Please cite the Published Version

Zhang, Di, Liu, Yuanwei, Dai, Linglong, Bashir, Ali Kashif, Nallanathan, Arumugam and Shim,
Byonghyo (2019) Performance Analysis of FD-NOMA-based Decentralized V2X Systems. IEEE
Transactions on Communications, 67 (7). pp. 5024-5036. ISSN 0090-6778

DOI: https://doi.org/10.1109/tcomm.2019.2904499
Publisher: Institute of Electrical and Electronics Engineers (IEEE)
Version: Accepted Version
Downloaded from: https://e-space.mmu.ac.uk/622899/
Additional Information: "(c) 2019 IEEE. Personal use of this material is permitted. Permission
from IEEE must be obtained for all other uses, including reprinting/ republishing this material for
advertising or promotional purposes, creating new collective works for resale or redistribution to
servers or lists, or reuse of any copyrighted components of this work in other works."

Enquiries:
If you have questions about this document, contact [email protected]. Please include the
URL of the record in e-space. If you believe that your, or a third party’s rights have
been compromised through this document please see our Take Down policy (available from
https://www.mmu.ac.uk/library/using-the-library/policies-and-guidelines)
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information:
DOI 10.1109/TCOMM.2019.2904499, IEEE Transactions on Communications
1

Performance Analysis of FD-NOMA-based

Decentralized V2X Systems
Di Zhang, Member, IEEE, Yuanwei Liu, Member, IEEE, Linglong Dai, Senior Member, IEEE,
Ali Kashif Bashir, Senior Member, IEEE, Arumugam Nallanathan, Fellow, IEEE, and Byonghyo Shim, Senior
Member, IEEE
Abstract—In order to meet the requirements of massively has received much attention with explosively growing devices
connected devices, different quality of services (QoSs), various
connecting to the wireless networks. With the help of cellular
transmit rates and ultra-reliable and low latency communications
(URLLC) in vehicle to everything (V2X) communications, we introduce network, C-V2X can connect more V2X devices [5], [6]; it can
a full duplex non-orthogonal multiple access (FDNOMA)-based establish the link among vehicles, smart infrastructures and
decentralized V2X system model. We then classify the V2X pedestrians. C-V2X operates in two modes. First, in the direct
communications into two scenarios and give their exact capacity
expressions. To solve the computation complicated problems of the
communications (DC) mode, V2X devices can directly
involved exponential integral functions, we give the approximate communicate with each other. Well-known examples include
closed-form expressions with arbitrary small errors. Numerical results vehicle to vehicle (V2V), vehicle to pedestrian (V2P)
indicate the validness of our derivations. Our analysis has that the communications. Second, in the network-based
accuracy of our approximate expressions is controlled by the division
of in the urban and crowded scenario, and the truncation point T in
communications (NC) mode, cellular base station (BS) is playing
the suburban and remote scenario. Numerical results manifest 1) the dominant role, and the V2X devices communicate with (or
Increasing the number of V2X device, NOMA power and Rician factor with the help of) the cellular, for instance, vehicle to network
value yields better capacity performance. 2) Effect of FD-NOMA is (V2N), vehicle to infrastructure (V2I) communications.
determined by the FD self-interference and the channel noise. 3) FD-
NOMA has better latency performance compared to other schemes. However, the current version of C-V2X (i.e., the long term
evolution V2X (LTEV2X)) cannot fully satisfy the requirements
Index Terms—Vehicle communications, V2X, full duplex,
nonorthogonal multiple access, capacity analysis. of low latency, various quality of services (QoSs) and different
transmit rates [6], [7].
In addition, the existing orthogonal frequency division
I. INTRODUCTION multiple access (OFDMA)-basced LTE-V2X systems need
A. Background orthogonality. Different from the static or non-mobility wireless
There are two distinct regimes in vehicle to everything (V2X) communications, moving vehicle caused Doppler effect is a vital
communications, i.e., the dedicated short-range problem for OFDMA-based LTE-V2X systems [8]. As is known,
communications (DSRC) [1], [2] and the cellular-V2X (C-V2X) [3], carrier frequency offset (CFO) caused by the Doppler effect will
[4]. lead to inter-carrier interference (ICI) to the OFDMbased
DSRC was popular in the past decades. Recently, C-V2X wireless communications [9]. In literature, there have been
various studies to solve the CFO compensation, see, e.g., [9],
Corresponding author: B. Shim {[email protected]}.
The work of D. Zhang and B. Shim is supported by the NRF grant funded by
[10]. However, because the oscillators can never be oscillating
the Korean government (MSIP2014R1A5A1011478), the work of D. Zhang is at the identical frequency, in OFDMA-based wireless
also supported by the Zhengzhou University Startup Foundation (32210907), communications, CFO side-effect always exists even for
the National Natural Science Foundation of China (61801435), the China
Postdoctoral Science Foundation (2018M633733). L. Dai is supported by the
nonmoving circumstance [9].
National Natural Science Foundation of China for Outstanding Young Scholars It is noticed that besides the OFDMA, some fifth generation
(Grant No. 61722109). (5G) technologies can be used to address the issues of low
Di Zhang is with the School of Information Engineering, Zhengzhou latency [11], various QoS and different transmit rates in V2X
University, Zhengzhou, 450001, China, and also with the Information System
Laboratory and Department of Electrical and Computer Engineering, Seoul communications. From the upper layer perspective, the
National University, Seoul, 08826, Korea (e-mail: di [email protected]). software-defined networks (SDNs) with its centralized control
Yuanwei Liu and Arumugam Nallanathan are with the EECS, Queen Mary plan and distributed multiple nodes are more suitable for
University of London, London, E14NS, U.K. (e-mail: {yuanwei.liu, vehicle communications. With the aid of machine learning and
a.nallanathan}@qmul.ac.uk).
Linglong Dai is with the Department of Electronic Engineering, Tsinghua
big data analysis, we can monitor all types of events and
University, Beijing 100084, China (e-mail: [email protected]). Ali Kashif maintain a global network status [12]. From the physical layer
Bashir is with the School of Computing, Mathematics, and Digital Technology, perspective, under the equal frequency resources constraint,
Manchester Metropolitan University, Manchester M15 6BH, U.K. (e-mail:
[email protected] ).
NOMA can accommodate more users comparing to the
Byonghyo Shim is with the Institute of New Media and Communications and orthogonal multiple access (OMA) scheme. Besides, these users
Department of Electrical and Computer Engineering, Seoul National University, can be with different QoS requirements [13], [14]. In addition,
Seoul, 08826, Korea (e-mail: [email protected]).
NOMA is insensitive to CFO effect caused by moving vehicles
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TCOMM.2019.2904499, IEEE Transactions on Communications
2
because of its non-orthogonal frequency. NOMA employs the In literature, various channel models are used for the ergodic
same resource block (RB) for multiple user’s transmission, capacity analysis, for instance, the κ − µ channel model [28],
which can alleviate the spectrum bottleneck of wireless [29] and the η − µ channel model [28]. However, obtaining the
communications [15]–[17]. NOMA can pair users with different closed-form capacity expression in these channel models is
transmit rates for simultaneous transmission [18]–[20]. On the difficult because of the involved infinity series operations.
other hand, while simultaneously transmitting and receiving Authors thus employed some special conditions and methods
information, full duplex (FD) can provide faster speed and to give the closed-form expressions, e.g., µ with positive integer
better spectrum efficiency (SE) performances [13]. Moreover, values [29] and the approximate method [30]. On the other
FD can offer reliable communications [21], which is useful for hand, the difficulty to obtain a closed-form expression with
V2X applications such as navigation and emergency message Rayleigh or Rician channel model lies in the involved
broadcasting. exponential integral functions. In order to solve this problem,
some approximate methods and algorithms have been
proposed, for instance, the Swamee and Ohija method for
B. Related Works and Motivations
exponential integral function [31] and the fast and accurate
Some antecedent works has been done on FD-NOMA. For algorithm for generalized exponential integral function [32].
instance, it was found that FD-NOMA can significantly suppress However, these methods are based on some special conditions
the co-channel interferences and achieve better performance (e.g., [32]), or with low accuracies (e.g., [31]). In this paper, we
gains compared to half duplex NOMA (HDNOMA) and give the approximate closed-form capacity expressions for both
orthogonal multiple access (OMA) [22]. Analysis and simulation Rayleigh and Rician channel models while taming the
results in [23] demonstrated that rate region performance of troublesome exponential integral functions.
FD-NOMA outperforms the one with NOMA. Analysis and In this work, we propose the FD-NOMA-based decentralized
simulation results in [24] indicates that FDNOMA improves the V2X system model, and also provide the capacity analysis to
5G’s system performance compared to HD-NOMA. Based on obtain the approximate closed-form capacity expressions with
the relaying system model, analysis and simulation results in high accuracy. We try to answer the following key questions.
[13] indicated that FD-NOMA outperforms HD-NOMA in terms • Can we use one solution to meet the requirements of V2X
of outage probability and ergodic sum rate in low signal to noise communications? If it is not possible, what about a
ratio (SNR) region, but displays an inferior performance in high combination of FD-NOMA techniques?
SNR region. • If the combination is feasible to satisfy the requirements of
In V2X communications, there are some existing works on V2X communications, what about the capacity and
NOMA-V2X and FD-V2X [25]–[27]. Based on the NOMA, the throughput performance of the V2X systems?
authors in [26] proposed the graph-based practical encoding • Is there any approximate expressions for the capacity
and joint belief propagation (BP) decoding techniques, which expressions with arbitrary small error and low
can achieve any rate pair close to the capacity region. B. Di et computational complexity?
al. in [25] employed NOMA for URLLC communications while
proposing a NOMA-based mixed centralized/distributed C. Contributions
(NOMA-MCD) scheme to reduce the resource collision. In [27],
The main contributions of this work can be summarized as
an optimal blind interference alignment scheme was proposed
follows:
for the coexisting of FD and HD modes. This scheme can
• The FD-NOMA-based decentralized V2X systems can
improve the sum rate performance in the finite SNR regime.
partly offload the cellular network1. Compared to OFDMA,
However, most of these studies on NOMA-V2X and FD-V2X
NOMA is insensitive to Doppler effect caused by moving
communications are based on the NC mode, which is a
vehicles. In addition, FD-NOMA can accommodate more
challenge for connecting massive V2X devices because of the
users with different QoSs and transmit rates for
cellular throughput restriction. Although the authors
simultaneous transmission and reception.
investigated the decentralized NOMA-V2X systems in [25],
• Based on the system model, we derive the exact system
there has been no capacity analysis for such a system. To the
ergodic capacity expressions and their approximate
best of our knowledge, a study investigating the impact of
closedform expressions for both scenarios. These
FDNOMA techniques on V2X systems is rare, which motivates
approximate closed-form expressions are with low
us to develop this treatise.
computational complexity and controllable arbitrary small
errors compared to the existing approximate expressions.

1 Besides C-V2X communications, there are other types of cellular

communications, our work can not offload all the cellular network load
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TCOMM.2019.2904499, IEEE Transactions on Communications
3
Insights from our analysis has 1) the accuracy of our see, e.g., [25]–[27], [34]–[36]. As discussed, not only the
simplified approximate expression in urban and crowded vehicles, V2X aims to connect everything on the road. In order
scenario is controlled by the associated division of (with to cope with this trend, in our FD-NOMA-based decentralized
respect to the exponential integral function E1(x)). 2) The V2X systems, all V2X devices (vehicle, pedestrian, traffic lights,
accuracy of our simplified approximate expression in etc.) are comprehensively included. The massive connected
suburban and remote scenario is controlled by the devices and their various applications are making the V2X
truncation point T (with respect to the exponential communications more complicated. To deal with this
integral function En(x)). • It is observed from our numerical intractable problem, in this work, we classify the V2X
results that: 1) the analytical results coincide with the communications into two scenarios: 1) the urban and crowded
Monte-Carlo based simulation results perfectly, which scenario and 2) the suburban and remote scenario.
demonstrates the validity of our derivations. 2) The system In urban and crowded scenario, Rayleigh fading can be used
capacity increases with the increasing allocated power as the channel model. This is due to the abundant reflection and
value, SNR and Rician factor values. 3) The FD self- refraction links between source and destination [37]. In
interference and the channel noise determine the effect contrast, Rician channel model is suitable for the suburban and
of FD-NOMA. 4) FD-NOMA has better latency performance remote scenario because of the less obstacles, where we can
compared to HD-NOMA and HD-orthogonal multiple always establish a dominant light of sight (LoS) path from
access (HD-OMA) schemes. source to destination [38].

D. Notations and Organization

Notations: In this article, we use upper case boldface letters

to denote matrices (e.g., A), and we use lower case boldface
letters to denote vectors (e.g., a). In addition, we use AT as the
transpose of A, a • b to denote the multiply by position
operation for two vector a and b. On the other hand, A ↔ B
means a transmit-receive pair with A and B transceivers on each
side working on FD mode, A → B the transmission procedure
from A to B, vice versa.
The remainder of the paper is organized as follows. In section
II, the FD-NOMA-based decentralized V2X system model is
proposed. We divide the V2X communications into different
scenarios in this section. We analysis the system capacity of Fig. 1: The M ↔ N FD-NOMA-based decentralized V2X system
different scenarios in section III. The numerical simulations are model. The communications among V2X devices can be
given by section IV, and conclusion is given in section V. accomplished by FD-NOMA working on the DC mode.

II. THE FD-NOMA-BASED DECENTRALIZED V2X SYSTEMS B. Received Signal and Power Allocation Scheme
In the FD-NOMA-based decentralized V2X systems, the
A. System Model channel matrix from M sources to N destinations is
The FD-NOMA-based decentralized V2X system model is
given in Fig. 1. This system is slightly different from the existing
ones in the following respects. A) Different from the existing
studies on FD-NOMA, no relaying systems are used because of
the vehicle’s limited energy. B) V2X devices can directly H
communicate with each other through DC mode without the
cellular’s help, and the required contents are obtained from where hi,j is the channel between source i and destination j.
neighboring V2X caches [33]. This system model thus has In this case, the received signal can be given as
shorter transmission distance and better latency performance √
[33]. The cellular network load can be reduced too. y = H p • x + n, (2)
Once can notice that to simplify the analysis, only V2V and
where p ∈ C
V2I communications are considered in the existing V2X studies,
√ M×1, is the allocated downlink NOMA power
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TCOMM.2019.2904499, IEEE Transactions on Communications
4
matrix, x ∈ CM×1 is the downlink transmit signal and n ∼ transmission going through all fading states, which is also called
CN(0,σ2IN) is the downlink channel noise. Under the condition the Shannon capacity since it is the average of instantaneous
that Hˆ = HT is the uplink channel with FD mode, uplink transmit
capacity over all states. In contrast, outage capacity is used to
describe the system performance under slowly varying
information with FD mode will be
channels with a constant instantaneous γ [41], [49]. Here in this
study, we adopt the ergodic capacity since V2X channels are
yˆ = Hˆ ppˆ • z + nˆ, (3) generally the time varying channels.
where z ∈ CN×1 is the uplink information. NOMA power and In the decentralized FD-NOMA V2X systems, transmission
channel noise vectors thus can be given as pˆ . channels are uncorrelated. In this case, the considered multiple
The total power received by destination n from all M sources is input multiple output NOMA (MIMO-NOMA) can be treated as
given by a sum of additive single input single output NOMA (SISONOMA)
pn = p1,n + p2,n + ... + pM,n. (4) links. Moreover, similar to prior works [13], [50], we adopt an
increasing order of the channel response, which means |hi,1|2
Similarly,
≤,...,|hi,j|2 ≤,...,≤ |hi,N|2,∀i ∈ [1,M],j ∈ [1,N], vice versa. In this case,
pˆn = pˆn,1 + pˆn,2 + ... + pˆn,M, (5)
after successive interference cancellation (SIC), NOMA co-
is the self-interference power when transmitting information to channel interference of the i-th user are from the (i + 1)-th user
M destinations from source n. to the N-th user [50].
Remark 1: The received signal is composed of the received According to Shannon theory [19], achievable capacity of
downlink information and its self-interference from the FD each destination can be given by (6), see the equation in the top
uplink. (1.10) On the other hand, transmission and reception of next page. Here yields the co-channel
processes in the FD-NOMA-based decentralized V2X systems interference from neighboring users after SIC, ηpˆi,k is the
are different from the centralized cellular-based selfinterference by FD uplink, σ2 is the channel noise power,
communications, i.e., each V2X destination can receive respectively. Additionally, η is the coefficient of self-
information with different NOMA power vectors from multiple interference with η ∈ [0,1], which makes our expressions
distributed sources. By invoking the FD-NOMA techniques for versatile to describe different schemes. For instance, in FD-
simultaneous transmission and reception, the power received NOMA scheme, large value of η denotes the strong FD self-
and transmitted by each V2X device are pn,pˆn. interference, and small value denotes the weak FD self-
interference. On condition that η = 0, the expression reduces to
III. ERGODIC CAPACITY ANALYSIS IN DIFFERENT SCENARIOS the pure NOMA expression. On the basis of (6), normalizing the
channel noise power value will give (7). Here ρ is the SNR, and
In literature, capacity analysis is to reveal the intuitive and we use αi,j,αi,l,αi,k to denote the allocated NOMA power
simple-to-compute capacity expressions for the wireless coefficient with FD transmission in line with a normalized
systems [39], [40]. In this regard, closed-form capacity channel noise power value. In the sequel, we adopt the
expression is of great importance. Generally, capacity can be normalized noise power.
classified into two different types, i.e., the ergodic (Shannon)
capacity and the outage capacity [41]. In time varying channels,
A. Ergodic Capacity Analysis in Urban and Crowded Scenario
on condition that the channel state information (CSI) is known
We first analyze the achievable sum capacity in urban and
at the receiver but not the transmitter, i.e., γ (signal to
crowded scenario. Note that we use the superscript a and c to
interference plus noise ratio (SINR)) is known for every time
distinguish different scenarios. In urban and crowded scenario,
slot. On condition that the CSI cannot be perfectly obtained
PDF of instantaneous signal to interference plus noise ratio
because of the feedback delay or channel estimation error [42]–
(SINR) in each time slot, say, γi,j, is given by
[44], the system performance is reduced. The hardware
impairment (HI) is another issue that reduce the system
performance. The impacts of imperfect CSI and HI to the , (8)
NOMA-based systems have been investigated in prior works, where
see, e.g., [45], [46]. In literature, there are some methods to
alleviate the side-effects of imperfect CSI and HI, for instance, , (9)
the deep learning-based CSI prediction [47]. In addition, some
is the averaged channel power gain of each destination. As is
transmission methods without CSI have been investigated, see,
well known, ergodic capacity is achieved by experiencing all the
e.g., [48].
channel fading states, which means
On condition that the distribution of γ is known at both the
transmitter and receiver. Ergodic capacity is defined by data
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TCOMM.2019.2904499, IEEE Transactions on Communications
5
(10) bk = . (15) θk −
θk−1
In addition, θk,k ∈ [0,n + 1] is given by 0 ≤ θ0 < θ1 <
. Besides, as,bs,θs are defined with the
same method, i.e.,
.

In the following theorem, we provide the exact ergodic capacity , (16)

expression of the FD-NOMA-based decentralized V2X system- cotθs−1 − cotθs
s. bs = , (17) θs −
Theorem 1: In urban and crowded scenario, the exact θs−1
achievable sum ergodic capacity of the FD-NOMA-based and . It is also worth
decentralized V2X systems is noting that the approximation accuracy is controlled by the
division of with θk and θs (associate with as,bs)2.
(11) Proof: See Appendix B.
In order to verify the tightness of this expression, we
where E1(x) is the exponential integral function that defined as compare the performances of the exact expression, the
(12) approximate expression and the well known Swamee and Ohija
approximation. Note that the Swamee and Ohija approximation
Additionally, we have γ¯i,j given as expression is given by [31]
(9).
(18)
Proof: See Appendix A.
Exact ergodic capacity expression in urban and crowded where
scenario is provided in Theorem 1. Since the exponential ,
integral function is involved, this expression thus is not given in (19)
closed-form. We thereby further pursue an approximate
closedform expression of the achievable capacity. As noticed, in (20)
(11), the only expression not given by closed-form is the
generalized exponential integral functions. In this case, our Here while using the approximate expression in Lemma 1, we
main focus is to find out a closed-form expression of E1(x).

, (6)

, (7)

Lemma 1: Closed-form expression (lower bound) of the divide the with 1000 segments, which means, θk−θk−1 =
generalized exponential integral function is given by . The simulation results are given by Fig. 2. As noticed, the
gap between the approximation and the exact form curves is
large. Although this approximation method is better than the
, (13) Swamee and Ohija approximation method, it is still unsuitable
where ak,bk are defined as to be adopted directly.
We notice from Appendix B that in our derivations, the only
, (14) issue that might bring in difference is the Jensen’s inequality,
i.e., in the derivations of Q(x)-function’s closed-form
cotθk−1 − cotθk expression, we use

2 It is worth noting that here in our analysis, the equal division of is used.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TCOMM.2019.2904499, IEEE Transactions on Communications
6

. (21)
Additionally, one can see from Fig. 2 that the approximation
curve displays a similar curvature to the exact curve. We can
expect that a coefficient factor to the closed-form expression
might improve the accuracy, i.e.,

. (22)
Consequently, our task is to find out a q satisfying

(23)
Here we use . After some manipulations, we notice
that when , the above condition is met (e.g.,
E1(1)| = |0.2193827−0.2193839| = 1.2187×10−6). We thus Fig. 2: Comparison of the exact, approximation and Swamee
have an approximate closed-form expression of E1(x) as and Ohija-based expressions.

. (24)
We further give the comparison results of the exact, improved
and approximate expressions, which is shown in Fig. 3.
Compared to the approximate results, the improved
approximate results coincide with the exact results perfectly,
which indicates the validity of our hypothesis. Closed-form
expression of CsumRay is given by the following corollary.
Corollary 1: By substituting (24) into (11), we obtain the
approximate closed-form expression of the achievable capacity
in urban and crowded scenario

.
(25)
Fig. 3: Comparison of the exact, improved and approximate
Remark 2: Insights from Corollary 1 is that the system ergodic
expressions.
capacity in urban and crowded scenario is determined by M,N,
γ¯i,j. The system capacity increases with M,N. The accuracy of
this approximate closed-form expression is determined by n,t. B. Ergodic Capacity Analysis in Suburban and Remote Scenario
That is, the divisions of . The validity of this approximate In the subsection III. A, we have obtained both exact and
expression will be verified by the following numerical results. approximate forms of the capacity of the FD-NOMA-based
decentralized V2X systems in urban and crowded scenario. In
this subsection, we focus on the system capacity analysis in the
suburban and remote scenario. We use K as the Rician factor
(which is the ratio between the deterministic and random
fastfading component). It is noticed that in Rician channel, we
have

. (26)
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TCOMM.2019.2904499, IEEE Transactions on Communications
7
where r2 yields the channel gain of LoS component, 2ω2 is the words, we may obtain an approximate expression with an
average channel power gain of all NLoS components. By arbitrary small error when
defining the total average power gain as γ¯ and following the
prior work in [51], PDF of γi,j can be given as
(33)
Insight from Corollary 2 has that the system capacity expression
is determined by M,N, γ¯i,j and K. With M,N increasing, the
Here I0(·) is the first kind modified Bessel function with zeroth system capacity always increases. The precise effects of γ¯i,j,K to
order. By following a similar procedure of the previous analysis, the capacity are still nonintuitive, which will be discussed in the
we can obtain Theorem 2. following section.
Theorem 2: Exact ergodic capacity expression of the
FDNOMA-based decentralized V2X systems in suburban and IV. NUMERICAL RESULTS
remote scenario is given by In this section, we perform the Monte Carlo simulations to
verify the validity of our analysis. We also perform simulations
to exposit the effects of different parameters to the system
capacity, and compare the performance between FD-NOMA
Here En(x) is the generalized exponential integral function and NOMA schemes based on the decentralized FD-
defined as [52] NOMAenabled V2X systems. Due to variable parameters, we
separately explain them and their values in the following
, (29) simulations.
We first check the validity of the derived capacity expressions
where Re(x) yields the real part of x. in (25) and (31). In these simulations, for the sake of
Proof: See Appendix C. compactness, one source with multiple destinations are used,
This expression is still intractable to use directly because of where the source employs the FD-NOMA scheme to serve these
the involved infinite factorial and generalized exponential destinations. We also assume that the allocated NOMA power
integral expressions. In order to tame this troublesome variance is growing linearly with a normalized noise variance
problem, we give one approximate expression with arbitrary value (e.g., with 4 users, the NOMA power vector is ai =
small error by invoking the truncation method in the sequel. [4,3,2,1]), where ai = [αi,1,...,αi,N]). Additionally, η = 0.1,αi,k = 5
We find that the following expression are used. As clearly shown by Fig. 4 and
Fig. 5, our analytical results3 and the MC results almost exactly
coincide, which demonstrates the validity of our analysis. For
(30) instance, in Fig. 4, with ρ = 15 dB, 1 ↔ 4, the MC and
has an upper ceiling approximation, as shown by Corollary 2. In App results are respectively 3.6865,3.6866 Bit/S/Hz. Under the
this case, the system capacity can be given by an approximate same condition, as shown in Fig. 5, MC result and App result are
expression with much lower computation complexity and 3.8458,3.8455 Bit/S/Hz, respectively. The differences are less
arbitrary small error, . than 0.001 Bit/S/Hz in both scenarios. We also observe that as
Corollary 2: By truncating the infinite series with regard to the values of N,ρ increases, the system capacity always
T, the capacity expression is approximately given as increases. By comparing Fig. 4 and Fig. 5, we notice that under
the same condition, capacities in suburban and remote scenario
always outperform the ones in urban and crowded scenario (for
instance, in 1 ↔ 3 case, SNR = 0 dB, ; SNR
= 30 dB, ). This is because of the less
The truncation error is propagation loss with a dominant LoS path between source and
destination in the suburban and remote scenario.
In order to verify the benefits of our analytical expressions,
we compare the consumed time of App and MC simulations in
Proof: See Appendix D. Table I with ρ = 15 dB as an example. In these simulations,
Remark 3: One can notice that the accuracy of the eight-core 3.4 GHz processors, 16 GB memory and windows 10
approximate expression in (31) is controlled by T. In other 64-bit operating system are used. The results are rounded off

3 App: approximate, MC: Monte Carlo.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TCOMM.2019.2904499, IEEE Transactions on Communications
8
to four decimal places. As shown in Table I, the consumed time also increases. This is because the higher K brings in a stronger
of our analytical expressions are about 106 times shorter than LoS component and a weaker multi-path propagation loss.
the MC simulations. Besides the effects of N,ρ,K, the effects of M and ai to the
In the next step, we check the effect of Rician factor K to the system capacity are also checked with: 1) a linearly growing
system capacity in suburban and remote scenario. In order to power value with M = 1 (i.e., a1 = [0.5,1,1.5],a2 = [1,2,3],a3 =
keep K as the only variable, we do some manipulations as [2,4,6]); 2) different NOMA power vectors with M = 2 (i.e., 2 ↔
follows: 1) we keep all variables consistent except K; 2) with 3,a1,a3 denote that two sources are transmitting information to
normalized noise power value and 3 destinations, we set ai = 3 destinations with FD-NOMA, where the NOMA power vector
[1,2,3]. The simulation results of the system capacity vs the are a1,a2, respectively). The simulation results are given by Fig.
destination number in suburban and remote scenario is given 7 and Fig. 8. As shown by the solid lines in both figures,
in Fig. 6. We notice that as the K increases, system capacity increasing the power values leads

Fig. 4: Comparison of the system achievable sum capacity

performances of App and MC results in urban and crowded
scenario. The approximate results are obtained according to
(25).

Fig. 5: Comparison of the system achievable sum capacity

performances with App and MC results in suburban and remote
scenario. The analytical results are obtained according to (31).
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TCOMM.2019.2904499, IEEE Transactions on Communications
9
TABLE I: Consumed time (second) of App and MC simulations with ρ = 15 dB.
App 0.0002 0.0001 0.0001 0.0001 0.0004 0.0000 0.0000 0.0000
Urban and crowded scenario MC 83.0272 89.2913 83.0986 80.8462 81.7240 87.6330 91.5975 87.3378
Suburban and remote App 0.0781 0.0153 0.0036 0.0034 0.0066 0.0035 0.4032 0.06056
scenario MC 92.9720 89.8830 89.6005 92.4459 95.3015 97.2475 95.9309 94.8763

Fig. 6: Comparison of the capacities with different power values NOMA, NOMA, FD-OMA and HD-OMA schemes in different Fig.
and source numbers. 8: Comparison of the capacities with different power values and
source numbers.

scenarios. The results are given in Fig. 9 and Fig. 10. In these
simulations, carrier bandwidth B = 100 MHz, ai = [3,2,1],η = 0.1
and αi,k = 0.1,1,10 are used. In order to be fair, we average the
allocated power in FD-OMA and HD-OMA schemes. As shown
in both figures, NOMA scheme has a better throughput
performance compared to OMA scheme. Moreover, with a
smaller value of αi,k, FD-NOMA always outperforms the other
schemes (HD-NOMA, FD-OMA, HD-OMA). However, the benefit
of FD-NOMA decreases while αi,k increasing. This is mainly due
to the increased FD self-interferences. We also notice that even
with a higher FD self-interference value, FD-NOMA
outperforms NOMA in low SNR scenario (i.e., ρ ∈ [0,5] dB). This
is because in low SNR scenario, channel noise is the dominant
factor compared to FD self-interference. In contrast, FD-NOMA
self-interference becomes the dominant factor in high SNR
Fig. 7: Comparison of the capacities with different power values scenario, NOMA scheme without FD selfinterference thus has a
and source numbers. better throughput performance. It is also worth noting that the
effective transmission time is limited because of the fast
moving V2X devices. FD-NOMA enabled bidirectional
to better capacity performance, which is due to the increased transmission can greatly reduce the transmission latency
SNR value. For instance, in 1 ↔ 3 case and SNR = 20 dB, we compared to other schemes. For example, compared to HD-
have . We can also confirm from NOMA and HD-OMA, FD-NOMA only needs a half latency time
both figures that as M increases, the system capacities also to transmit the same amount of data by its simultaneous
increase. transmission and reception scheme.
Finally, we compare the achievable throughputs with FD-
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TCOMM.2019.2904499, IEEE Transactions on Communications
10
V. CONCLUSION scenario is determined by the associated division of (with
In this article, we proposed the FD-NOMA-based respect to exponential integral function (E1(x)), and the
decentralized V2X systems. We classified the V2X accuracy of simplified approximate expression in suburban and
communications remote scenario is determined by the truncation point T (with
respect to generalized exponential integral function (En(x)).
Numerical results demonstrate the validity and effectiveness of
our analytical results. Compared to MC method, the consumed
time is greatly reduced by our Approximation expressions.
Simulation results also demonstrated that the system capacity
performance can be enhanced by increasing the number of V2X
devices, NOMA power and Rician factor (suburban and remote
scenario), and the effectiveness of FD-NOMA is determined by
the FD self-interference and the channel noise. In addition, FD-
NOMA can greatly reduce the system latency compared to
other schemes.

APPENDIX A: PROOF OF THEOREM 1

Firstly, according to the integration by parts method, we have

Fig. 9: System achievable throughput comparisons with

FDNOMA, NOMA, FD-OMA and HD-OMA schemes in urban and
crowded scenario.

(A.1)
So far the expression is still intractable. In the next step, we
recall the alternative generalized exponential integral
expression [52]

Fig. 10: System achievable throughput comparisons with (A.2)

FDNOMA, NOMA, FD-OMA and HD-OMA schemes in suburban
and remote scenario.
,

By substituting (A.2) into (A.1), and further summarizing the

into two typical scenarios, i.e., the urban and crowd scenario
result with M sources and N destinations, we can safely arrive
and the suburban and remote scenario, and then derived the
the final expression.
exact system capacity expressions in both scenarios. To tackle
This completes the proof.
down the capacity expression’s intractable calculations in both
APPENDIX B: PROOF OF LEMMA 1
scenarios, we further obtained their simplified approximate
expressions. Insights of our analysis are that the accuracy of our As noticed, E1(x) can be rewritten as
simplified approximate expression in urban and crowded
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TCOMM.2019.2904499, IEEE Transactions on Communications
11
The approximate accuracy of this lower bound expression is
(B.1) controlled by the interval gap between each pair of [θk−1,θk].
It is also noticed that the following equality holds [53] Moreover, it is known that the following equality holds
Z
sin−2 xdx = −cotx + C. (B.7)

Substituting it into (B.6), we thus have

(B.2)
Thus in the next step, our work is to seek a closed-form . (B.8)
expression for the Q-function. Actually, there are various Finally, by substituting (B.2) into (B.1), E1(x) can be given as
closedform expressions to capture the lower or upper bounds
of the Q-function, for instance, the chernoff bound

, (B.3)

the improved exponential bound

. (B.4)

However, the integral is still intractable while substituting those

expressions into (B.1), an alternative method is needed.
According to prior work, by adopting the Craig’s form, we have
[53]
(B.9)

As the final inequality is obtained while substituting the ap-√

proximate expression of Q( 2bkx), definitions of as,bs,θs thus are
similar as prior definitions of ak,bk,θk, i.e., as =

and 0 ≤ θ0 < θ1 < ...θs < ... <

This completes the proof.
(B.5)

with ak,bk are defined as cot

θθkk−−1θ−k−cot1 θk .
Then by substituting the Jensen’s inequality [54] to (B.5), we APPENDIX C: PROOF OF THEOREM 2
have the lower bound expression of Q-function as

It is noticed that the PDF of γi,j in Rician channel condition can

be given by [51]

Additionally, it is worth noting that θk,k ∈ [0,n + 1] is given by

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TCOMM.2019.2904499, IEEE Transactions on Communications
12
By following a similar derivation procedure as in Theorem 1, we
have the following equation

As shown here, approximate error mainly comes from the

infinite expression series with regard to m. According to prior
Ci,jc = E[log2(1 + ¯γi,j)]
work in [56], [57], En(x) monotonically decreasing in n giving
Z +∞ (C.2) c
equal x. In this case, by putting part
= log2(1 + γi,j)f (γi,j)dγi,j.
0 aside, we have the following expression

and substituting equations (C.4) and (C.5) into this expression, (D.2)
the derivations of capacity expression of the FD-NOMA-based .
decentralized V2X systems can be given as:
It is noticed that giving constant values of γ¯i,j and
then becomes a constant coefficient. Consequently, we focus
on the function

. (D.3)
By some mathematical manipulations, it is found that there
existing x0, so that f0(x0) = 0 with f00(x0−) > 0,f00(x0+) < 0.
Additionally, observation has that f(x) rapidly converges to 0
after x0 (e.g., f(100) = 6.9966 × 10−125). This gives approximate
capacity expression of (31) with an arbitrary small error .
This completes the proof.

REFERENCES
[1] Z. Zhou, H. Yu, C. Xu, Y. Zhang, S. Mumtaz, and J. Rodriguez, “Dependable
content distribution in D2D-based cooperative vehicular networks: A big
data-integrated coalition game approach,” IEEE Trans. Intell. Transp. Syst.,
vol. 19, no. 3, pp. 953–964, Mar. 2018.
[2] J. B. Kenney, “Dedicated short-range communications (DSRC) standards
in the united states,” Proc. IEEE, vol. 99, no. 7, pp. 1162–1182, Jul. 2011.
[3] B. Walke, W. Mende, and G. Hatziliadis, “CELLPAC: A packet radio protocol
applied to the cellular GSM mobile radio network,” in IEEE VTC, May 1991,
pp. 408–413.
[4] W. Zhu, T. Hellmich, and B. Walke, “Dcap, a decentral channel access
(C.3) protocol: performance analysis,” in IEEE VTC, May 1991, pp. 463–468.
Here the second equality is due to the modified Bessel function [5] Z. Zhu, Z. Wang, Z. Chu, D. Zhang, and B. Shim, “Robust energy harvest
balancing optimization with V2X-SWIPT over MISO secrecy channel,”
of the zeroth order expression [55] Comput. Netw., vol. 137, no. 4, pp. 61–68, Jun. 2018.
[6] Z. Xu, X. Li, X. Zhao, M. H. Zhang, and Z. Wang, “DSRC versus 4GLTE for
connected vehicle applications: A study on field experiments of vehicular
communication performance,” J. Adv. Transp., vol. 2017, pp. 1–10, Aug.
. (C.4) 2017.
Additionally, the third equality is because [55] [7] G. Americas, “Cellular V2X communications towards 5G,” 5G Americas,
Tech. Rep., Mar. 2018.
[8] L. Liang, H. Peng, G. Y. Li, and X. Shen, “Vehicular communications: A
physical layer perspective,” IEEE Trans. Veh. Tech., vol. 66, no. 12, pp.
10647–10659, Dec. 2017.
[9] B. Hui, J. Kim, H. Chung, I. Kim, and H. Lee, “Efficient doppler mitigation
By further summarizing this expression with M sources and N for high-speed rail communications,” in IEEE ICACT, Jan. 2016, pp. 634–
destinations, we can finally arrive at (31). This completes the 638.
proof. [10] A. Gomaa and L. M. A. Jalloul, “Receiver architecture for frequency offset
correction and I/Q imbalance compensation in equal bandwidth
APPENDIX D: PROOF OF COROLLARY 2 contiguous carrier aggregation,” IEEE Trans. Veh. Technol., vol. 65, no. 10,
pp. 8730–8735, Oct. 2016.
The remaining section after a truncation with regard to T is
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TCOMM.2019.2904499, IEEE Transactions on Communications
13
[11] H. Ji, S. Park, J. Yeo, Y. Kim, J. Lee, and B. Shim, “Ultra-reliable and [32] G. Navas-Palencia, “Fast and accurate algorithm for the generalized
lowlatency communications in 5G downlink: Physical layer aspects,” IEEE exponential integral eν(x) for positive real order,” Numerical Algorithms,
Wireless Commun., vol. 25, no. 3, pp. 124–130, Jun. 2018. vol. 77, no. 2, pp. 603–630, Feb. 2018.
[12] J. Wu, M. Dong, K. Ota, J. Li, and Z. Guan, “Big data analysis-based secure [33] D. Zhang, Z. Zhou, S. Mumtaz, J. Rodriguez, and T. Sato, “One integrated
cluster management for optimized control plane in softwaredefined energy efficiency proposal for 5G IoT communications,” IEEE Internet
networks,” IEEE Trans. Netw. Service Manag., vol. 15, no. 1, pp. 27–38, Things J., vol. 3, no. 6, pp. 1346–1354, Dec. 2016.
Mar. 2018. [34] A. Bazzi, B. M. Masini, and A. Zanella, “How many vehicles in the LTE-V2V
[13] X. Yue, Y. Liu, S. Kang, A. Nallanathan, and Z. Ding, “Exploiting full/half- awareness range with half or full duplex radios?” in 2017 15th
duplex user relaying in NOMA systems,” IEEE Trans. Commun., vol. 66, no. International Conference on ITS Telecommunications (ITST), May 2017,
2, pp. 560–575, Feb. 2018. pp. 1–6.
[14] W. Liang, Z. Ding, Y. Li, and L. Song, “User pairing for downlink [35] P. Wang, B. Di, H. Zhang, K. Bian, and L. Song, “Cellular V2X
nonorthogonal multiple access networks using matching algorithm,” IEEE communications in unlicensed spectrum: Harmonious coexistence with
Trans. Commun., vol. 65, no. 12, pp. 5319–5332, Dec. 2017. VANET in 5G systems,” IEEE Trans. Wireless Commun., 2018.
[15] D. Zhang, Y. Liu, Z. Ding, Z. Zhou, A. Nallanathan, and T. Sato, [36] M. Boban, K. Manolakis, M. Ibrahim, S. Bazzi, and W. Xu, “Design aspects
“Performance analysis of non-regenerative Massive-MIMO-NOMA relay for 5G V2X physical layer,” in IEEE CSCN, Oct. 2016, pp. 1–7.
systems for 5G,” IEEE Trans. Commun., vol. 65, no. 11, pp. 4777–4790, [37] R. J. C. Bultitude and G. K. Bedal, “Propagation characteristics on
Nov. 2017. microcellular urban mobile radio channels at 910 MHz,” IEEE J. Sel. Areas
[16] L. Dai, B. Wang, Y. Yuan, S. Han, C. l. I, and Z. Wang, “Non-orthogonal Commun., vol. 7, no. 1, pp. 31–39, Jan. 1989.
multiple access for 5G: solutions, challenges, opportunities, and future [38] Q. Zhang, T. Q. S. Quek, and S. Jin, “Scaling analysis for massive MIMO
research trends,” IEEE Commun. Mag., vol. 53, no. 9, pp. 74–81, Sep. systems with hardware impairments in Rician fading [early access:]
2015. https://ieeexplore.ieee.org/document/8350341/,” IEEE Trans. Wireless
[17] H. Zhang, N. Yang, K. Long, M. Pan, G. K. Karagiannidis, and V. C. M. Leung, Commun., Apr. 2018.
“Secure communications in NOMA system: Subcarrier assignment and [39] C. E. Shannon, “A mathematical theory of communication,” Bell Labs Syst.
power allocation,” IEEE J. Sel. Areas Commun., vol. 36, no. 7, pp. 1441– Tech. J., vol. 27, no. 3, pp. 379–423, Jul. 1948.
1452, Jul. 2018. [40] S. Weber, J. G. Andrews, and N. Jindal, “An overview of the transmission
[18] H. Zhang, N. Yang, K. Long, M. Pan, G. K. Karagiannidis, and A. capacity of wireless networks,” IEEE Trans. Commun., vol. 58, no. 12, pp.
Nallanathan, “Energy efficient resource allocation for secure NOMA 3593–3604, Dec. 2010.
networks,” in IEEE VTC Spring, Jun. 2018, pp. 1–6. [41] S. Choudhury and J. D. Gibson, “Information transmission over fading
[19] Y. Liu, M. Elkashlan, Z. Ding, and G. K. Karagiannidis, “Fairness of user channels,” in IEEE GLOBECOM, Nov. 2007, pp. 3316–3321.
clustering in MIMO non-orthogonal multiple access systems,” IEEE [42] H. Y. H. Q. N. Thomas L Marzetta, Erik G Larsson, “Fundamentals of
Commun. Lett., vol. 20, no. 7, pp. 1465–1468, Jul. 2016. massive mimo,” Cambridge, UK, 2016.
[20] F. Fang, H. Zhang, J. Cheng, and V. C. M. Leung, “Energy-efficient resource [43] S. S. Wei Xiang, Kan Zheng, “5G mobile communications,” AG,
scheduling for NOMA systems with imperfect channel state information,” Switzerland, 2017.
in IEEE ICC, May 2017, pp. 1–5. [44] T. Van Chien, C. Molln, and E. Bjrnson, “Large-scale-fading decoding in
[21] A. Sabharwal, P. Schniter, D. Guo, D. W. Bliss, S. Rangarajan, and R. cellular massive MIMO systems with spatially correlated channels [early
Wichman, “In-band full-duplex wireless: Challenges and opportunities,” access:] https://ieeexplore.ieee.org/document/8585131,” IEEE Trans.
IEEE J. Sel. Areas Commun., vol. 32, no. 9, pp. 1637–1652, Sep. 2014. Commun., 2018.
[22] Z. Ding, P. Fan, and H. V. Poor, “On the coexistence between full-duplex [45] P. T. M. D. Z. L. L. X. Li, J. Li and J. Jin, “Joint impact of hardware
and NOMA [early access:] impairments and imperfect CSI on cooperative SWIPT NOMA
https://ieeexplore.ieee.org/document/8306094/,” IEEE Wireless multirelaying systems,” in IEEE ICCC, Aug. 2018, pp. 95–99.
Commun. Lett., Mar. 2018. [46] D. Z. X. L. J. L. C. C. C. C. Deng, X. Zhao, “Performance analysis of noma-
[23] H. Huang, J. Xiong, J. Yang, G. Gui, and H. Sari, “Rate region analysis in a based relaying networks with transceiver hardware impairments,” vol. 12,
full-duplex-aided cooperative nonorthogonal multiple-access system,” no. 9, pp. 4295–4316, Sep. 2018.
IEEE Access, vol. 5, pp. 17869–17880, Aug. 2017. [47] C. Luo, J. Ji, Q. Wang, X. Chen, and P. Li, “Channel state information
[24] M. F. Kader, S. Y. Shin, and V. C. M. Leung, “Full-duplex non-orthogonal prediction for 5G wireless communications: A deep learning approach
multiple access in cooperative relay sharing for 5G systems,” IEEE Trans. [early access:] https://ieeexplore.ieee.org/document/8395053,” IEEE
Veh. Tech., vol. 67, no. 7, pp. 5831–5840, Jul. 2018. Trans. Netw. Sci. Eng., 2018.
[25] B. Di, L. Song, Y. Li, and G. Y. Li, “Non-orthogonal multiple access for high- [48] S. A. Jafar, “Blind interference alignment,” IEEE J. Sel. Topics Signal
reliable and low-latency V2X communications in 5G systems,” IEEE J. Sel. Process., vol. 6, no. 3, pp. 216–227, Jun. 2012.
Areas Commun., vol. 35, no. 10, pp. 2383–2397, Oct. 2017. [49] A. Goldsmith, “Wireless communciations,” New York, USA, 2005.
[26] B. W. Khoueiry and M. R. Soleymani, “An efficient NOMA V2X [50] Y. Liu, Z. Ding, M. Elkashlan, and H. V. Poor, “Cooperative nonorthogonal
communication scheme in the internet of vehicles,” in IEEE VTC, Jun. multiple access with simultaneous wireless information and power
2017, pp. 1–7. transfer,” IEEE J. Sel. Areas Commun., vol. 34, no. 4, pp. 938– 953, Apr.
[27] M. Yang, S. W. Jeon, and D. K. Kim, “Interference management for inband 2016.
full-duplex vehicular access networks,” IEEE Trans. Veh. Technol., vol. 67, [51] H. C. Yang and M. S. Alouini, “Order statistics in wireless communications:
no. 2, pp. 1820–1824, Feb. 2018. Diversity, adaptation, and scheduling in mimo and ofdm systems,” New
[28] M. D. Yacoub, “The κ−µ distribution and the η−µ distribution,” IEEE York, USA, 2011.
Antennas and Propag. Mag., vol. 49, no. 1, pp. 68–81, Feb. 2007. [52] I. S. Gradshteyn and I. M. Ryzhik, “Table of integrals, series and products,”
[29] N. Bhargav, S. L. Cotton, and D. E. Simmons, “Secrecy capacity analysis London, UK, 2015.
over κ c µ fading channels: Theory and applications,” IEEE Trans. [53] A. A. Alkheir and M. Ibnkahla, “An accurate approximation of the
Commun., vol. 64, no. 7, pp. 3011–3024, Jul. 2016. exponential integral function using a sum of exponentials,” IEEE Commun.
[30] X. Li, J. Li, L. Li, J. Jin, J. Zhang, and D. Zhang, “Effective rate of miso systems Lett., vol. 17, no. 7, pp. 1364–1367, Jul. 2013.
over κ - µ shadowed fading channels,” IEEE Access, vol. 5, pp. 10605– [54] M. Wu, X. Lin, and P. Y. Kam, “New exponential lower bounds on the
10611, Jun. 2017. Gaussian Q-Function via Jensen’s inequality,” in IEEE VTC Spring, May
[31] G. P. Huy, “Revisit of well function approximation and an easy graphical 2011, pp. 1–5.
curve matching technique for theis solution,” Groundwater, vol. 41, no. 3,
pp. 387–390, Dec. 2005.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TCOMM.2019.2904499, IEEE Transactions on Communications
14
[55] A. L. Giuseppa Alfano, A. M. Tulino, and S. Verdu, “Mutual information Linglong Dai (M’11-SM’14) received the B.S. degree
and eigenvalue distribution of MIMO Rician channels,” in IEEE ISITA, from Zhejiang University in 2003, the M.S. degree (with
Parma, Italy, Oct. 2004, pp. 1–6. the highest honor) from the China Academy of
[56] J. Zhang, L. Dai, X. Zhang, E. Bjonson, and Z. Wang, “Achievable¨ rate of Telecommunications Technology in 2006, and the
Rician large-scale MIMO channels with transceiver hardware Ph.D. degree (with the highest honor) from Tsinghua
impairments,” IEEE Trans. Veh. Tech., vol. 65, no. 10, pp. 8800–8806, Oct. University, Beijing, China, in 2011. From 2011 to 2013,
2016. he was a Post-Doctoral Research Fellow with the
[57] C. Chiccoli, S. Lorenzutta, and G. Maino, “Recent results for generalized Department of Electronic Engineering, Tsinghua
exponential integrals,” Comput. Math. Appl., vol. 19, no. 5, pp. 21 – 29, University, where he was an Assistant Professor from
1990. 2013 to 2016 and has been an Associate Professor
since 2016. He co-authored the book “mWave Mas-
sive MIMO: A Paradigm for 5G” (Academic Press, Elsevier, 2016). He has
published over 60 IEEE journal papers and over 40 IEEE conference papers. He
also holds 16 granted patents. His current research interests include massive
MIMO, millimeter-wave communications, NOMA, sparse signal processing, and
machine learning for wireless communications. He has received five IEEE Best
Paper Awards at the IEEE ICC 2013, the IEEE ICC 2014, the IEEE ICC 2017, the
IEEE VTC 2017-Fall, and the IEEE ICC 2018. He has also received the Tsinghua
University Outstanding Ph.D. Graduate Award in 2011, the Beijing Excellent
Doctoral Dissertation Award in 2012, the China National Excellent Doctoral
Dissertation Nomination Award in 2013, the URSI Young
Scientist Award in 2014, the IEEE Transactions on Broadcasting Best Paper
Award in 2015, the Electronics Letters Best Paper Award in 2016, the National
Natural Science Foundation of China for Outstanding Young Scholars in 2017,
Di Zhang (S’13-M’17) is an Assistant Professor with the the IEEE ComSoc Asia-Pacific Outstanding Young Researcher Award in 2017,
Zhengzhou University, Zhengzhou, China, and also a and the IEEE ComSoc Asia-Pacific Outstanding Paper Award in 2018. He
visiting researcher with the Seoul National University, currently serves as an Editor of the IEEE Transactions on Communications, the
Seoul, Korea. He received his Ph.D. degree from the IEEE Transactions on Vehicular Technology, and the IEEE Communications
Waseda University, Tokyo, Japan, in 2017. He visited Letters. Particularly, he is dedicated to reproducible research and has made a
the National Key Laboratory of large amount of simulation code publicly available.
Alternate Electrical Power System with Renewable Ali Kashif Bashir (M’15-SM’16) is a Senior Lecturer at
Energy Sources, Beijing, China, from 2015 to 2017, and School of Computing, Mathematics, and Digital
the National Chung Hsing University, Taichung, Technology, Manchester Metropolitan University,
Taiwan, in 2012. He serves as the editor of the KSII U.K. He is a senior member of IEEE and Distinguished
Speaker of ACM. His past assignments include:
TRANSACTIONS ON INTERNET AND INFORMATION SYSTEMS,
Associate Professor of Information and
has served as the guest editors of the IEEE NETWORK, the IEEE
Communication Technologies, Faculty of Science and
ACCESS, the IET INTELLIGENT TRANSPORT SYSTEMS, TPC members of many IEEE
Technology, University of the Faroe Islands, Denmark;
flagship conferences, such as ICC and WCNC. His research interests include 5G
Osaka University, Japan; Nara National College of
wireless networks, internet of things and telemedicine.
Technology, Japan; the National Fusion Research
Yuanwei Liu (S’13-M’16) received the B.S. and M.S.
Institute, South Korea; Southern Power Company
degrees from the Beijing University of Posts and Ltd., South Korea, and the Seoul Metropolitan Government, South Korea.
Telecommunications in 2011 and 2014, respectively,
He received his Ph.D. in computer science and engineering from Korea
and the Ph.D. degree in electrical engineering from the
Queen Mary University of London, U.K., in 2016. He University, South Korea. MS from Ajou University, South Korea and BS from
was with the Department of Informatics, King’s College University of Management and Technology, Pakistan. He is author of over 80
London, from 2016 to 2017, where he was a Post- peer-reviewed articles. He is supervising/co-supervising several graduate (MS
Doctoral Research Fellow. He has been a Lecturer and PhD) students. His research interests include internet of things, wireless
(Assistant Professor) with the School of Electronic networks, distributed systems, network/cyber security, cloud/network
Engineering and Computer Science, Queen Mary function virtualization, etc. He is serving as the Editor-in-chief of the IEEE FUTURE
University of London, since 2017. DIRECTIONS NEWSLETTER.
His research interests include 5G wireless networks, Internet of Things, He is editor of several journals and also has served/serving as guest editor on
machine learning, stochastic geometry, and matching theory. He received the several special issues in journals of IEEE, Elsevier, and Springer. He has served
Exemplary Reviewer Certificate of the IEEE WIRELESS COMMUNICATION LETTERS in as chair (program, publicity, and track) chair on several conferences and
2015 and the IEEE TRANSACTIONS ON COMMUNICATIONS and the IEEE TRANSACTIONS workshops. He has delivered several invited and keynote talks, and reviewed
ON WIRELESS COMMUNICATIONS in 2016 and 2017. He has served as a TPC Member the technology leading articles for journals, such as the IEEE TRANSACTIONS
for many IEEE conferences, such as GLOBECOM and ICC. He currently serves as ON INDUSTRIAL INFORMATICS, the IEEE COMMUNICATION MAGAZINE, the
an Editor of the IEEE COMMUNICATIONS LETTERS, IEEE INTERNET OF THINGS, and the IEICE Journals,
IEEE TRANSACTIONS ON COMMUNICATIONS, the IEEE COMMUNICATIONS and conferences, such as the IEEE INFOCOM, the IEEE ICC, the IEEE GLOBECOM,
LETTERS and the IEEE ACCESS. He is also a guest editor for IEEE JSTSP special issue and the IEEE Cloud of Things.
on “Signal Processing Advances for Non-Orthogonal Multiple Access in Next
Generation Wireless Networks”.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TCOMM.2019.2904499, IEEE Transactions on Communications
15
View publication stats
Arumugam Nallanathan (S’97-M’00-SM’05-F’17) is
Professor of Wireless Communications and Head of
the Communication Systems Research (CSR) group in
the School of Electronic Engineering and Computer
Science at Queen Mary University of London since
September 2017. He was with the Department of
Informatics at Kings College London from December
2007 to August 2017, where he was Professor of
Wireless Communications from April 2013 to August
2017 and a Visiting Professor from September 2017.
He was an Assistant Professor in the Department
of Electrical and Computer Engineering, National University of Singapore from
August 2000 to December 2007. His research interests include 5G Wireless
Networks, Internet of Things (IoT) and Molecular Communications. He
published nearly 400 technical papers in scientific journals and international
conferences. He is a co-recipient of the Best Paper Awards presented at the
IEEE International Conference on Communications 2016 (ICC’2016) and IEEE
Global Communications Conference 2017 (GLOBECOM’2017). He is an IEEE
Distinguished Lecturer. He has been selected as a Web of Science Highly Cited
Researcher in 2016.
He is an Editor for IEEE TRANSACTIONS ON COMMUNICATIONS. He was an Editor
for IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS (2006-2011), IEEE
TRANSACTIONS ON VEHICULAR TECHNOLOGY (2006-
2017), IEEE WIRELESS COMMUNICATIONS LETTERS and IEEE SIGNAL PROCESSING LETTERS.
He served as the Chair for the Signal Processing and Communication Electronics
Technical Committee of IEEE Communications Society and Technical Program
Chair and member of Technical Program Committees in numerous IEEE
conferences. He received the IEEE Communications Society SPCE outstanding
service award 2012 and IEEE Communications Society RCC outstanding service
award 2014.
Byonghyo Shim (S’95-M’97-SM’09) received the B.S.
and M.S. degrees in control and instrumentation
engineering from Seoul National University, South
Korea, in 1995 and 1997, respectively, and the M.S.
degree in mathematics and the Ph.D. degree in
electrical and computer engineering from the
University of Illinois at Urbana-Champaign, IL, USA, in
2004 and
2005, respectively. From 1997 and 2000, he was an
Officer (First Lieutenant) and also an Academic
Fulltime Instructor with the Department of Electronics
Engineering, Korean Air Force Academy. From 2005 to
2007, he was a Staff Engineer with Qualcomm Inc., San Diego, CA, USA. From
2007 to 2014, he was an Associate Professor with the School of Information and
Communication, Korea University, Seoul. Since 2014, he has been with the
Seoul National University, where he is currently a Professor with the
Department of Electrical and Computer Engineering. His research interests
include wireless communications, statistical signal processing, compressed
sensing, and machine learning. He is an elected member of the Signal
Processing for Communications and Networking Technical Committee of the
IEEE Signal Processing Society. He was a recipient of the M. E. Van Valkenburg
Research Award from University of Illinois in 2005, the
Hadong Young Engineer Award from the IEIE in 2010, the Irwin Jacobs Award
from Qualcomm and KICS in 2016, and the Shinyang research award from SNU
in 2017. He has served as an Associate Editor of the IEEE
TRANSACTIONS ON SIGNAL PROCESSING, the IEEE TRANSACTIONS ON
COMMUNICATIONS, the IEEE WIRELESS COMMUNICATIONS LETTERS, and the JOURNAL OF
COMMUNICATIONS AND NETWORKS, and a Guest Editor of the IEEE JOURNAL ON
SELECTED AREAS IN COMMUNICATIONS.