You are on page 1of 11

Alexandria Engineering Journal (2023) 64, 667–677

H O S T E D BY
Alexandria University

Alexandria Engineering Journal


www.elsevier.com/locate/aej
www.sciencedirect.com

ORIGINAL ARTICLE

Uplink power control scheme for spectral efficiency


maximization in NOMA systems
Bilal Ur Rehman a,*, Mohammad Inayatullah Babar a, Gamil Abdel Azim b,
Muhammad Amir a, Hesham Alhumyani c, Mohammed S. Alzaidi d,
Majid Alshammari e, Rashid Saeed f

a
Department of Electrical Engineering, Faculty of Electrical and Computer Engineering, University of Engineering and Technology,
Peshawar, Pakistan
b
Department of Computer Science, Faculty of Computer Science, Canal Suez University, Egypt
c
Department of Computer Engineering, College of Computers and Information Technology, Taif University, P.O. Box 11099,
Taif 21944, Saudi Arabia
d
Department of Electrical Engineering, College of Engineering, Taif University, Taif 21944, Saudi Arabia
e
Department of Information Technology, College of Computers and Information Technology, Taif University, P.O. Box 11099,
Taif 21944, Saudi Arabia
f
Department of Computer Engineering, Taif University, Saudi Arabia

Received 6 July 2022; revised 3 November 2022; accepted 19 November 2022


Available online 14 December 2022

KEYWORDS Abstract Non-orthogonal multiple access (NOMA) has recently gained much research attention
5G; as a possible multiple access solution for fifth-generation (5G) wireless communication. The NOMA
Uplink; can further enhance spectral performance compared with orthogonal multiple access (OMA). This
Non-orthogonal multiple paper focuses on the joint optimization problem of power control and user grouping by considering
access (NOMA); the uplink NOMA system. The power control issues are solved using a linear programming
User grouping; approach and three alternative algorithms solve user grouping: Randomly, 2-Opt, and the Hybrid.
Power control The numerical results show that the proposed solution for power control and user-grouping
achieves significant spectral performance compared to OMA and existing schemes reported in
the literature with lower system complexity for uplink NOMA systems.
Ó 2022 THE AUTHORS. Published by Elsevier BV on behalf of Faculty of Engineering, Alexandria
University. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/
licenses/by-nc-nd/4.0/).

1. Introduction

In recent years, non-orthogonal multiple access (NOMA) has


* Corresponding author. emerged as a viable method for obtaining maximum spectral
E-mail address: bur@uetpeshawar.edu.pk (B.U. Rehman). efficiency. Hence, the NOMA method has been employed in
q
Research Article fifth-generation (5G) and beyond 5G (B5G) wireless
https://doi.org/10.1016/j.aej.2022.11.030
1110-0168 Ó 2022 THE AUTHORS. Published by Elsevier BV on behalf of Faculty of Engineering, Alexandria University.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
668 B.U. Rehman et al.

communication networks [1–6]. The core principle of NOMA The importance of power-control and user-grouping for the
is to employ the power domain (PD) multiplexing approach sum rates of paired users in NOMA uplink is comprehensively
to effectively utilize the spectrum among multiple users sharing covered in [26]. Also, a few studies [27–29] have discussed the
the same Resource Block (RB). For this reason, the NOMA implementation of NOMA in the multi-cell scenario for sum-
technique differs significantly from the classic orthogonal mul- rate maximization in the uplink case. Furthermore, the itera-
tiple access (OMA) approach involving time/frequency/code tive water-filling algorithm and Dinklebach algorithm are
domain multiplexing. The NOMA scheme has shown a higher employed in [30,31] respectively, to obtain a sub-optimal
ergodic sum rate even though the outage output is strongly power allocation solution for a multi-user with an efficient
dependent on user requirement and resource allocation [7] as clustering approach regarding the NOMA uplink scheme.
compared to the OMA scheme. Also, the NOMA can simulta- Moreover, some authors [32,33] have proposed power alloca-
neously handle multiple users having different channel condi- tion approaches to increase the energy efficiency for NOMA
tions, which leads to massive connectivity, improved user in an uplink scenario. Besides that, a recent study [34] has also
fairness, and lower latency than OMA [8]. Furthermore, the investigated the trade-off between spectral- efficiency and
NOMA approach is more compatible with current and next- energy efficiency for a multi-cluster, multi-user NOMA uplink.
generation communication systems since no significant modifi- In addition, a few game theory-based power control schemes
cations are required in its current architecture. Therefore, [35,36] have been presented recently for NOMA uplink that
numerous standardized procedures have recently adopted the achieved relatively high spectral efficiency, energy efficiency,
NOMA techniques, such as the layered division multiplexing and network efficiency to address NOMA’s user-clustering
(LDM) technique employed in digital television and the and power allocation issues (downlink and uplink) jointly.
multi-user superposition transmission (MUST) technique For the NOMA uplink scheme, Ali and fellows [37] came up
implemented by third-generation partnership project long- with a low-complexity algorithm to obtain closed-form solu-
term evolution (3GPP-LTE) [9]. tions systems and achieved noteworthy results. In [38], the
Effectual user-grouping and power control are essential in authors addressed the optimal user clustering and power-
achieving a high sum rate and improved user fairness in the control problem for the multi-input single-output multiple-
NOMA systems. In some NOMA studies, user pairing is done carrier NOMA uplink approach for maximizing the through-
such that a user in the center with a more robust channel- gain put of the weighted system using monotonic optimization.
is paired with a user at the edge having a low channel gain [10]. Zhang et al. [39] studied the joint problem of the user-
However, such user pairing can result in mid-user pairing grouping issue with power-control and user decoding order
issues between edge and center users in a network. Hence, for NOMA in an uplink scenario to intensify the multiple
some techniques are investigated in [11] to address this issue. users’ achievable sum rate (ASR). Initially, an optimal solution
Another study [12] provided a systematic overview to shed for both power control and user decoding order is obtained in
light on the complexities related to users’ grouping and power closed-form. After that, the optimal and sub-optimal user-
control for meeting reliability constraints and user target rates clustering solution is derived with varied computational com-
in NOMA systems. The users’ grouping and power control are plexity. To maximize system throughput and reduce IoT
currently more explored for downlink NOMA systems [13–17] (Internet of Things) application latency, the resource alloca-
whereas limited literature has considered the case of uplink tion problem is formulated. The power allocation was also
NOMA. studied to ensure fairness among users [40]. Also, meta-
For uplink NOMA, a sub-optimal power allocation solu- heuristic algorithms are implemented to solve the optimization
tion based on single-user water-filling (SUWF) [18] is proposed problem in [41] for uplink NOMA systems to maximize the
for maximizing the system throughput by taking a minimum- network’s spectral performance. Furthermore, the sub-
rate requirement and maximum transmit power constraints optimal solution has also been obtained for the user-
into account. A power back-off scheme [19] for uplink is then grouping problem [42] to attain high spectral performance.
introduced to attain high spectral efficiency by distinguishing In [43], a dynamic power allocation mechanism is presented
the users having similar signal strength within a cluster. An to improve the sum rate for uplink NOMA transmission,
optimal user-pairing approach is proposed in [20] with a fixed and a comparison to the fixed power allocation mechanism
power-control scheme. Also, a cumulative distribution func- is also provided. Further, the user pairing and power alloca-
tion (CDF) based fair-resource scheduling is suggested in tion scheme are studied to maximize the sum rate at layer-1
[21] for NOMA uplink, where the highest CDF value determi- and layer-2 for two users uplink NOMA transmission in
nes the users’ selection within every time slot. Another study [44]. In addition, some recent studies [45–47] also considered
[22] introduced a dynamic power control method for down- the uplink scenarios and joint optimization problems for
linking NOMA and uplinking NOMA for achieving a higher sum-rate maximization.
spectral performance due to a compromise between user- Uplink NOMA has received much less attention than
fairness and throughput. Moreover, Power Controlled downlink NOMA. However, the research in this area is still
System-wide Utility Maximization (PCSUM) algorithms [23] in its infancy, and more studies are required to effectively
are introduced for addressing cell-association problems and address the joint power-control issue and user-pairing/
power-allocation jointly in NOMA uplink. Zhai and Du [24] grouping for the NOMA scheme in an uplink scenario. This
utilized the maximum weight independent set (MWIS) algo- paper attempts to fill this research gap by proposing an effi-
rithm for obtaining optimum power levels for sum-rate maxi- cient approach that optimizes user-pairing/grouping and
mization. Another joint approach [25] is presented regarding power control based on the user’s minimum transmission rate
payload power-control and pilot that effectively alleviated requirement in the NOMA uplink system for maximizing
the error-propagation while enhancing the data detection achievable spectral efficiency. An analytical framework for
performance. power-control problems and user-grouping are examined
Uplink power control scheme for spectral efficiency maximization in NOMA systems 669

jointly by considering an uplink scenario. Furthermore, this 3. Fundamentals of uplink NOMA transmission
research investigates the power control issue by employing a
linear- programming approach. In addition, the user- This section covers the concept of two-users uplink NOMA
grouping problem is solved using three algorithms, namely transmission with a single base station (BS). The NOMA
the 2-Opt, randomly, and the hybrid. Our proposed scheme transmission in the uplink scenario is shown in Fig. 1, where
is compared with OMA and current schemes addressed in the power-domain multiplexing is performed in an almost dif-
the literature for the uplink NOMA. Simulation results ferent manner from downlink NOMA transmission as the BS
showed a noticeable increase in spectral performance with employed super-position coding to accomplish power domain
lower system complexity for uplink NOMA transmission. multiplexing in downlink NOMA. However, users’ broadcast
The following are the key contributions of this article: power is solely limited by their battery capacity in the uplink.
Therefore, both users can broadcast with their full power.
 We formulate an optimization problem of power control as The simultaneous transmission of users 1 and 2 to their cor-
a linear programming problem, and user grouping is based responding signals denoted by s1 (user-1) and s2 (user-2) to their
on the users’ minimum rate requirement; as a result, it max- desired BS. The signals of both users received yz [2] at the BS
imizes the network’s spectral performance. can be represented as:
 We also studied and employed a user-grouping approach to
grouping the users efficiently with no constraint on the X
2 pffiffiffiffiffi

number of users allocated to the same PRB/group. Further- yz ¼ Pi gi si þ nz ð1Þ


i¼1
more, we proposed two cases; the first is the three algo-
rithms using the linear programming approach to solve where nz denotes the Gaussian noise and w represents the
the power control problem and the second is the same using power spectral-density. The signal of the users i and transmis-
the power control solution [39]. sion power is represented by si with Ejðsi Þ2 j ¼ 1 and Pi , where
i ¼ 1; 2.
This research study is divided into different sections as The signal s1 is nearer to the BS and regarded as the stron-
follows: The recent work and the fundamental concepts gest, experiencing greater channel-gain, whereas the signal s2 is
regarding the NOMA scheme, particularly for the uplink the weak user that is far away from the BS, experiencing lower-
case, are presented in Section 2 and Section 3. The system channel gain. The SIC is performed to decode the user’s signal/
model for uplink NOMA transmission is provided in Sec- information at the BS in uplink NOMA transmission. Hence.
tion 4. Section 5 formulate the spectral-efficiency maximiza- the BS decode the signal s1 firstly by assuming the signal s2 as
tion issue for uplink NOMA systems. The proposed solution interference [3]. For signal s2 , subtract the first signal s1 , which
is provided in Section 6, where the power control issue is is decoded already, and then decode the signal s2 . Therefore, in
solved through a linear programming approach and the uplink case, signal s1 is experiencing interference from signal
user-pairing/grouping issue is solved with three alternative s2 , while signal s2 is interference-free transmission since signal
algorithms. The numerical results are provided in Section 7 s1 is decoded before signal s2 . In case of downlink NOMA
to evaluate the performance of the proposed approach. transmission, the operation of the decoding is different as a
Finally, the conclusion of this research study is drawn in result the signal s2 experience interference from signal s1 ,
the Section 8. whereas signal s1 is interference-free transmission, since signal
s2 is decoded prior to signal s1 .
2. Related work For perfect operation of SIC at the BS, the attainable data-
rates represented as Ri , where i ¼ 1; 2 for a given transmission
bandwidth of 1 Hz is as follows:
NOMA was initially studied for uplink-transmissions in [48],
!
where the power-control method was used at the transmitter P1 jg1 j2
(user) and minimum mean squared error (MMSE)-based SIC R1 ¼ log2 1 þ ð2Þ
P2 jg2 j2 þ w
operation was used at the receiver (base station). In [18], a
joint problem of sub-carrier scheme and power-allocation !
was investigated. Mainly, a sub-optimal approach was P2 jg2 j2
obtained to enhance the spectral-performance of NOMA R2 ¼ log2 1þ ð3Þ
w
users. For two-user NOMA transmission uplink, closed-
form equations for sum-throughput and outage probability It should be noted that, for an effective uplink NOMA scheme,
with static powers [19] were developed. Furthermore, the the channel-gains of users must be different sufficiently. If both
author compares their results to those of a time division mul- users have comparable channel-gains, the BS will be unable to
tiple access (TDMA) based OMA systems and found that differentiate their signals in the power domain. In that scenar-
any NOMA user in a network is always at outage if the tar- io, an efficient power control scheme is required. That is, the
get sum-rates are not properly selected. Hence, the conclusion users must transmit with differing extents of power. Another
was reached in [7] for NOMA downlink transmission. Aside key observation from the preceding explanation is that the
from previous works, the robust user-scheduling method was SIC order is reversed in the NOMA uplink scenario. There-
proposed in [49] for NOMA transmissions in an uplink sce- fore, the signal/information of the closest user is first decoded,
nario, where all users in a network have distinct channel- whereas, in downlink NOMA, the signal/information of the
gain and by using single-carrier frequency division multiple distant user is decoded first. In addition, BS can often be pro-
access (SC-FDMA) were leveraged to achieve efficient user- vided with adequate processing in terms of energy and power.
grouping. Hence, uplink NOMA [2] can accommodate a more consider-
670 B.U. Rehman et al.

Fig. 1 Two users uplink NOMA transmission.

able number of users than downlink NOMA. Furthermore, the ment. In the case of NOMA, the users belonging to a similar
BS can utilize more computationally complex decoding algo- PRB/group can transmit their PRB, whereas the users belong-
rithms to minimize the impact of interferences on the users. ing to a distinct group can transmit different PRB in OMA. As
For that reason, NOMA in the uplink case is preferable to a result, different N signals can be received yk at the BS is
NOMA in the downlink case for large machine-type communi- mathematical as follows:
cation [2].
X
N
yk ¼ Hk;n sn þ nk ð4Þ
4. System model for uplink NOMA n¼1

Hk;n ¼ Uk;n gn aen ð5Þ


Let us consider a NOMA based transmission for the uplink
scenario, which consists of a single cell represented by C, where pffiffiffiffiffiffiffiffi  pffiffiffiffi
where aen ¼ an P; an 2 ½0; 1 and aen 2 0; P . The power con-
C ¼ 1. In C, a base station (BS) is positioned at the centre and trol coefficient and total transmission power is denoted by an
serves to connect with N distinct NOMA users as shown in and P for all N in a network.
Fig. 2. We assume that BS and all users N are implemented  The signal from N is indicated
with a single-antenna. Further, a BS is fully aware of channel as sn ð1 6 n 6 NÞ, where E jsn j2 ¼ 1. The response of the
state information (CSI) and serves all NOMA users N within channel is Gaussian which is distributed between N and BS
the coverage area. The number of physical resources blocks and represented by gn , whereas, white Gaussian noise with
(PRBs) in-terms of time/frequency/code is denoted by K, an average power is described by nk and r2 in the k  th group.
where K < N. Hence, the allocation of N is arranged in K such The representation of N assigned to the K  th group is
that each N in a network fully maintains the data-rate require- denoted by Uk;n and mathematically described as:

1; iftheusern 2 groupk
Uk;n ¼ ð6Þ
0; otherwise
Therefore,

gn aen ; if the usern 2 groupk
Hk;n ¼ ð7Þ
0; otherwise
To perform the decoding operation, the BS applies Successive
Interference Cancellation (SIC) for each PRB to acquire the
desired signal/information of the users belonging to the same
group. For user n, the decoding order is represented by Dk;n
in the k  th group/PRB. Consider that Dk;n ¼ a > 0 illustrate
that for each n to be decoded is the a  th one in k  th group.
Hence, for each n, the mathematical representation for
spectral-efficiency can be represented as:
0 1
B C
B C
B C
B C
B jgn j2 an c C
Rn ¼ log2 B
B1 þ
C
C ð8Þ
B XN
C
B jgj j aj c þ 1C
2
B C
@ j–n A
dk;j > dk;n > 0

where, transmission power to noise-ratio is denoted by c and


Fig. 2 Uplink NOMA transmission. c ¼ P=r2 . We consider any two users n and j in a cell for
Uplink power control scheme for spectral efficiency maximization in NOMA systems 671

decoding order. The decoding of user n is prior to user j in a Dk;n 2 h; fan gRt ðkÞ ð12aÞ
group, which is represented as Dk;j > Dk;n . It is assumed that
the BS has more processing power and is fully aware [2,3] Rn P rn ; n 2 Nk ð12bÞ
i.e. channel-state-information (CSI) to decode effectively the 0 6 an 6 1; n 2 Nk ð12cÞ
signals/information. Moreover, each user n in a network
should be transmitted an and Uk;n to address the power- where Nk , indicates the set of all possible combinations in the
control issue and user-grouping. Therefore, for each user n k  th PRB.
belongs to the k  th group, the mathematical representation
of spectral-efficiency is given as: 6.1. Decoding order approach for uplink NOMA
X
Rðt kÞ ¼ Rn ð9Þ
Uk;n ¼1 In order to apply SIC operation, all users’ signals/information
is decoded by the receiver in descending order based on
0 1 channel-condition. For example, in the NOMA [37] uplink sce-
X
Rðt kÞ ¼ log2 @1 þ jgn j2 an cA ð10Þ nario, users with better channel conditions (strong-user) are
Uk;n ¼1 first decoded at the base station. In contrast, a user with poor
channel condition (weak-user) is decoded at the base station
The above Eqs. (9) and (10) show that the spectral perfor- last. For this reason, the user with a better channel condition
mance of individual user n in a network is affected by the (strong user) experiences interference from other users in the
decoding order while there is no influence on each group.. network, whereas users with poor channel condition (weak
user) experience zero interference.
5. Optimization problem formulation For optimal decoding order [39,42], the users N belong to
same PRB/group are based on the value of On . The decoding
In this research, we jointly formulated the power-control prob- order of each user in a network depends on power control
lem, user-grouping and user’s decoding-order under their min- [39,42] scheme result in different feasible regions can be as
imum required constraint to maximize the spectral- follows:
performance. For each n in a network, the minimum rate 1
requirement is denoted by rn . Hence, the spectral- On ¼ jgn j2 1 þ ð13Þ
wn
performance maximization problem [20,39,42] can be formu-
lated as: where

  X
K wn ¼ 2rn  1 ð14Þ
Uk;n ; Dk;n 2 h; fan gRt ¼ Rðt kÞ ð11aÞ
k¼1
Hence, an Eqs. 13 and 14 illustrates that any user n in a net-
C1 : 0 6 an 6 1; 8n ð11bÞ work is decoded first according to the highest On [39,42].
C2 : Rn P rn ; 8n ð11cÞ
6.2. Uplink power control as a linear programming approach
C3 : Uk;n 2 f0; 1g; 8k; 8n ð11dÞ
XK
We solve the power control issue as an optimization by consid-
C4 : Uk;n ¼ 1; 8n ð11eÞ
ering a linear programming approach. The power control opti-
k¼1
mization problem is represented as follows:
where Dk;n represents the decoding order and h indicates all P1 :
possible combinations of users’ decoding orders. C1 defines
the transmission power considering the upper bound of all user  X M
aj jgj j2 aj c ð15aÞ
n in a network. C2 represents the achievable minimum rate- j¼1
requirement. Finally, C3 and C4 denote the user indicator
and ensure that user n are allocated to the group. C1 : 0 6 aj 6 1; 8j ð15bÞ
!
X
M
6. Spectral efficiency maximization for uplink NOMA systems C2 : jgj j2 aj c P wj jgk j2 ak c þ 1 ; 8j ð15cÞ
k¼jþ1

The optimization variables an ; Uk;n , and Dk;n are intimately 


where, aj is the power control coefficient and
linked to find the best approach for the problem discussed in
j 2 f1; 2; 3; . . . c; Mg. The Eq. (15a) and Eq. (15c) shows linear-
Eq. (11a), making the problem complicated. Moreover, the
ity and represented in-terms of SNR formulations. Therefore,
user-grouping Dk;n variables are combinatorial-integer. So,
an Eq. (15c) can be expressed as follows:
first, solve the combinatorial issue of the power-control prob- !
lem, and the users’ decoding order afterward apply an appro- 2
XM
2
priate approach for user-grouping. In the case of any fixed jgj j aj c  wj jgk j ak c þ 1 P 0 ð16Þ
k¼jþ1
scheme of user-grouping, the value of Uk;n is un-related among
all distinct groups regarding both decoding order and power Hence, the Eq. (16) can be represented in matrix form as
control. follows:
672 B.U. Rehman et al.
2 3 2 3
jg1 j2 w1 jg2 j2 w1 jg3 j2       w1 jgm j2 2 a 3 2 w 3 a1
6 7 1 1 6 . 7
6 0 jg2 j 2 2
w2 jg3 j       w2 jgm j2 76 7 6 7 6 .. 7 2 3
6 76 a2 7 6 w2 7 6 7 1
6 76 7 6 7 6 7
6 .. 76 7 6 7 1 0 1 0 6 aM 7 6 . 7
6 . 0 jg3 j2   w3 jgm j2 76a 7 6w 7 6 7 ¼ 4 .. 5 ð26Þ
6 76 3 7 6 3 7 0 1 0 1 6 7
c6
6 .. ..
76 . 7 P 6 . 7
76 . 7 6 . 7
6 t1 7
6 . 7
6 6 . 7 6 . 7 1
6 . . 0    7 76 7 6 7
6 . 7
4 . 5
6 76 7 6 7
6 .. .. .. .. .. .. 76 ... 7 6 ... 7
6 . . . . . . 74 5 4 5 tM
4 5
.. .. .. am wm Hence, Eqs. (25) and (26) can be represented as
. . .   jgm j2
 
ð17Þ ½ A I I ½ x  ¼ b1 ð27Þ

8 with
>
< 0 i>j 8
>
< 0 i>j
Set ai;j ¼ jgi j2 i¼j ð18Þ
>
: Set ai;j ¼ jgi j þ 1
2
i¼j ð28Þ
wi jgj j2 i<j >
:
wi jgj j2 i<j
Then, Eq. (17) can be expressed as
1 0
Aa P b ð19Þ and I ¼ ; b1i ¼ bi þ 1 8i 2 f1; 2; 3; . . . c; Mg.
0 1
1 0 0 1
a1 w1 =c Finally, the standard form of optimization problem P1 is
B. C B C
with a ¼ @ .. Aand b ¼ @ ... A,
expressed as:
P1 :
aM wM =c
Finally, the optimization problem P1 can be represented as: Ct x ð29aÞ
P1 : Bx ¼ y ð29bÞ
Ct x ð20aÞ xP0 ð29cÞ
Aa P b ð20bÞ where, x ¼ fa1 ; a2 ; . . . c; am ; S1 ; S2 ; . . . c; Sm ; t1 ; t2 ; . . . c; tm g;
 
06a61 ð20cÞ 2
C ¼ jgi j c; for i ¼ 1; 2; . . . c; M and Ci ¼ 0;for i ¼ M þ 1;
t

 
where, Ci ¼ jhi j2 c . . . . c; 3M.
In this way, we obtained the solution to the power control
The power control optimization problme P1 can be repre- problem using a linear programming approach.
sented in the standard form of linear programming. Hence,
Eq. (20b) can be expressed as follows: 6.3. User-pairing/grouping algorithms
X
M
ai;j aj P bi ; 8i 2 f1; 2; 3; . . . c; Mg ð21Þ
j¼1
The user-pairing/grouping approach is compatible with the
standard application requirements of the NOMA scheme,
where an Eq. (21) can be represented in standard form as: where multiple concurrent user transmissions are inefficient
X
M to apply to all users at the same time due to the additional sys-
ai;j aj  Si ¼ bi ; Si P 0 8i 2 f1; 2; 3; . . . c; Mg ð22Þ tem complexity. On the other hand, multiple user-pairing/
j¼1 grouping techniques are experiencing different system imple-
where, ai P 0 and ai 6 1; 8i 2 f1; 2; 3; . . . c; Mg, ti are slack mentation complexity and network environment. Moreover,
variables and si are suprplus variables, therefore an adequate user-grouping mechanism is one of the major con-
ai þ ti ¼ 1; 8i 2 f1; 2; 3; . . . c; Mg. Finally, the standard form cerns for operational NOMA, particularly in the uplink case.
of Eq. (20b) and (20c) as follows: The user-pairing/grouping schemes should be compatible with
the appropriate user’s decoding-order and power control
Aa  IS ¼ b ð23Þ scheme to ensure minimum-rate requirement (rn ) to attain high
system performance with lower computational-time coopera-
a tively [39,42]. Hence, an exhaustive-search approach is
½A I ¼ ½b ð24Þ
s employed for the NOMA uplink scenario to determine the best
3 2 solution for users’ grouping. Finally, the suggested algorithms
a1 (the randomly, 2-Opt, and the hybrid) are employed to gener-
6 . 7 ate a sub-optimal solution illustrated in Algorithms 1–3 for a
6 .. 7 2 3
6 7 b1 fixed user-pairing/grouping scheme. Hence, arrange users n
6 7
a11 a1M 1 0 6 aM 7 6 . 7
6 7¼6 . 7 ð25Þ according to Eqs. 13 and 14 to obtain the desired solution.
0 1 6 7 4 . 5
aM1 aMM 6 s1 7 For user-pairing/grouping problems, a feasible solution is pro-
6 . 7 bM
6 . 7 vided at the start of the algorithms.
4 . 5 The system-complexity of the randomly algorithm is
 
sM OðNKÞ, whereas hybrid and 2-Opt algorithms are O N2 [42].
Uplink power control scheme for spectral efficiency maximization in NOMA systems 673

Algorithm 1. Randomly Algorithm


Table 1 Uplink NOMA simulation parameters.
Parameter Value
Total Number of Cells ðCÞ 1
1: procedure INITIALIZE VARIABLES AND LIST ALL THE USERS WITH
Total Number of Users ðNÞ 10
DECREASING ORDER OF On Number of PRB/Group ðKÞ 3
Input: C; N; c; fgn g; fsn g. Minimum transmission rate requirement ðrn Þ 1:5bits=s=Hz
  
Output: Uk;n {optsolution ¼ so ; optmax f so } Transmission bandwidth ðBÞ 1Hz
2: for i ¼ 1: totalIterations do Transmission power to noise ratio ðcÞ 30dB
3: Generate a feasible solution so
4: for n ¼ 1: N do
5: Measure fðso Þ (According to Eq. (9) and (10) to find
ðk;nÞ
Rt , for k ¼ 1; 2; 3; . . . c; K) 7.1. Case 1
6: if ðfðso ÞÞ P optmax then
7: optmax ¼ fðso Þ
8: optsolution ¼ so The power control problem is based on the optimal solution of
the standard form of linear programming with the user-
grouping solution by considering Algorithms 1–3.
Fig. 3 compares spectral efficiency versus the number of
Algorithm 2. 2-Opt Algorithm iterations of the 2-Opt, randomly, and the hybrid algorithms.
It demonstrates that the spectral efficiency increased with
increasing iterations for all the algorithms. Furthermore, it is
clear from the results that the hybrid algorithm attained the
1: procedure Initialize variables and List all the users with maximum spectral efficiency well before the competing alter-
decreasing order of On . natives compared to the 2-Opt and randomly algorithms.
Input: C; N; c; fgn g; fsn g; randomsolutionðso Þ.
The Fig. 4 represent the case 1 with rn ¼ 1:1bits=s=Hz for
Output: Uk;n {optsolutionoptmax}
100 iterations. The suggested NOMA strategy outperforms
2: where a random solution so1 , for user i and user j, where i
and j are different and belong to different groups u and the OMA system in terms of spectral efficiency. Furthermore,
v; swapði; jÞ. The result is a new solution for user grouping so2 the proposed solution is nearer to the optimal-solution, indi-
3: for n ¼ 1: N do cating that the proposed solution for power-control problems
4: if ðfðso2 ÞÞ P ðfðso1 ÞÞ then and user-grouping may reach near optimal-solution with lower
5: optmax ¼ fðso2 Þ system-complexity. The attainments of benchmark-solutions
6: optsolution ¼ so2 [20] are also reported for NOMA uplink systems. NOMA
users’ spectral-efficiency is considerably lower with a fixed
power-control scheme [20] than with the intended NOMA sys-
tems. By imparting the suggested power-control strategy, the
Algorithm 3. Hybrid Algorithm spectral efficiency of NOMA in [20] is closer to the optimal-
solution. It is mentioned that the method in [20] has some lim-
itations due to which it is suited for two users only, whereas the
suggested user-grouping solution is applied to multiple users.
1: procedure Initialize variables and List all the users with
decreasing order of On
Input: C; N; c; fgn g; fsn g, opt solution in
  
Output: Uk;n {optsolution ¼ so ; optmax f so }
2: Applying Algorithm 2, where the initial solution is the opt
solution obtained by Algorithm 1.

7. Simulation results

The performance of the NOMA uplink scenario using the pro-


posed power-control strategy and user-grouping algorithms is
evaluated in this section. For the simulation process, the
MATLAB tool is used to determine the outcome of the pro-
posed solution. The parameters attained from the literature
[20,39,42] used in the simulation are provided in Table 1.
The user’s channel and position are randomly distributed,
where the user’s distance from BS follows uniform distribution
[39,42].
In this part, we offer two cases for accessing the execution
of the suggested solution for NOMA in the uplink transmis- Fig. 3 Convergence of 2-Opt, Randomly and Hybrid
sion. The cases are discussed as follows: algorithms.
674 B.U. Rehman et al.

Fig. 6 Illustration of Spectral efficiency with Randomly, Hybrid,


Fig. 4 Illustration of Spectral efficiency versus c for 100 2-Opt and Optimal for 50 iterations.
iterations.

The Figs. 5–8 represent the case 1 with rn ¼ 1:5bits=s=Hz


for iterations 10; 50; 100 and 500. The Fig. 5 depict the
spectral-efficiency with increasing N for 10 iterations. The
spectral efficiency increases as the number of users increase
for fixed numbers of PRBs/groups. Hence, it is also a shred
of evidence that our proposed approach achieves better than
OMA systems. Furthermore, the proposed solution attains
higher spectral-efficiency than the OMA and uplink schemes
with lower time-complexity. Furthermore, the proposed 2-
Opt algorithm is nearer to the optimal-solution as the users
increase compared to randomly and hybrid algorithms. In
addition, for the higher number of iterations, the hybrid and
randomly algorithms provide better performance in-terms of
spectral-efficiency compared to the 2-Opt algorithm as shown
in Figs. 6–8. Furthermore, Figs. 5–8 imply that the BS does not

Fig. 7 Illustration of Spectral efficiency with Randomly, Hybrid,


2-Opt and Optimal for 100 iterations.

take full advantage of the spectrum resources in the OMA


scheme since the spectrum is shared by multiple users not only
to be assigned to one user.

7.2. Case 2

The power control problem is based on the solution reported


in [39,42]. The Figs. 9–12 represent the case 2 with iterations
10; 50; 100 and 500. The Fig. 9 illustrate the spectral-
efficiency versus N for 10 iterations. Furthermore, the spectral
efficiency increase as the users increases for a fixed number of
PRBs/groups. For 10 iterations, the 2-Opt algorithm achieves
a better outcome and is nearer to the optimal-solution as the
number of users increases compared to randomly and hybrid
Fig. 5 Illustration of Spectral efficiency with Randomly, Hybrid, algorithms. In addition, the NOMA approach achieve better
2-Opt and Optimal for 10 iterations. attainment than the OMA approach in-terms of spectral-
Uplink power control scheme for spectral efficiency maximization in NOMA systems 675

Fig. 8 Illustration of Spectral efficiency with Randomly, Hybrid, Fig. 10 Illustration of Spectral efficiency with Randomly,
2-Opt and Optimal for 500 iterations. Hybrid, 2-Opt and Optimal for 50 iterations.

Fig. 9 Illustration of Spectral efficiency with Randomly, Hybrid, Fig. 11 Illustration of Spectral efficiency with Randomly,
2-Opt and Optimal for 10 iterations. Hybrid, 2-Opt and Optimal for 100 iterations.

efficiency for NOMA in uplink system. The Figs. 10–12 repre- approach by employing the randomly, 2-Opt, and hybrid algo-
sent the spectral-efficiency with varying N for iterations 50; 100 rithms, as shown in Figs. 5–8. Case 2 is based on the method
and 500. It is noted that as the iterations increases the hybrid mentioned in [39,42] and shown in Figs. 9–12. The simulation
and randomly algorithms are closer to the optimal solution findings based on the proposed approach (Case 1) represent
compared to 2-Opt algorithm as shown in Figs. 10–12. that spectral efficiency is significantly greater than that of
The two cases are proposed to evaluate the performance in OMA systems and considerably closer to the optimal solution
spectral efficiency. Case 1 is based on the linear programming with lower computational complexity.
676 B.U. Rehman et al.

[4] S. Timotheou, I. Krikidis, Fairness for non-orthogonal multiple


access in 5g systems, IEEE Signal Process. Lett. 22 (10) (2015)
1647–1651.
[5] Q. Yang, H.-M. Wang, D.W.K. Ng, M.H. Lee, Noma in
downlink sdma with limited feedback: Performance analysis and
optimization, IEEE J. Sel. Areas Commun. 35 (10) (2017) 2281–
2294.
[6] M. Zeng, A. Yadav, O.A. Dobre, G.I. Tsiropoulos, H.V. Poor,
Capacity comparison between mimo-noma and mimo-oma with
multiple users in a cluster, IEEE J. Sel. Areas Commun. 35 (10)
(2017) 2413–2424.
[7] Z. Ding, Z. Yang, P. Fan, H.V. Poor, On the performance of
non-orthogonal multiple access in 5g systems with randomly
deployed users, IEEE Signal Process. Lett. 21 (12) (2014) 1501–
1505.
[8] Y. Saito, Y. Kishiyama, A. Benjebbour, T. Nakamura, A. Li, K.
Higuchi, Non-orthogonal multiple access (noma) for cellular
future radio access, 2013 IEEE 77th vehicular technology
conference (VTC Spring), IEEE, 2013, pp. 1–5.
[9] Z. Ding, X. Lei, G.K. Karagiannidis, R. Schober, J. Yuan, V.K.
Bhargava, A survey on non-orthogonal multiple access for 5g
Fig. 12 Illustration of Spectral efficiency with Randomly,
networks: Research challenges and future trends, IEEE J. Sel.
Hybrid, 2-Opt and Optimal for 500 iterations.
Areas Commun. 35 (10) (2017) 2181–2195.
[10] Z. Ding, P. Fan, H.V. Poor, Impact of user pairing on 5g
nonorthogonal multiple-access downlink transmissions, IEEE
Trans. Veh. Technol. 65 (8) (2015) 6010–6023.
8. Conclusion [11] M.B. Shahab, M. Irfan, M.F. Kader, S. Young Shin, User
pairing schemes for capacity maximization in non-orthogonal
The NOMA scheme has received much attention recently due multiple access systems, Wireless Communications and Mobile
Computing 16 (17) (2016) 2884–2894.
to its high spectral performance and plays a crucial role in
[12] M.B. Shahab, S.Y. Shin, User pairing and power allocation for
increasing future network capacity. This research addresses a
non-orthogonal multiple access: Capacity maximization under
joint optimization problem of power control and user group- data reliability constraints, Physical Communication 30 (2018)
ing with minimum rate requirements by considering the uplink 132–144.
NOMA transmission to maximize spectral performance. The [13] S.R. Islam, M. Zeng, O.A. Dobre, K.-S. Kwak, Resource
power control problem has been solved using a linear pro- allocation for downlink noma systems: Key techniques and open
gramming approach. Also, the user-pairing/grouping issue is issues, IEEE Wirel. Commun. 25 (2) (2018) 40–47.
solved using three alternative algorithms, namely the ran- [14] Y. Fu, Y. Chen, C.W. Sung, Distributed power control for the
domly, 2-Opt, and the hybrid. Simulation results show spectral downlink of multi-cell noma systems, IEEE Trans. Wireless
efficiency is much nearer to the optimal solution and more Commun. 16 (9) (2017) 6207–6220.
[15] E. Erturk, O. Yildiz, S. Shahsavari, N. Akar, Power allocation
remarkable than the OMA systems. Furthermore, the sug-
and temporal fair user group scheduling for downlink noma,
gested strategies outperform reported earlier techniques in
Telecommunication Systems (2021) 1–14.
the research, making them a viable competitor for future wire- [16] R. Chen, F. Shu, K. Lei, J. Wang, L. Zhang, User clustering and
less technology. In the future, this study might be broadened to power allocation for energy efficiency maximization in downlink
explore NOMA-enabled heterogeneous networks (HetNet) to non-orthogonal multiple access systems, Applied Sciences 11 (2)
enhance network performance. (2021) 716.
[17] W.F. Alghasmari, L. Nassef, Power allocation evaluation for
Declaration of Competing Interest downlink non-orthogonal multiple access (noma), IJACSA)
International Journal of Advanced Computer Science and
Applications 11 (4).
The authors declare that they have no known competing
[18] M. Al-Imari, P. Xiao, M.A. Imran, R. Tafazolli, Uplink non-
financial interests or personal relationships that could have
orthogonal multiple access for 5g wireless networks, 2014 11th
appeared to influence the work reported in this paper. international symposium on wireless communications systems
(ISWCS), IEEE, 2014, pp. 781–785.
References [19] N. Zhang, J. Wang, G. Kang, Y. Liu, Uplink nonorthogonal
multiple access in 5g systems, IEEE Commun. Lett. 20 (3) (2016)
[1] M. Al-Imari, M.A. Imran, P. Xiao, Radio resource allocation 458–461.
for multicarrier low-density-spreading multiple access, IEEE [20] M.A. Sedaghat, R.R. Müller, On user pairing in uplink noma,
Trans. Veh. Technol. 66 (3) (2016) 2382–2393. IEEE Trans. Wireless Commun. 17 (5) (2018) 3474–3486.
[2] S.R. Islam, N. Avazov, O.A. Dobre, K.-S. Kwak, Power- [21] G. Zhanyang, W.T. Toor, H. Jin, Cdf-based scheduling for
domain non-orthogonal multiple access (noma) in 5g systems: uplink non-orthogonal multiple access, 2018 IEEE 87th
Potentials and challenges, IEEE Commun. Surv. Tutor. 19 (2) Vehicular Technology Conference (VTC Spring), IEEE, 2018,
(2016) 721–742. pp. 1–5.
[3] L. Dai, B. Wang, Y. Yuan, S. Han, I. Chih-Lin, Z. Wang, Non- [22] Z. Yang, Z. Ding, P. Fan, N. Al-Dhahir, A general power
orthogonal multiple access for 5g: solutions, challenges, allocation scheme to guarantee quality of service in downlink
opportunities, and future research trends, IEEE Commun. and uplink noma systems, IEEE transactions on wireless
Mag. 53 (9) (2015) 74–81. communications 15 (11) (2016) 7244–7257.
Uplink power control scheme for spectral efficiency maximization in NOMA systems 677

[23] L.P. Qian, Y. Wu, H. Zhou, X. Shen, Joint uplink base station [37] M.S. Ali, H. Tabassum, E. Hossain, Dynamic user clustering and
association and power control for small-cell networks with non- power allocation for uplink and downlink non-orthogonal
orthogonal multiple access, IEEE Trans. Wireless Commun. 16 multiple access (noma) systems, IEEE access 4 (2016) 6325–6343.
(9) (2017) 5567–5582. [38] Y. Sun, D.W.K. Ng, J. Zhu, R. Schober, Robust and secure
[24] D. Zhai, J. Du, Spectrum efficient resource management for resource allocation for full-duplex miso multicarrier noma
multi-carrier-based noma networks: A graph-based method, systems, IEEE Trans. Commun. 66 (9) (2018) 4119–4137.
IEEE Wireless Communications Letters 7 (3) (2017) 388–391. [39] J. Zhang, L. Zhu, Z. Xiao, X. Cao, D.O. Wu, X.-G. Xia,
[25] Z. Wei, D.W.K. Ng, J. Yuan, Joint pilot and payload power Optimal and sub-optimal uplink noma: Joint user grouping,
control for uplink mimo-noma with mrc-sic receivers, IEEE decoding order, and power control, IEEE Wireless
Commun. Lett. 22 (4) (2018) 692–695. Communications Letters 9 (2) (2019) 254–257.
[26] I. Azam, M.B. Shahab, S.Y. Shin, User pairing and power [40] R. Karem, M. Ahmed, F. Newagy, Resource allocation in
allocation for capacity maximization in uplink noma, in: 2019 uplink noma-iot based uav for urllc applications, Sensors 22 (4)
42nd International Conference on Telecommunications and (2022) 1566.
Signal Processing (TSP), IEEE, 2019, pp. 690–694. [41] B. ur Rehman, M.I. Babar, A.W. Ahmad, M. Amir, W.
[27] Z. Zhang, H. Sun, R.Q. Hu, Downlink and uplink non- Shahjehan, A.S. Sadiq, S. Mirjalili, A.A. Dehkordi, Joint user
orthogonal multiple access in a dense wireless network, IEEE grouping and power control using whale optimization algorithm
J. Sel. Areas Commun. 35 (12) (2017) 2771–2784. for noma uplink systems, PeerJ Computer Science 8 (2022) e882.
[28] M. Salehi, H. Tabassum, E. Hossain, Meta distribution of sir in [42] B.U. Rehman, M.I. Babar, A.W. Ahmad, H. Alhumyani, G.
large-scale uplink and downlink noma networks, IEEE Trans. Abdel Azim, R.A. Saeed, S. Abdel Khalek, Joint power control
Commun. 67 (4) (2018) 3009–3025. and user grouping for uplink power domain non-orthogonal
[29] H. Tabassum, E. Hossain, J. Hossain, Modeling and analysis of multiple access, Int. J. Distrib. Sens. Netw. 17 (12) (2021),
uplink non-orthogonal multiple access in large-scale cellular 15501477211057443.
networks using poisson cluster processes, IEEE Trans. [43] X. Wang, T. Xu, T. Zhou, H. Hu, Dynamic power allocation
Commun. 65 (8) (2017) 3555–3570. strategy for uplink non-orthogonal multiple access systems,
[30] R. Ruby, S. Zhong, H. Yang, K. Wu, Enhanced uplink resource Comput. Commun. 184 (2022) 36–41.
allocation in non-orthogonal multiple access systems, IEEE [44] M. Bello, A. Chorti, I. Fijalkow, Optimal power allocation and
Trans. Wireless Commun. 17 (3) (2017) 1432–1444. user pairing for layer-1 and layer-2 sum-rate maximization in
[31] M. Zeng, A. Yadav, O.A. Dobre, H.V. Poor, Energy-efficient noma uplink networks.
power allocation for uplink noma, 2018 IEEE Global [45] H. Zheng, S. Hou, H. Li, Z. Song, Y. Hao, Power allocation and
Communications Conference (GLOBECOM), IEEE, 2018, pp. user clustering for uplink mc-noma in d2d underlaid cellular
1–6. networks, IEEE Wireless Communications Letters 7 (6) (2018)
[32] M. Zeng, N.-P. Nguyen, O.A. Dobre, Z. Ding, H.V. Poor, 1030–1033.
Spectral-and energy-efficient resource allocation for multi- [46] F. Guo, H. Lu, D. Zhu, Z. Gu, Joint user association, grouping
carrier uplink noma systems, IEEE Trans. Veh. Technol. 68 and power allocation in uplink noma systems with qos
(9) (2019) 9293–9296. constraints, in: ICC 2019-2019 IEEE International Conference
[33] F. Fang, Z. Ding, W. Liang, H. Zhang, Optimal energy efficient on Communications (ICC), IEEE, 2019, pp. 1–6.
power allocation with user fairness for uplink mc-noma systems, [47] Y.-H. Chen, Y.-F. Chen, S.-M. Tseng, D.-F. Tseng, Low
IEEE Wireless Communications Letters 8 (4) (2019) 1133–1136. complexity user selection and power allocation for uplink noma
[34] X. Tian, Y. Huang, S. Verma, M. Jin, U. Ghosh, K.M. Rabie, beamforming systems, Wireless Pers. Commun. 111 (3) (2020)
D.-T. Do, Power allocation scheme for maximizing spectral 1413–1429.
efficiency and energy efficiency tradeoff for uplink noma systems [48] Y. Endo, Y. Kishiyama, K. Higuchi, Uplink non-orthogonal
in b5g/6g, Physical Communication 43 (2020) 101227. access with mmse-sic in the presence of inter-cell interference,
[35] E. Balevi, R.D. Gitlin, Pareto optimization for uplink noma 2012 international symposium on wireless communication
power control, 2018 Wireless Telecommunications Symposium systems (ISWCS), IEEE, 2012, pp. 261–265.
(WTS), IEEE, 2018, pp. 1–5. [49] A. Li, A. Benjebbour, X. Chen, H. Jiang, H. Kayama, Uplink
[36] S. Riaz, U. Park, Power control for interference mitigation by non-orthogonal multiple access (noma) with single-carrier
evolutionary game theory in uplink noma for 5g networks, frequency division multiple access (sc-fdma) for 5g systems,
Journal of the Chinese Institute of Engineers 41 (1) (2018) 18– IEICE Transactions on Communications 98 (8) (2015) 1426–
25. 1435.

You might also like