Content uploaded by Yanshi Sun
Author content
All content in this area was uploaded by Yanshi Sun on Mar 26, 2021
Content may be subject to copyright.
1
A New Design of Hybrid SIC for Improving
Transmission Robustness in Uplink NOMA
Yanshi Sun, Zhiguo Ding, Fellow, IEEE, Xuchu Dai
Abstract—Conventionally, it was thought that outage probabil-
ity error floors are unavoidable in uplink non-orthogonal multiple
access (NOMA) regardless which type of successive interference
cancellation (SIC) is adopted. However, a recent work breaks
such myths by introducing the concept of hybrid SIC, which
dynamically chooses the decoding order according to the users’
channel conditions and quality of service requirements. However,
there is still a limitation in this existing work that the error floors
can only be avoidable under a stringent condition on users’ target
rates. To this end, this letter proposes a new hybrid SIC scheme
by exploiting the benefit of power control, which guarantees no
error floor without any conditions on users’ target rates.
Index Terms—non-orthogonal multiple access, dynamic suc-
cessive interference cancellation, error floor
I. INTRODUCTION
NOn-orthogonal multiple access (NOMA) technique has
received extensive attention in the past few years, due
to its superior spectral efficiency and capability for realizing
massive connectivity [1]–[5]. The key idea of NOMA is to
allow multiple users access to one channel resource block
simultaneously, which is strictly prohibited in conventional
orthogonal multiple access (OMA). However, due to the
channel sharing among users, NOMA can potentially suffer
severe inter-user interference. To suppress the impact of inter-
user interference, the most widely used method is successive
interference cancellation (SIC), where the users’ signals are
decoded in a successive manner [6].
Note that the decoding order of the users’ signals plays
an vital role in SIC techniques. The most well-known SIC
methods in NOMA are mainly based on the following two
types of criteria. One type is based on the channel state
information (CSI) [6], including instantaneous CSI [7], [8] and
statistical CSI [9], [10], where users with better CSI is decoded
first. The other type is based on quality of service (QoS)
requirement, where users with more stringent QoS are decoded
first, while other users are often opportunistically served and
decoded later [11]–[13]. Unfortunately, as discussed in [14],
the outage probabilities achieved by these methods suffers
Copyright (c) 2015 IEEE. Personal use of this material is permitted.
However, permission to use this material for any other purposes must be
obtained from the IEEE by sending a request to pubs-permissions@ieee.org.
The work of Y. Sun and X. Dai was supported by the National Natural
Science Foundation of China under Grant 61971391.
Y. Sun and X. Dai are with the CAS Key Laboratory of Wireless-Optical
Communications, University of Science and Technology of China, Hefei,
230026, China. (email: sys@mail.ustc.edu.cn, daixc@ustc.edu.cn).
Z. Ding is with the Department of Electrical and Electronic Engi-
neering, the University of Manchester, Manchester M13 9PL, U.K. (e-
mail:zhiguo.ding@manchester.ac.uk).
from severe outage probability error floors1, which reduces
the reception reliability of NOMA transmission in practical
systems. Such error floors were usually thought unavoidable in
the implementation of NOMA. However, recent works in [14]–
[16] prove that this view is wrong. Specifically, a new SIC
scheme is proposed, named hybrid SIC, which dynamically
switches the decoding orders according to both the users’ QoS
requirements and channel conditions. It has been shown that
the hybrid SIC scheme in [14]–[16] can avoid the outage
probability error floors. Particularly, a detailed comparison
among hybrid SIC and conventional SIC methods is provided
in [14].
However, a drawback of the work in [14]–[16] is that the
error floors can only be avoidable under a strict condition on
users’ target rates, which might not be realistic in some scenar-
ios. To address this issue, this letter proposes a more general
hybrid SIC scheme, by introducing dynamic power control into
the existing scheme. Excitingly, in the new proposed scheme,
the error floors are avoided without any constraints on the
users’ target rates. To reveal the advantage of the proposed
scheme, comprehensive analysis and detailed comparisons to
the existing scheme in [14]–[16] are developed in this letter.
II. SY ST EM MO DE L
Consider an uplink cognitive radio inspired NOMA (CR-
NOMA) scenario with one base station, one primary user
U0, and one secondary user U1.U0has a stringent QoS
requirement that its preset target data rate R0has to be
satisfied. Meanwhile, U1is opportunistically served without
degrading the primary user’s QoS. It is assumed that each
node is equipped with a single antenna.
A novel hybrid SIC NOMA scheme is proposed to the
considered scenario, where U0and U1are served simultane-
ously by occupying the same resource block, which is solely
occupied by U0in OMA2. The received signal at the base
station is given by
y=gs0+βhs1+n, (1)
where gand hare the channels of U0and U1, respectively.
Rayleigh fading is considered, and gand hare modeled as in-
dependent and identically distributed (i.i.d) circular symmetric
complex Gaussian (CSCG) random variables with zero mean
1We say an outage error floor exists, when the outage probability achieved
by a certain user doesn’t approach zero as SNR goes infinity.
2In the considered OMA scheme, U0is allocated with a dedicated resource
block, say B1, and U1is allocated with another resource block, say B2, which
is orthogonal to B1.
2
TABLE I: SUMMARY OF KEY PARAMETERS
Notation Description
g, h the channel of U0and U1
P0, P1maximum transmit power of U0and U1
βadjustable power coefficient of U1
R0, R1the preset target rate of U0and U1
ϵ0, ϵ1ϵ0= 2R0−1,ϵ1= 2R1−1
and unit variance3.si(i= 0,1) is the transmitted signal of
Ui, each of which is independently coded with Gaussian code
book. The power of siis denoted by Pi, i.e., E{|si|2}=Pi.n
is the Gaussian background noise, whose power is assumed to
be normalized. β∈[0,1] is the adjustable power coefficient of
U1, which means the effective transmit power of U1is βP1.
SIC is carried out at the base station to decode the two
users’ signals. The novelty of the proposed scheme is that the
decoding order and the transmit power of the secondary user
are dynamically chosen, depending on the relationship among
g,h, and R0, as described in the following.
To begin with, we introduce a threshold denoted by τ(g),
which is a function of g. Note that the data rate supported
by U0’s channel in OMA is log(1 + P0|g|2), the definition of
τ(g)can be divided into two cases:
•when log 1 + P0|g|2≥R0,τ(g)is defined as the
maximum interfering power with which the target rate
R0can still be satisfied, i.e.,
log 1 + P0|g|2
τ(g) + 1=R0(2)
•when log 1 + P0|g|2< R0,τ(g)is set to be zero.
Based on the above two cases, τ(g)can be further briefly
expressed as follows:
τ(g) = max 0, P0|g|2/ϵ0−1,(3)
where ϵ0= 2R0−1.
Remark 1. Note that with τ(g)as the interference, U0can
still achieve the same outage performance as in OMA in which
the resource block is solely occupied by U0, i.e.,
Plog 1+ P0|g|2
τ(g)+1<R0
=Plog 1+P0|g|2
< R0
,(4)
where the left term is outage probability of the case with
interference, and the right term is the outage probability
achieved by OMA.
According to the relationship between P1|h|2and τ(g), the
channel condition of the secondary user U1can be categorized
as follows:
•Type I: The received signal power of U1at the base
station is less or equal to τ(g), i.e., P1|h|2≤τ(g). This
type is defined as same as in the benchmark scheme [15].
It is preferable to decode U1’s signal at the second stage
3It is assumed that U1can overhear the information exchange between U0
and the base station, and hence U1has the knowledge of g. Besides, U1can
also obtain the knowledge of hthrough the pilot signal transmitted by the
base station.
of SIC and β= 1, which yields the following data rate
4RI= log(1 + P1|h|2).
•Type II: The received signal power of U1at the base sta-
tion is larger than τ(g), i,e. P1|h|2> τ(g). For this type,
the benchmark scheme [15] focuses U1’s signal to be de-
coded at the first stage of SIC and β= 1, and the data rate
is RII ,1= log 1 + P1|h|2
P0|g|2+1 . Differently, the proposed
scheme in this letter considers an additional strategy: for
P1|h|2> τ(g),U1transmits with a fraction of its power
P1, denoted by P′
1=βP1, such that P′
1|h|2=τ(g). As
a result, U1’s signal can still be decoded at the second
stage of SIC and the data rate is RII ,2= log(1 + τ(g)).
Hence, for P1|h|2> τ(g), the data rate of the proposed
scheme is RII = max{RII,1, RII ,2}, whereas that of the
benchmark scheme is only ¯
RII =RII,1.
Based on the above discussions, the achievable rate of the
proposed hybrid SIC NOMA scheme can be expressed as
follows:
R=RI, P1|h|2≤τ(g)
RII , P1|h|2> τ (g).(5)
From the above description, it can be easily seen that
the main difference between the proposed scheme and the
benchmark scheme [15] is that the proposed scheme can apply
optional power control to type II secondary user, while the
benchmark scheme does not.
Note that the premise of the proposed scheme is that U0
should achieve the same performance as in OMA. To achieve
this, U1can adopt adaptive rate transmission. Besides, as will
be shown later, U1can also transmit with a fixed rate in the
high SNR regime. The remainder of the paper will focus on
the performance achieved by U15.
III. OUTAG E ANALYSIS FOR THE PROP OS ED SCHEME
Given a target rate R1, the outage probability achieved by
U1is given by Pout
1=P(R < R1),which can be further
written as follows:
Pout
1=Pout
1,I +Pout
1,II ,(6)
where
Pout
1,I =PP1|h|2≤τ(g), RI< R1(7)
is the probability of the event that U1is a type I secondary
user and is in outage, and
Pout
1,II =PP1|h|2> τ (g), RI I < R1(8)
is the probability of the event that U1is a type II secondary
user and is in outage. Closed form expressions for Pout
1,I and
Pout
1,II are provided in the following two lemmas.
4Note that the data rate is the maximum rate of reliable communication
supported by the channel [17], which has the following two fold meanings.
When U1transmit with adaptive rate, the data rate is the achievable rate.
When U1transmit with a fixed rate, the data rate is the effective capacity.
5The proposed hybrid SIC scheme can be extended to serve multiple users
by applying hybrid NOMA. Specifically, in hybrid NOMA, users are divided
into multiple groups, each of which consists of two users. Different groups
are allocated with orthogonal resource blocks. In each group, the two users
are served by the proposed hybrid SIC scheme.
3
Lemma 1. Given a target rate R1, the probability of the event
that U1is a type I secondary user and is in outage can be
expressed as follows:
Pout
1,I =e−ϵ0
P0−e−ϵ1P0+ϵ0P1(ϵ1+1)
P0P1(9)
−e1
P1ϵ0P1
P0+ϵ0P1e−P0+ϵ0P1
P0P1−e−(P0+ϵ0P1)(ϵ1+1)
P0P1,
where ϵ1= 2R1−1.
Proof: Please refer to Appendix A.
Lemma 2. Given a target rate R1, the probability of the event
that U1is a type II secondary user and is in outage can be
expressed as follows:
Pout
1,II =1 −e−ϵ0
P0−e−ϵ1
P1P1
ϵ1P0+P11−e−(ϵ0P0+P1)ϵ0(ϵ1+1)
P0P1
(10)
+e1
P1ϵ0P1
P0+ϵ0P1e−P0+ϵ0P1
P0P1−e−(P0+ϵ0P1)(ϵ1+1)
P0P1.
Proof: Please refer to Appendix B.
The overall outage probability achieved by the secondary
user can be easily obtained by combing Pout
1,I and Pout
1,II , and
is provided in the following theorem.
Theorem 1. For a given target rate R1, the outage probability
achieved by U1in the proposed scheme is given by
Pout
1=1−e−ϵ1
P1P1
ϵ1P0+P11−e−ϵ0(ϵ1P0+P1)(ϵ1+1)
P0P1(11)
−e−ϵ1P0+ϵ0P1(ϵ1+1)
P0P1.
By applying Taylor seizes ex≈1−x(x→0), the
asymptotic outage performance achieved by the proposed
scheme in the high SNR regime can be obtained as in the
following corollary.
Corollary 1. In the high SNR regime, i.e., P0=P1→ ∞,
the outage probabilities Pout
1,I ,Pout
1,II and Pout
1can be approx-
imated as follows: Pout
1,I ≈ϵ1
P1, P out
1,II ≈0, P out
1≈ϵ1
P1.
Remark 2. Recall that, in the benchmark scheme [15], outage
probability error floors can be avoided only if the following
condition is satisfied: ϵ0ϵ1≤1. However, in the proposed
scheme, error floors can be avoided for arbitrary choices of
ϵ0and ϵ1, which is the main contribution of the proposed
scheme.
Remark 3. Consider the case where U1is served in OMA, i.e.,
a unique resource block, is allocated to U1. The use of OMA
means that the achieved outage probability of U1can be shown
as Pout
1,OM A =Plog(1 + P1|h|2)< R1. It is straightforward
to prove that Pout
1,OM A approaches ϵ1
P1, for P1→ ∞, which is
the same as the result shown in Corollary 1. In other words,
without occupying an additional resource block, the secondary
user in the proposed scheme achieves the same asymptotic
outage performance as if it is allocated with an additional
resource block.
Note that the outage probability can be used for not only
characterizing how likely a fixed rate cannot be satisfied, but
also further deriving the ergodic rate for the case with adaptive
rate transmission. When U1adopts adaptive rate transmission,
the ergodic rate achieved by U1, which is defined as
˜
R=Eg,h{R},(12)
can be expressed as follows [18, Appendix D]:
˜
R=1
ln 2 ∞
0
1−Pout
1
1 + ϵ1
dϵ1.(13)
IV. COMPARISON BETWEEN THE PROPO SE D SCHEME AND
BENCHMARK SCH EM E
In this section, a more detailed comparison between the
proposed scheme [15] and the benchmark scheme is provided.
It is straightforward that, for type I secondary user, the
proposed scheme achieves the same rate as the benchmark
scheme. However, for type II secondary user, it is possible
that the proposed scheme can achieve a higher instantaneous
rate (i.e., RII >¯
RII ) while consuming less power compared
to the benchmark scheme, as long as RII ,2> RII ,1. Thus, it
is interesting to investigate how likely the event RII >¯
RII
occurs, which is characterized in the following proposition.
Proposition 1. Given that U1is type II secondary user, the
conditional probability that RII >¯
RII is given by
Pbeat =PRII >¯
RII , U1is type II
P(U1is type II),(14)
where
PRII >¯
RII , U1is type II(15)
=e1
P1ϵ0P1
P0+ϵ0P1
e−P0+ϵ0P1
P0P1−√πϵ0P1
P0
e
1
p1+(P0−ϵ0P0+ϵ0P1)2
4ϵ0P1P2
0
×1
2−1
2erf √ϵ0
√P1
+P0−ϵ0P0+ϵ0P1
2P0√ϵ0P1,
erf(·)denotes the error function and
P(U1is type II) = 1 −e−ϵ0
P0+e1
P1ϵ0P1
P0+ϵ0P1
e−P0+ϵ0P1
P0P1.(16)
Proof: Please refer to Appendix C.
Please note that the superscript “beat” in (14) means that
the proposed scheme achieves a higher instantaneous rate than
the benchmark scheme (i.e., RII >¯
RII ), when U1is a type
II secondary user.
Remark 4. In the high SNR regime, when P0=P1→
∞, for type II secondary user, the proposed scheme al-
most surely outperforms the benchmark scheme [15], i.e.,
limP0=P1→∞ Pbeat = 1.
Remark 5. In the low SNR regime, when P0=P1→0,
for type II secondary user, the proposed scheme almost surely
achieves the same performance as the benchmark scheme [15],
i.e., limP0=P1→0Pbeat = 0.
4
0 10 20 30 40
SNR (in dB)
10-5
10-4
10-3
10-2
10-1
100
Outage probabilities
proposed scheme, R0= 1 BPCU
proposed scheme, R0= 4 BPCU
benchmark scheme, R0= 1 BPCU
benchmark scheme, R0= 4 BPCU
solid lines: sim
dash-dotted lines: ana
dashed line: approximation
(a) R1= 1 BPCU
0 10 20 30 40
SNR (in dB)
10-4
10-3
10-2
10-1
100
Outage probabilities
proposed scheme, R1= 1 BPCU
proposed scheme, R1= 4 BPCU
benchmark scheme, R1= 1 BPCU
benchmark scheme, R1= 4 BPCU
solid lines: sim
dash-dotted lines: ana
dashed lines: approximation
(b) R0= 1 BPCU
Fig. 1: Outage probabilities achieved by the proposed scheme
and the benchmark scheme [15].
The advantage of the proposed scheme over the benchmark
scheme [15] can be straightforwardly demonstrated by the
following example.
Example 1. When |g|2= 10,|h|2= 10,ϵ0= 2,P0=P1=
1. The threshold is chosen as τ(g)=4, In the benchmark
scheme [15], the achievable rate of U1is ¯
RII =RII,1=
log 1 + 10
11 . While in the proposed scheme, by costing only
2
5P1,U1’s data rate is RII =RII ,2= log(1 + 4), which is
much larger than that of the benchmark scheme.
V. NUMERICAL RE SU LTS
In this section, numerical results are presented to demon-
strate the performance achieved by the proposed scheme
and also verify the accuracy of the developed performance
analysis.
Fig. 1 shows the outage probabilities achieved by the
proposed and benchmark schemes. As shown in Fig. 1, the
accuracy of Theorem 1and the approximations at high SNR
in Corollary 1 can be verified. The two figures in Fig. 1
show that the proposed scheme achieves much lower outage
probabilities compared to the benchmark scheme, particularly
at high SNR. Moreover, it is noteworthy that the advantage
of the proposed scheme is more significant in the cases with
ϵ0ϵ1>1. As shown in the figure, the benchmark scheme
avoids error floors only in the case where R0=R1= 1
BPCU, whereas the proposed scheme avoids error floors in
all cases. Another interesting observation in Fig. 1(a) is that
0 10 20 30 40 50
SNR in dB (P0=P1)
0.4
0.5
0.6
0.7
0.8
0.9
1
P(U1is a type II user)
R
0
= 1 BPCU
R
0
= 2 BPCU
R
0
= 4 BPCU
R
0
= 8 BPCU
solid lines: sim
dashed lines: ana
Fig. 2: The probability of that U1is a type II secondary user
0 10 20 30 40 50
SNR in dB (P0=P1)
0
0.2
0.4
0.6
0.8
1
Pbeat
R
0
= 1 BPCU
R
0
= 2 BPCU
R
0
= 4 BPCU
R
0
= 8 BPCU
solid lines: sim
dashed lines: ana
Fig. 3: Pbeat when U1is a type II secondary user
the curves with different choices of R0coincide in the high
SNR regime. This observation is consistent with Corollary 1,
where the impact of R0on Pout
1diminishes in the considered
high SNR regime.
Fig. 2 shows the probability of that U1is a type II secondary
user, and Fig. 3 shows the probability that the proposed
scheme achieves higher instantaneous rates compared to the
benchmark scheme when U1is a type II secondary user.
Simulation results perfectly match the analytical results, which
verifies the accuracy of Proposition 1. As shown in Fig. 2,
P(U1is a type II user)decreases with transmit SNR, while
increases with the target data rate of the primary user U0. As
shown in Fig. 3, Pbeat begins with a very small value at low
SNR, then increases with the SNR, and finally approaches 1
at high SNR, which is consistent with the conclusions made
below Proposition 1.
Fig. 4 shows the power consumption of the proposed
scheme. In the benchmark scheme [15], U1transmits with full
power P1, whereas in the proposed scheme, U1can transmit
with βP1(0< β ≤1). As shown in the figure, in most
of the cases, βis less than 1, which means that the proposed
scheme consumes less power than the benchmark scheme. For
example, when SNR is 20 dBm and R0= 4 BPCU, only half
of P1is consumed by the proposed scheme. Instead, as shown
in Fig. 1, the outage probability achieved by the proposed
scheme is much lower than the benchmark scheme.
Fig. 5 shows the comparison between the proposed and
benchmark schemes in terms of ergodic rate. As shown in
5
0 10 20 30 40
SNR (in dB)
0
0.2
0.4
0.6
0.8
1
β
R
0
= 1 BPCU
R
0
= 2 BPCU
R
0
= 4 BPCU
R
0
= 8 BPCU
Fig. 4: Power consumption of the proposed scheme.
0 10 20 30 40 50
SNR in dB (P
0
=P
1
)
0
2
4
6
8
10
12
14
16
Ergodic rates (BPCU)
OMA
proposed scheme, R0=0.1 BPCU
proposed scheme, R0=0.5 BPCU
proposed scheme, R0=2 BPCU
proposed scheme, R0=4 BPCU
dashed lines:
benchmark scheme
Fig. 5: Ergodic rates achieved by the proposed scheme and
the benchmark scheme [15].
the figure, when R0= 0.1BPCU, the proposed scheme
achieves almost the same ergodic rates as OMA, where U1
is allocated with an additional resource block. At low SNR,
the proposed scheme performs similarly to the benchmark
scheme in terms of ergodic rates. In contrast, at high SNR,
the proposed scheme achieves larger ergodic rates than the
benchmark scheme. However, it is noteworthy that the gap
between the two schemes is insignificant when R0is very
small, but the gap becomes significant as R0increases.
VI. CONCLUSION
In this letter, a novel hybrid SIC based uplink NOMA
scheme has been proposed to opportunistically serve an sec-
ondary user without degrading the primary user’s QoS. It has
been shown that the proposed scheme can avoid an outage
probability error floor at high SNR, regardless of the choice
of the target rates, which has not been realized by the existing
SIC strategies.
APPENDIX A
PROO F FO R LEMMA 1
Pout
1,I can be calculated as follows:
Pout
1,I =PP1|h|2≤τ(g), P1|h|2< ϵ1(17)
=PP0|g|2
ϵ0−1>0, P1|h|2<min P0|g|2
ϵ0−1, ϵ1
=P0<P0|g|2
ϵ0−1< ϵ1, P1|h|2<P0|g|2
ϵ0−1
+PP0|g|2
ϵ0−1≥ϵ1, P1|h|2< ϵ1
=ϵ0(ϵ1+1)
P0
ϵ0
P0P0y
ϵ0P1−1
P1
0
e−xe−ydxdy
+∞
ϵ0(ϵ1+1)
P0ϵ1
P1
0
e−xe−ydxdy,
After some integration manipulations, a closed-form expres-
sion for Pout
1,I can be obtained and the proof is complete.
APP EN DI X B
PROO F FO R LEM MA 2
Pout
1,II can be calculated as follows:
Pout
1,II =PP1|g|2> τ (g),max P1|h|2
P0|g|2+ 1, τ (g)< ϵ1
(18)
=Pτ(g)< P1|h|2< ϵ1P0|g|2+ϵ1, τ (g)< ϵ1
(a)
=P0< P1|h|2< ϵ1P0|g|2+ϵ1, g ≤ϵ0
P0
+PP0|g|2
ϵ0−1< P1|h|2< ϵ1P0|g|2+ϵ1,
P0|g|2
ϵ0−1< ϵ1, g > ϵ0
P0
=ϵ0
P0
0ϵ1P0y
P1+ϵ1
P1
0
e−xe−ydxdy
+ϵ0(ϵ1+1)
P0
ϵ0
P0ϵ1P0y
P1+ϵ1
P1
P0y
ϵ0P1−1
P1
e−xe−ydxdy,
where step (a) is obtained by dividing the event into two cases,
one is τ(g) = 0 and the other is τ(g)>0. Finally, after
some integration manipulations, the expression for Pout
1,II can
be obtained and the proof is complete.
APP EN DI X C
PROO F FO R PROPOSITION 1
The numerator in (14) can calculated as follows:
PRII >¯
RII , U1is type II(19)
=Pτ(g)>0, P1|h|2> τ(g),P1|h|2
P0|g|2+ 1 < τ(g)
=PP0|g|2−ϵ0
ϵ0P1
<h<P2
0|g|4+P0(1−ϵ0
)|g|2−ϵ0
ϵ0P1
, g > ϵ0
P0
=∞
ϵ0
P0
P2
0y2
+P0(1−ϵ0
)y−ϵ0
ϵ0P1
P0y−ϵ0
ϵ0P1
e−xe−ydxdy.
The denominator in (14) can be calculated as
P(U1is type II)(20)
=PP1|h|2> τ(g)
=P(τ(g) = 0) + PP1|h|2> τ (g), τ (g)>0
=ϵ0
P0
0
e−xdx +∞
ϵ0
P0∞
P0y
ϵ0P1−1
P1
e−xe−ydxdy.
After some integration manipulations, the expressions (15) and
(16) can be obtained and the proof is complete.
6
REF ER EN CE S
[1] 3GPP TR36.859 V13.0.0, “Study on Downlink Multiuser Superposition
Transmission (MUST) for LTE (Release 13),” Dec. 2015.
[2] 3GPP, TR38.812 V16.0.0, “Study on Non-Orthogonal Multiple Access
(NOMA) for NR,” Dec. 2018.
[3] B. Makki, K. Chitti, A. Behravan, and M.-S. Alouini, “A survey of
NOMA: Current status and open research challenges,” IEEE Open
Journal of the Commun. Society, vol. 1, pp. 179–189, 2020.
[4] Y. Fu, M. Zhang, L. Sala¨
un, C. W. Sung, and C. S. Chen, “Zero-Forcing
Oriented Power Minimization for Multi-Cell MISO-NOMA Systems: A
Joint User Grouping, Beamforming, and Power Control Perspective,”
IEEE J. Select. Areas Commun., vol. 38, no. 8, pp. 1925–1940, Jun.
2020.
[5] H. Wang, Y. Fu, R. Song, Z. Shi, and X. Sun, “Power minimization
precoding in uplink multi-antenna NOMA systems with jamming,” IEEE
Trans. Green Commun. Netw., vol. 3, no. 3, pp. 591–602, May 2019.
[6] K. Higuchi and Y. Kishiyama, “Non-orthogonal access with random
beamforming and intra-beam SIC for cellular MIMO downlink,” in IEEE
Veh. Tech. Conf.,, Las Vegas, NV, US, Sep. 2013, pp. 1–5.
[7] Y. Gao, B. Xia, K. Xiao, Z. Chen, X. Li, and S. Zhang, “Theoretical
analysis of the dynamic decode ordering sic receiver for uplink noma
systems,” IEEE Commun. Lett., vol. 21, no. 10, pp. 2246–2249, Jun.
2017.
[8] N. Zhang, J. Wang, G. Kang, and Y. Liu, “Uplink Nonorthogonal
Multiple Access in 5G Systems,” IEEE Commun. Lett., vol. 20, no. 3,
pp. 458–461, Jan. 2016.
[9] B. Xia, J. Wang, K. Xiao, Y. Gao, Y. Yao, and S. Ma, “Outage
Performance Analysis for the Advanced SIC Receiver in Wireless
NOMA Systems,” IEEE Trans. Veh. Technol., vol. 67, no. 7, pp. 6711–
6715, Mar. 2018.
[10] M. Chen and A. Burr, “Multiuser detection for uplink non-orthogonal
multiple access system,” IET Communications, vol. 13, no. 19, pp. 3222–
3228, Dec. 2019.
[11] F. Zhou, Y. Wu, Y.-C. Liang, Z. Li, Y. Wang, and K.-K. Wong, “State
of the art, taxonomy, and open issues on cognitive radio networks with
NOMA,” IEEE Wireless Commun., vol. 25, no. 2, pp. 100–108, 2018.
[12] S. Dhakal, P. A. Martin, and P. J. Smith, “NOMA With Guaranteed
Weak User QoS: Design and Analysis,” IEEE Access, vol. 7, pp. 32884–
32 896, Mar. 2019.
[13] Z. Ding, R. Schober, P. Fan, and H. V. Poor, “Simple Semi-Grant-Free
Transmission Strategies Assisted by Non-Orthogonal Multiple Access,”
IEEE Trans. Commun., vol. 67, no. 6, pp. 4464–4478, Mar. 2019.
[14] Z. Ding, R. Schober, and H. V. Poor, “Unveiling the Importance of SIC
in NOMA Systems: Part I–State of the Art and Recent Findings,” IEEE
Commun. Lett., (submitted) Available online at arXiv:2005.10215.
[15] ——, “A new QoS-guarantee strategy for NOMA assisted semi-grant-
free transmission,” IEEE trans. Wireless Commun., submitted) Available
online at arXiv:2004.12997.
[16] ——, “Unveiling the Importance of SIC in NOMA Systems: Part II:
New Results and Future Directions,” IEEE Commun. Lett., (submitted)
Available online at arXiv:2005.10217.
[17] D. Tse and P. Viswanath, Fundamentals of wireless communication.
Cambridge university press, 2005.
[18] Y. Sun, Z. Ding, X. Dai, and O. A. Dobre, “On the performance
of network NOMA in uplink CoMP systems: a stochastic geometry
approach,” IEEE Trans. Commun., vol. 67, no. 7, pp. 5084–5098, Jul.
2019.