a1 a2 21
× (48) Total data rate
a2 1 − ξ ηE E = . (52)
Total energy consumption
In the same way, combing (39) and (47), the asymptotic For FD/HD NOMA energy efficiency, the total data rate is
expression for ergodic sum rate of HD NOMA with direct denotes as sum throughput from the BS to D1 and D2 and

Authorized licensed use limited to: ULAKBIM UASL - Yakin Dogu Universitesi. Downloaded on July 17,2024 at 09:09:22 UTC from IEEE Xplore. Restrictions apply.
568 IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 66, NO. 2, FEBRUARY 2018

from D1 to D2 . The total power consumption is denoted as

the sum of the transmitted power Ps at the BS and Pr at D1 .
Based on results in Section III-A4, III-B3 and IV-A4,
IV-B3, the energy efficiency of user relaying for FD/HD
NOMA systems are expressed as
RFD
η
FD
= , (53)
T Ps + T Pr
and
2RHD
HD
η = , (54)
T Ps + T Pr
respectively. T denotes the transmission time for the
entire communication process.  ∈ (l_nodir, l_dir,
t_nodir, t_dir ). ηl_nodir and ηl_dir are system energy effi-
ciencies without/with direct link in delay-limited transmission
mode, respectively. ηt _nodir and ηt _dir are system energy effi-
ciencies without/with direct link in delay-tolerant transmission Fig. 2. Outage probability versus the transmit SNR without direct link.
mode, respectively.

VI. N UMERICAL R ESULTS

In this section, simulation results are provided to validate
our analytical expressions derived in the previous section, and
further evaluate the performance of FD/HD user relaying in
NOMA systems. Without loss of generality, we assume that
the distance between BS and D2 is normalized to unity, i.e.
 S D2 = 1.  S D1 = d −α and  D1 D2 = (1 − d)−α , where d
is the normalised distance between BS and D1 setting to be
d = 0.3 and α is the pathloss exponent setting to be α = 2.
The power allocation coefficients of NOMA are a1 = 0.2 and
a2 = 0.8 for D1 and D2 , respectively.

A. Without Direct Link

For user relaying without direct link, the target rate is set
to be R1 = 3, R2 = 0.5 bit per channel use (BPCU) for D1 Fig. 3. System throughput in delay-limited transmission mode versus SNR
and D2 , respectively. The performance of conventional OMA with different values of LI without direct link.
is shown as a benchmark for comparison, in which the total
communication process is finished in three slots. In the first
slot, the BS sends information x 1 to D1 and sends x 2 to D1 well approximate the exact performance curves in the high
in the second slot. In the last slot, D1 decodes and forwards SNR region. It is shown that error floors exist in FD NOMA,
the information x 2 to D2 . which verify the conclusion in Remark 3 and obtain zero
1) Outage Probability: Fig. 2 plots the outage probability of diversity order. This is due to the fact that there is loop
two users versus SNR without direct link and the value of LI interference in FD NOMA. Another observation is that HD
is assumed to be E{|h L I |2 } = −15 dB. The exact theoretical NOMA and OMA are superior to FD NOMA in the high SNR
curves for the outage probability of two users for FD/HD region. Therefore, we can select different operation mode for
NOMA are plotted according to (10), (14) and (12), (17), user relaying according to the different SNR levels in practical
respectively. Obviously, the exact outage probability curves cooperative NOMA systems.
match precisely with the Monte Carlo simulation results. It is Fig.3 plots the system throughput versus SNR in delay-
observed that the outage performance of FD NOMA exceeds limited transmission mode without direct link. The solid curves
HD NOMA and OMA on the condition of low SNR region. represent throughput for FD/HD NOMA without direct link
This is because LI is not dominant impact factor in the low which are obtained from (23) and (24), respectively. We can
SNR region for FD NOMA and answers the first question observe that FD NOMA achieves a higher throughput than
we raised in the introduction part. Especially, one can also HD NOMA and OMA, since FD NOMA has the low values
observed that the outage behavior of D2 for HD NOMA of LI. It is worth noting that increasing the values of LI from
outperforms HD OMA [32, eq. (8)]. The asymptotic outage −20 dB to −10 dB reduce the system throughput in high SNR
probability curves of two users for HD NOMA are plotted region. This is because FD NOMA converges to an error floor
according to (20) and (22), respectively. The asymptotic curves in high SNR region.

Authorized licensed use limited to: ULAKBIM UASL - Yakin Dogu Universitesi. Downloaded on July 17,2024 at 09:09:22 UTC from IEEE Xplore. Restrictions apply.
YUE et al.: EXPLOITING FULL/HALF-DUPLEX USER RELAYING IN NOMA SYSTEMS 569

Fig. 5. Outage probability versus the transmit SNR with direct link.
Fig. 4. Rates versus the transmit SNR without direct link.

2) Ergodic Rate: Fig. 4 plots the ergodic sum rate of FD/HD

NOMA without direct link versus SNR and the value of LI is
assumed to be E{|h L I |2 } = −10 dB. The red and blue solid
curves denote the achievable rates of D1 for FD/HD NOMA,
respectively. The dashdotted curves denote the achievable rates
of D2 for FD/HD NOMA, respectively. One can observe that
the achievable rate of D1 for FD NOMA is superior to HD
NOMA in the low SNR region. This phenomenon can be
also explained is that LI has little effect on achievable rate
of D1 in the low SNR region. On the contrary, due to the
influence of LI, the ergodic rate of D1 converge to a throughput
ceiling in the high SNR region. Another observation is that
the achievable rate of D2 for FD NOMA exceeds the HD
NOMA. This is due to the fact that the communication process
is completed over one slot time for FD NOMA. It is also
shown that throughput ceilings exist in FD/HD NOMA for D2 , Fig. 6. Outage probability versus the transmit SNR for different values of
which verify the conclusion in Remark 5. The dashed curves LI with direct link.
denote the asymptotic ergodic sum rate for FD/HD NOMA, diversity order by using the direct link, which overcomes the
corresponding to the analytical results derived in (40) and (41), problem of zero diversity order inherent to FD cooperative
respectively. An important observation is that FD NOMA can systems. This phenomenon answers the second question we
achieve the maximal ergodic sum rate corresponding to HD raised in the introduction part. More importantly, one can
NOMA and OMA in the low SNR region. The reason is that observe that the performance of FD NOMA is superior to
FD NOMA can improve system spectrum efficiency in the low HD NOMA in the low SNR region, whilst the performance is
SNR region. This phenomenon answers the third question we inferior to HD NOMA in the high SNR region. Additionally,
raised in the introduction part. for D2 , considering the impact of RI between relaying link and
direct link, we plots the corresponding outage probability of
B. With Direct Link D2 based on (25) denoted by blue dash-dotted curves. As can
For user relaying with direct link, the target rate is set to be seen from Fig. 5, these simulation results almost match with
be R1 = 2, R2 = 1 BPCU for D1 and D2 , respectively. analytical result derived in (27) by utilizing the upper bound
The performance of conventional HD NOMA is shown as a SINR in low SNR region. However, with the increase of RI
benchmark for comparison, which has been discussed in [25]. levels κ, the RI-based simulation results for D2 converge to a
1) Outage Probability: Fig. 5 plots the outage probability constant and provide an error floor in high SNR region. Hence,
of two users versus SNR and the value of LI is assumed to be the effect of RI should be carefully addressed in practical
E{|h L I |2 } = −15 dB. The exact outage probability curves of FD NOMA systems. Another observation is that the outage
two users for FD NOMA are given by Monte Carlo simulations behavior of D2 for FD/HD NOMA outperforms HD OMA
and perfectly match with the analytical results derived in (10) [39, eq. (13)]. That is due to the fact that NOMA can provide
and (27). The approximated outage probability curve for D2 more spectral efficiency compared to OMA.
is plotted according to (28) is practically indistinguishable Fig. 6 plots the outage probability of the two users versus
from the exact expression. We observe that D2 obtains one different values of LI from −20 dB to −10 dB. We see that LI

Authorized licensed use limited to: ULAKBIM UASL - Yakin Dogu Universitesi. Downloaded on July 17,2024 at 09:09:22 UTC from IEEE Xplore. Restrictions apply.
570 IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 66, NO. 2, FEBRUARY 2018

Fig. 7. System throughput in delay-limited transmission mode versus SNR Fig. 9. System energy efficiency in delay-limited transmission mode, where
with different LI with direct link. Ps = Pr =10 W, and T = 1.

for FD/HD NOMA in the low SNR region. This can be

explained as the direct link between BS and D2 exists and
improves system reliability. One can observed from figure,
as the RI value increases, the achievable rate of D2 becomes
smaller, such as, setting κ from 0.5 to 1. In addition to consider
the effect of LI, it is also important to design effective rake
receiver at D2 for FD NOMA system.

C. Energy Efficiency
Fig. 9 plots the system energy efficiency versus SNR in
delay-limited transmission mode for user relaying in NOMA
systems. The dashed curves, representing user relaying without
direct link for FD/HD NOMA are obtained from (53), (23)
and (54), (24) with throughput in delay-limited transmission
mode, respectively. The solid curves, representing representing
user relaying with direct link for FD/HD NOMA are obtained
Fig. 8. Rates versus the transmit SNR with direct link. from (53), (29) and (54), (30) with throughput in transmission
mode, respectively. It can be seen that the energy efficiency of
user relaying for FD/HD NOMA in delay-limited transmission
strongly affect the performance of FD NOMA systems. With mode is FD > HD in the low SNR region. The reason is that
the values of LI increasing, the superiority of FD NOMA is FD NOMA can achieve larger throughput than that of HD
no longer apparent. Therefore, it is important to consider the NOMA at this transmission mode. This phenomenon answers
influence of LI when designing practical FD NOMA systems. the fourth question we raised in the introduction part.
Fig. 7 plots system throughput versus SNR in delay-limited Fig. 10 plots the system energy efficiency versus SNR in
transmission mode with direct link. The solid curves, repre- delay-tolerant transmission mode for user relaying in NOMA
senting FD NOMA, is obtained from (29). The dashed curve, systems. The dashed curves, representing user relaying without
representing HD NOMA, is obtained from (30). Observe that direct link for FD/HD NOMA are obtained from (40), (53) and
FD NOMA also outperform HD NOMA in the low SNR (41), (54) with throughput in delay-tolerant mode, respectively.
region. The reason is that in low SNR region, the outage The solid curves, representing user relaying with direct link
probability is small and has no effect on the throughput, which for FD/HD NOMA are obtained from (48), (53) and (48),
only depends on the fixed transmission rates at the BS. (54) with throughput in delay-tolerant mode, respectively.
2) Ergodic Rate: Fig. 8 plots the ergodic sum rate of HD/FD We observe that user relaying with direct link has a higher
NOMA with direct link versus SNR and the value of LI is energy efficiency compared to without direct link for FD/HD
assumed to be E{|h L I |2 } = −10 dB. The dashed curves denote NOMA in the low SNR region. This is because that the direct
the asymptomatic ergodic sum rate for FD/HD NOMA based link improves system throughput at this transmission mode.
on the analytical results derived in (48) and (49), respectively. Additionally, it is worth noting that HD NOMA achieves the
It is observed that the asymptomatic ergodic sum rate is larger higher system energy efficiency in the high SNR region. This is

Authorized licensed use limited to: ULAKBIM UASL - Yakin Dogu Universitesi. Downloaded on July 17,2024 at 09:09:22 UTC from IEEE Xplore. Restrictions apply.
YUE et al.: EXPLOITING FULL/HALF-DUPLEX USER RELAYING IN NOMA SYSTEMS 571

Furthermore, substituting (2), (6) and (8) to (A.1), J11 and

J12 can be calculated as
 
γtFh 2D |h 0 |2 a2
J11 = Pr |h 2 | <2
− , |h 0 | < τ1
2
ρ |h 0 |2 a1 ρ + 1
 τ1  FD 
γt h 2 ya2
= F|h 2 |2 − f 2 (y) d y
0 ρ ya1 ρ + 1 |h 0 |
 FD

γth
τ
 τ1 − 1 2 ya2
− ya ρ+1
− 1 1 − y 2 ρ 1
= 1−e 0 − e 0e d y . (A.2)
0
 0
 !
1

Based on (A.2), using x = yρa1 +1, 1 can be calculated as

γ FD  τ1 ρa1 +1 a2 (x−1)
1 − ρ
th 2
− ρax−1
1 = e 2 e 
1 0 e a1 ρ2 x d x
Fig. 10. System energy efficiency in delay-tolerant transmission mode, where
0 1
∞  τ1 ρa1 +1
Ps = Pr =10 W, and T = 1. eϕ (−1)n −a x
= xne 1 ρ2 x dx, (A.3)
φ2 n!(0 ρa1 )n
n=0 1  !
due to the fact that HD NOMA can provide a larger system 2
throughput, while FD NOMA converges to the throughput γthF D
ceiling in the high SNR region. where ϕ = ρa110 − ρ22 − φ1 , φ1 = a1−a
ρ2 and φ2 = a1 ρ0 .
2

Note that (A.3) is obtained by using Binomial theorem.

VII. C ONCLUSION Furthermore, using z = 1x , 2 is given by
This paper has investigated FD/HD user relaying in coop-  1 1 −a
a2 z
erative NOMA system and two cooperative relaying scenarios 2 = e 1 ρ2 dz
1
τ1 ρa1 +1
z n+2
have been considered insightfully. The performance of FD/HD
n+1
user relaying for NOMA system was characterized. The (−1)2n+1 φ1
closed-form expressions of outage probability for two users = (Ei (ψ) −Ei (φ1 ))
(n + 1)!
have been derived. Due to the influence of LI, the diversity
(1 + a1 ρτ1 )n+1 eψ ψ k − eφ φ1k
n
orders achieved by two user were zeros for FD NOMA. + , (A.4)
Therefore, the direct link between BS and far user was utilized (n + 1) n · · · (n + 1 − k)
k=0
to convey information and one diversity order was obtained
for the far user. Based on the analytical results, it was shown where (A.4) can be obtained by using [38, eq. (3.351.4)].
that FD NOMA was superior to HD NOMA in low SNR Substituting (A.3) into (A.2), J11 is written as
region rather than in the high SNR region. The superior of ∞
− 1
τ
(−1)n eϕ
(−1)2n+1 φ1 n+1
FD NOMA was not apparent with the values of LI increasing. J11 = 1−e 0 − (Ei (ψ)
Furthermore, the expressions of ergodic sum rate for FD/HD (n + 1)!
n!φ2 n+1
n=0
user relaying were derived.The results showed that FD NOMA n

(1 + a1 ρτ1 )n+1 eψ ψ k − eφ1 φ1 k
achieved a higher sum rate than HD NOMA in the low − Ei (φ1 )) + .
SNR region. In addition, the system energy efficiencies for (n + 1) n · · · (n + 1 − k)
k=0
FD/HD user relaying were discussed in different transmission (A.5)
modes.
After some algebraic manipulations, J12 is calculated as
 ∞  ∞
A PPENDIX A 1 − y 1 − x
P ROOF OF T HEOREM 3 J12 = e LI e 1 dxdy
0 LI τ1 (yρ+1) 1
τ
Based on (26), the outage probability of D2 can be − 1
= χe 1 , (A.6)
expressed as
    where χ = 1 +τ11ρLI .
PDF2D,dir = Pr γ DM2 RC < γtFh 2D Pr γ D2 →D1 > γtFh 2D
  !  ! Similarly, J13 is given by
J11 J12  τ  τ 
− 1 − 1
  J13 = 1 − e 0 1 − χe 1 . (A.7)
+ Pr γ1,D2 < γtFh 2D , γ D2 →D1 < γtFh 2D . (A.1)
  ! Combining (A.5), (A.6) and (A.7), we can obtain (27).
J13 The proof is completed.

Authorized licensed use limited to: ULAKBIM UASL - Yakin Dogu Universitesi. Downloaded on July 17,2024 at 09:09:22 UTC from IEEE Xplore. Restrictions apply.
572 IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 66, NO. 2, FEBRUARY 2018

A PPENDIX B where  = 1. We focus on the high SNR approximation of

P ROOF OF T HEOREM 4 J1 , which is given by
 
To obtain (31), the ergodic rate of D1 for FD NOMA is |h 1 |2 a2
expressed as J1 ≈ min , |h 2 | ρ .
2
(C.2)
|h 1 |2 a1 + |h L I |2
⎡ ⎛ ⎞⎤   !
Y
⎢ ⎜ |h 1 |2 a1 ρ ⎟ ⎥
⎢ ⎜ ⎟⎥
FD
RD = E ⎢log ⎜1 + ⎟ ⎥ The CDF of Y is expressed as
1
⎣ ⎝  |h L I |2 ρ + 1 ⎠⎦  
  !   |h 1 |2 a2
 ∞ X FY (y) = 1 − Pr |h 2 |2 ρ > y Pr >y .
1 1 − FX (x)   ! |h 1 |2 a1 + |h L I |2
= dx (B.1) 
  !
ln 2 0 1+x 1
2
where  = 1. (C.3)
The CDF of X is calculated as follows 1 and 2 are given by
    
x |h L I |2 ρ + 1 a2 
FX (x) = Pr |h 1 | <
2 − y
1 = U − y 1 − e ρ2 , (C.4)
a1 ρ a1
 ∞  x(zρ+1)
1 − z a1 ρ 1 − y and
= e LI e 1 d ydz  
0 L I 0 1 a2 (a2 − ya1 ) 1
a1 1 2 = U −y , (C.5)
= 1−
− x
e a1 ρ1 . (B.2) a1 (a2 − ya1 ) 1 + y L I
x L I + a1 1
respectively, where U (x) is unit step function as
Substituting (B.2) into (B.1), the ergodic rate of D1 is 1, x > 0
written as U (x) = .
0, x < 0
 ∞
FD 1 1 a1 1 − x Substituting (C.4) and (C.5) into (C.3), the CDF of Y is
R D1 = e a1 ρ1 d x given by
ln 2 0 1 + x a1 1 + x L I
 ∞ − x   − ρx
1 −a1 1 e a1 ρ1 a2 e 2 (a2 − xa1 ) 1
= dx FY (y) = 1 − U −y . (C.6)
ln 2 0 (1 + x) ( L I − a1 1 ) a1 (a2 − xa1 ) 1 + x L I
  !
J1 Base on (C.6), a high SNR approximation of the ergodic
 ∞ − x
rate for D2 is written as
1 a1 1  L I e a1 ρ1
+ dx . (B.3)  a2
ln 2 (a1 1 +x L I ) ( L I −a11 ) − y
e ρ2 (a2 − ya1) 1
0  ! F D,∞ 1 a1 1
R D2 ,nodir = dy
J2 ln 2 0 1 + y (a2 − ya1) 1 + y L I
⎛
Based on [38, eq. (3.352.4)] and applying some polynomial
 a2
1 ⎜
− y
expansion manipulations, J1 and J2 are given by ⎜ a1 1 a2 1 e ρ2
= ⎜ dy
  ln 2 ⎝ 0 1 + y yξ + a2 1
a1 1 e a1 ρ1
1
−1   !
J1 = Ei , (B.4) J2
⎞
 L I − a1 1 a1 ρ1
 a2
⎟
− ρy
and a1 1 ya11 e ⎟ 2
− d y ⎟, (C.7)
1   0 1 + y yξ + a2 1 ⎠
a1 1 e ρ L I −1   !
J2 = E . (B.5)
 L I − a1 1 i ρ L I J3

where ξ = ( L I − a1 1 ).
Substituting (B.4) and (B.5) into (B.3), we can obtain (31).
Applying [38, eq. (3.352.1)] and some polynomial expan-
The proof is completed.
sion manipulations, J2 and J3 can be calculated as
  a2 − y  a2 − y

A PPENDIX C a2 1 a1 e ρ2 a1 ξ e ρ2
J2 = dy − dy
P ROOF OF T HEOREM 5 a2 1 − ξ 0 1+y 0 ξ y + a2 1
&     
The proof starts by providing the ergodic rate of D2 as a2 1 1 −1 −1
follows: = e ρ2
Ei − Ei
a2 1 − ξ ρa1 2 ρ2
⎡ ⎛ ⎞⎤     '

a2  1 a2 ξ + a1 a2 1 a2 1
⎢ ⎜ ⎟⎥ −e ρ 2 ξ Ei − −Ei − .
RDFD
2 ,nodir
= E ⎣log ⎝1 + min γ D2 →D1 , γ2,D2 ⎠⎦, (C.1) ρa1 ξ 2 ρ2 ξ
  !
J1 (C.8)

Authorized licensed use limited to: ULAKBIM UASL - Yakin Dogu Universitesi. Downloaded on July 17,2024 at 09:09:22 UTC from IEEE Xplore. Restrictions apply.
YUE et al.: EXPLOITING FULL/HALF-DUPLEX USER RELAYING IN NOMA SYSTEMS 573

 a2 − y  a2 − y 
a1 1 a1 e ρ2 a1 a2 1 e ρ2 and
J3 = dy − dy  
ξ − a2 1 0 1+y 0 ξ y + a2 1 a2 (a2 − ya1 ) 1
⎧ 2 = U −y , (E.5)
     a1 (a2 − ya1 ) 1 + y L I
a1 1 ⎨
1 −1 −1
= e ρ2 Ei − Ei respectively.
ξ − a2 1 ⎩ ρa1 2 ρ2 Substituting (E.4) and (E.5) into (E.3), the CDF of Y is
⎫ given by
a2  1     ⎬  
a2 1 e ρ2 ξ
a2 ξ +a1 a2 1 a2 1 a2 (a2 − ya1) 1
− Ei − − Ei − . FY (y) = 1 − U −y . (E.6)
ξ ρa1 ξ 2 ρ2 ξ ⎭ a1 (a2 − ya1) 1 + y L I
(C.9) Base on (E.6), a high SNR approximation of the ergodic
rate for D2 is written as
Substituting (C.8) and (C.9) into (C.7), we can obtain (34).
 a2
The proof is completed. F D,∞ 1 a1 1 (a2 − ya1) 1
R D2 ,dir = dy
A PPENDIX D ln 2 0 1 + y (a2 − ya1) 1 + y L I
⎛
P ROOF OF C OROLLARY 4
 a2
We can rewrite (35) as follows: 1 ⎜ ⎜ a1 a2 1
⎡ ⎛ ⎞⎤ = ⎜ dy
ln 2 ⎝ 0 (1 + y) (yξ + a2 1 )
1 ⎢ ⎜  ⎟⎥   !
HD
RD 2 ,dir
= E ⎣log ⎝1 + min γ D2 →D1 , γ2,D2 ⎠⎦, (D.1) J2
2   ! ⎞
J1
 a2
⎟
where  = 0.
a1 ya1 1 ⎟
− dy⎟, (E.7)
At the high SNR region, J1 is approximated as (1 + y) (yξ + a2 1 ) ⎠
0  !
  J3
a2
J1 ≈ min , |h 2 | ρ .
2
(D.2)
a1 where ξ = ( L I − a1 1 ).
  !
Y
After some algebraic manipulations, J2 and J3 are obtained
    as follows:
− y     
Therefore, FY (y) = 1 − e ρ2 U aa21 − y can be easily a2 1 a2 ξ
obtained. As such, the approximated ergodic rate of D2 for J2 = ln 1 + − ln 1 + . (E.8)
a2 1 − ξ a1 a1 1
HD NOMA at high SNR is given in (36).     
a1 1 a2 1 ξ a2
J3 = ln 1+ −ln 1+ . (E.9)
A PPENDIX E a2 1 −ξ ξ a1 1 a1
P ROOF OF T HEOREM 6 Substituting (E.8) and (E.9) into (E.7), we can obtain (45).
The proof starts by providing the ergodic rate of D2 as The proof is completed.
follows: ⎡ ⎛ ⎞⎤
 ⎟⎥ R EFERENCES
⎢ ⎜
RDFD
= E ⎢log ⎜1 + min γ D →D , γ M RC ⎟⎥, (E.1)
2 ,dir ⎣ ⎝ 2 1 D2 ⎠⎦ [1] X. Yue, Y. Liu, S. Kang, A. Nallanathan, and Z. Ding, “Outage
  ! performance of full/half-duplex user relaying in NOMA systems,” in
J1 Proc. IEEE Int. Commun. Conf. (ICC), Paris, France, May 2017,
pp. 1–6.
where  = 1. We focus on the high SNR approximation of [2] Y. Cai, Z. Qin, F. Cui, G. Y. Li, and J. A. McCann, “Modulation and
J1 , which is given by multiple access for 5G networks,” CoRR, Feb. 2017. [Online]. Available:
  https://arxiv.org/abs/1702.07673
|h 1 |2 a2 a2 [3] Proposed Solutions for New Radio Access, Mobile and Wireless Commu-
J1 ≈ min , |h 2 | ρ +
2
. (E.2) nications Enablers for the Twenty-twenty Information Society (METIS)
|h 1 |2 a1 + |h L I |2 a1 Deliverable D.2.4, Feb. 2015.
  ! [4] Y. Saito, Y. Kishiyama, A. Benjebbour, T. Nakamura, A. Li, and
Y K. Higuchi, “Non-orthogonal multiple access (NOMA) for cellular
The cumulative distribution function (CDF) of Y is expressed future radio access,” in Proc. IEEE Veh. Technol. Conf. (VTC-Spring),
Dresden, Germany, Jun. 2013, pp. 1–5.
as
    [5] H. Nikopour and H. Baligh, “Sparse code multiple access,” in Proc.
a2 |h 1 |2 a2 IEEE Annu. Int. Symp. Pers., Indoor, Mobile Radio Commun. (PIMRC),
FY (y)= 1−Pr |h 2 | ρ + ≥ y ×Pr
2
≥y . London, U.K., Sep. 2013, pp. 332–336.
a1 |h 1 |2 a1 +|h L I |2 [6] S. Chen, B. Ren, Q. Gao, S. Kang, S. Sun, and K. Niu, “Pattern
  !   ! division multiple access—A novel nonorthogonal multiple access for
1 2 fifth-generation radio networks,” IEEE Trans. Veh. Technol., vol. 66,
(E.3) no. 4, pp. 3185–3196, Apr. 2017.
[7] Z. Yuan, G. Yu, W. Li, Y. Yuan, X. Wang, and J. Xu, “Multi-user
1 and 2 are given by shared access for Internet of Things,” in Proc. IEEE Veh. Technol.
    Conf. (VTC Spring), Nanjing, China, May 2016, pp. 1–5.
a2 − ρ1 y− aa2 [8] Study on Downlink Multiuser Superposition Transmation for LTE,
1 = 1 − U y − e 2 1 , (E.4) document TR 36.859, 3rd Generation Partnership Projet (3GPP),
a1 Mar. 2015.

Authorized licensed use limited to: ULAKBIM UASL - Yakin Dogu Universitesi. Downloaded on July 17,2024 at 09:09:22 UTC from IEEE Xplore. Restrictions apply.
574 IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 66, NO. 2, FEBRUARY 2018

[9] Z. Ding et al., “Application of non-orthogonal multiple access in LTE [31] D. P. M. Osorio, E. E. B. Olivo, H. Alves, J. C. S. S. Filho, and
and 5G networks,” IEEE Commun. Mag., vol. 55, no. 2, pp. 185–191, M. Latva-Aho, “Exploiting the direct link in full-duplex amplify-and-
Feb. 2017. forward relaying networks,” IEEE Signal Process. Lett., vol. 22, no. 10,
[10] M. Al-Imari, P. Xiao, M. A. Imran, and R. Tafazolli, “Uplink pp. 1766–1770, Oct. 2015.
non-orthogonal multiple access for 5G wireless networks,” in Proc. [32] T. Kwon, S. Lim, S. Choi, and D. Hong, “Optimal duplex mode for
IEEE Int. Symp. Wireless Commun. Syst. (ISWCS), Barcelona, Spain, DF relay in terms of the outage probability,” IEEE Trans. Veh. Technol.,
Aug. 2014, pp. 781–785. vol. 59, no. 7, pp. 3628–3634, Sep. 2010.
[11] N. Zhang, J. Wang, G. Kang, and Y. Liu, “Uplink nonorthogonal multiple [33] T. Riihonen, S. Werner, and R. Wichman, “Hybrid full-duplex/half-
access in 5G systems,” IEEE Commun. Lett., vol. 20, no. 3, pp. 458–461, duplex relaying with transmit power adaptation,” IEEE Trans. Wireless
Mar. 2016. Commun., vol. 10, no. 9, pp. 3074–3085, Sep. 2011.
[12] Z. Ding, Z. Yang, P. Fan, and H. V. Poor, “On the performance of [34] Z. Zhang, Z. Ma, M. Xiao, Z. Ding, and P. Fan, “Full-duplex device-to-
non-orthogonal multiple access in 5G systems with randomly deployed device-aided cooperative nonorthogonal multiple access,” IEEE Trans.
users,” IEEE Signal Process. Lett., vol. 21, no. 12, pp. 1501–1505, Veh. Technol., vol. 66, no. 5, pp. 4467–4471, May 2017.
Dec. 2014. [35] I. Krikidis, H. A. Suraweera, P. J. Smith, and C. Yuen, “Full-duplex relay
[13] Z. Ding, P. Fan, and H. V. Poor, “Impact of user pairing on 5G selection for amplify-and-forward cooperative networks,” IEEE Trans.
nonorthogonal multiple-access downlink transmissions,” IEEE Trans. Wireless Commun., vol. 11, no. 12, pp. 4381–4393, Dec. 2012.
Veh. Technol., vol. 65, no. 8, pp. 6010–6023, Aug. 2016. [36] T. M. Cover and J. A. Thomas, Elements of Information Theory, 6th ed.
New York, NY, USA: Wiley, 1991.
[14] Y. Liu, Z. Ding, M. Elkashlan, and J. Yuan, “Nonorthogonal multiple
[37] C. Zhong, H. A. Suraweera, G. Zheng, I. Krikidis, and Z. Zhang,
access in large-scale underlay cognitive radio networks,” IEEE Trans.
“Wireless information and power transfer with full duplex relaying,”
Veh. Technol., vol. 65, no. 12, pp. 10152–10157, Dec. 2016.
IEEE Trans. Commun., vol. 62, no. 10, pp. 3447–3461, Oct. 2014.
[15] Y. Tian, S. Lu, A. Nix, and M. Beach, “A novel opportunistic NOMA [38] I. S. Gradshteyn and I. M. Ryzhik, Table of Integrals, Series, and
in downlink coordinated multi-point networks,” in Proc. IEEE Veh. Products, 6th ed. New York, NY, USA: Academic, 2000.
Technol. Conf. (VTC Fall), Boston, MA, USA, Sep. 2015, pp. 1–5. [39] Z. Bai, J. Jia, C. X. Wang, and D. Yuan, “Performance analysis
[16] Z. Yang, Z. Ding, P. Fan, and Z. Ma, “Outage performance for dynamic of SNR-based incremental hybrid decode-amplify-forward cooperative
power allocation in hybrid non-orthogonal multiple access systems,” relaying protocol,” IEEE Trans. Commun., vol. 63, no. 6, pp. 2094–2106,
IEEE Commun. Lett., vol. 20, no. 8, pp. 1695–1698, Aug. 2016. Jun. 2015.
[17] P. Botsinis et al., “Quantum-aided multi-user transmission in
non-orthogonal multiple access systems,” IEEE Access, vol. 4,
pp. 7402–7424, 2016.
[18] J. Choi, “On the power allocation for a practical multiuser superposition Xinwei Yue received the master’s degree from
scheme in NOMA systems,” IEEE Commun. Lett., vol. 20, no. 3, Henan Normal University, China, in 2013. He is
pp. 438–441, Mar. 2016. currently pursuing the Ph.D. degree with the School
[19] Y. Liu, Z. Qin, M. Elkashlan, Y. Gao, and L. Hanzo, “Enhancing of Electronic and Information Engineering, Beihang
the physical layer security of non-orthogonal multiple access in large- University.
scale networks,” IEEE Trans. Wireless Commun., vol. 16, no. 3, His research interests include wireless communi-
pp. 1656–1672, Mar. 2017. cations theory, non-orthogonal multiple access, and
[20] J. N. Laneman, D. N. C. Tse, and G. W. Wornell, “Cooperative cooperative networks.
diversity in wireless networks: Efficient protocols and outage behav-
ior,” IEEE Trans. Inf. Theory, vol. 50, no. 12, pp. 3062–3080,
Dec. 2004.
[21] J. Choi, “Non-orthogonal multiple access in downlink coordinated
two-point systems,” IEEE Commun. Lett., vol. 18, no. 2, pp. 313–316, Yuanwei Liu (S’13–M’16) received the B.S. and
Feb. 2014. M.S. degrees from the Beijing University of Posts
[22] J. B. Kim and I. H. Lee, “Non-orthogonal multiple access in coordinated and Telecommunications in 2011 and 2014, respec-
direct and relay transmission,” IEEE Commun. Lett., vol. 19, no. 11, tively, and the Ph.D. degree in electrical engineering
pp. 2037–2040, Nov. 2015. from the Queen Mary University of London, U.K.,
[23] J. B. Kim and I. H. Lee, “Capacity analysis of cooperative relaying in 2016. He is currently a Post-Doctoral Research
systems using non-orthogonal multiple access,” IEEE Commun. Lett., Fellow with the Department of Informatics, King’s
vol. 19, no. 11, pp. 1949–1952, Nov. 2015. College London, U.K.
[24] J. Men, J. Ge, and C. Zhang, “Performance analysis of nonorthog- His research interests include 5G wireless net-
onal multiple access for relaying networks over Nakagami-m fading works, Internet of Things, stochastic geometry and
channels,” IEEE Trans. Veh. Technol., vol. 66, no. 2, pp. 1200–1208, matching theory. He has served as a TPC member
Feb. 2017. for many IEEE conferences such as GLOBECOM and VTC. He received the
[25] Z. Ding, M. Peng, and H. V. Poor, “Cooperative non-orthogonal mul- Exemplary Reviewer Certificate of the IEEE W IRELESS C OMMUNICATION
tiple access in 5G systems,” IEEE Commun. Lett., vol. 19, no. 8, L ETTER in 2015 and the IEEE T RANSACTIONS ON C OMMUNICATIONS in
pp. 1462–1465, Aug. 2015. 2017. He currently serves as an Editor of the IEEE C OMMUNICATIONS
[26] Y. Liu, Z. Ding, M. Elkashlan, and H. V. Poor, “Cooperative L ETTERS and the IEEE A CCESS .
non-orthogonal multiple access with simultaneous wireless information
and power transfer,” IEEE J. Sel. Areas Commun., vol. 34, no. 4,
pp. 938–953, Apr. 2016. Shaoli Kang was a Project Manager with the
[27] H. Ju, E. Oh, and D. Hong, “Improving efficiency of resource usage China Academy of Telecommunications Technol-
in two-hop full duplex relay systems based on resource sharing and ogy (CATT), with a focus on research and develop-
interference cancellation,” IEEE Trans. Wireless Commun., vol. 8, no. 8, ment of TDSCDMA, from 2000 to 2005. She was
pp. 3933–3938, Aug. 2009. with the Communication Center of System Research,
[28] T. Riihonen, S. Werner, and R. Wichman, “Mitigation of loopback self- University of Surrey, as a Research Fellow, where
interference in full-duplex MIMO relays,” IEEE Trans. Signal Process., she was involved in research on projects from
vol. 59, no. 12, pp. 5983–5993, Dec. 2011. EPSRC and OFCOM and leading the Antenna and
[29] Z. Zhang, X. Chai, K. Long, A. V. Vasilakos, and L. Hanzo, “Full duplex Propagation Club. In 2007, she came back to CATT
techniques for 5G networks: Self-interference cancellation, protocol and acted as the Vice Chief Engineer of TDD
design, and relay selection,” IEEE Commun. Mag., vol. 53, no. 5, Research and Development product line, with a
pp. 128–137, May 2015. focus on speeding up the standard and industrial progress of TDD technology.
[30] Q. Wang, Y. Dong, X. Xu, and X. Tao, “Outage probability of Since 2011, she has been with the Wireless Innovation Center and acted as
full-duplex AF relaying with processing delay and residual self- the Head Expert, leading the 5G research in CATT. She is currently the Head
interference,” IEEE Commun. Lett., vol. 19, no. 5, pp. 783–786, Expert of 5G standardization with the Wireless Innovation Center. She has
May 2015. applied over 50 patents and authored over 20 papers.

Authorized licensed use limited to: ULAKBIM UASL - Yakin Dogu Universitesi. Downloaded on July 17,2024 at 09:09:22 UTC from IEEE Xplore. Restrictions apply.
YUE et al.: EXPLOITING FULL/HALF-DUPLEX USER RELAYING IN NOMA SYSTEMS 575

Arumugam Nallanathan (S’97–M’00–SM’05– Zhiguo Ding (S’03–M’05) received the B.Eng.

F’17) was an Assistant Professor with the Depart- degree in electrical engineering from the Beijing
ment of Electrical and Computer Engineering, University of Posts and Telecommunications in
National University of Singapore, from 2000 to 2000, and the Ph.D. degree in electrical engineering
2007. He was with the Department of Informatics, from the Imperial College London in 2005. From
King’s College London, from 2007 to 2017, where 2005 to 2014, he was with Queen’s University
he was a Professor of wireless communications from Belfast, the Imperial College London, and Newcastle
2013 to 2017. He has been a Professor of wire- University. Since 2014, he has been with Lancaster
less communications with the School of Electronic University as a Chair Professor. Since 2012, he has
Engineering and Computer Science, Queen Mary also been an Academic Visitor with Princeton Uni-
University of London, since 2017. He has authored versity. His research interests are 5G networks, game
over 350 technical papers in scientific journals and international conferences. theory, cooperative and energy harvesting networks, and statistical signal
His research interests include 5G wireless networks, Internet of Things, processing.
and molecular communications. He is an IEEE Distinguished Lecturer. He Dr. Ding received the Best Paper Award at the IET Communications Confer-
received the IEEE Communications Society SPCE Outstanding Service Award ence on Wireless, Mobile and Computing in 2009, the IEEE Communication
in 2012 and the IEEE Communications Society RCC Outstanding Service Letter Exemplary Reviewer in 2012, and the EU Marie Curie Fellowship
Award in 2014. He was a co-recipient of the Best Paper Award presented 2012–2014. He is an Editor of the IEEE Transactions on Communications,
at the IEEE International Conference on Communications in 2016 and the the IEEE T RANSACTIONS ON V EHICULAR T ECHNOLOGY, and the Journal
IEEE International Conference on Ultra-Wideband in 2007. He was selected of Wireless Communications and Mobile Computing. He was an Editor of the
as a Web of Science (ISI) Highly Cited Researcher in 2016. He served as the IEEE W IRELESS C OMMUNICATION L ETTERS and the IEEE C OMMUNICA -
Chair for the Signal Processing and Communication Electronics Technical TION L ETTERS from 2013 to 2016.
Committee of the IEEE Communications Society and the Technical Program
Chair and a member of technical program committees in numerous IEEE
conferences. He was an Editor of the IEEE T RANSACTIONS ON W IRELESS
C OMMUNICATIONS from 2006 to 2011, the IEEE W IRELESS C OMMUNICA -
TIONS L ETTERS , and the IEEE S IGNAL P ROCESSING L ETTERS . He is an
Editor of the IEEE T RANSACTIONS ON C OMMUNICATIONS and the IEEE
T RANSACTIONS ON V EHICULAR T ECHNOLOGY.

Authorized licensed use limited to: ULAKBIM UASL - Yakin Dogu Universitesi. Downloaded on July 17,2024 at 09:09:22 UTC from IEEE Xplore. Restrictions apply.