Optimal Power Allocation for Downlink NOMA
Heterogeneous Networks to Improve Sum Rate and
Outage Probability
2022 IEEE India Council International Subsections Conference (INDISCON) | 978-1-6654-6601-1/22/$31.00 ©2022 IEEE | DOI: 10.1109/INDISCON54605.2022.9862842
Sachin Trankatwar
Department of Electrical and Electronics Engineering
Birla Institute of Technology and Science, Pilani
Hyderabad, India
strankatwar@gmail.com
Abstract—Heterogeneous networks (HetNets) have an in-
creased data rate, improve quality-of-service (QoS), lower la-
tency, and use less power for future wireless networks. Despite the
benefits, there are various problems in HetNets, with interference
being one of the most prominent. Densification in HetNets in-
creases interference while decreasing rate and outage probability.
As a result, non-orthogonal multiple access (NOMA) based
HetNets may be employed to minimize cross-tier interference.
Thus, decreasing outage probability and increasing the system
sum rate in order to fulfill future wireless communication needs.
Power allocation (PA) is critical for increasing system throughput
and reducing outages in a NOMA network. Hence, this study
comes up with an optimal PA for downlink transmission NOMA-
based HetNets in order to optimize the system’s sum rate and
outage probability while adhering to the minimal user rate
constraint. We derived a generalized optimum PA coefficient
equation for small cell users of NOMA-based HetNets. Then,
utilizing the PA coefficient equation, we presented an algorithm
to optimize the sum rate and minimize the outage probability. To
apply our algorithm in real-world wireless networks, we ensure
that our algorithm is both fast and minimal in complexity. Finally,
we illustrate simulation outcomes for the proposed method and
compare them to OMA systems.
Index Terms—Heterogeneous networks (HetNets), Non orthog-
onal multiple access (NOMA), Power allocation (PA), Sum rate,
Outage probability, Quality of service (QoS)
I. I NTRODUCTION
The widespread use of devices in wireless networks and
the demand for high-speed internet have made heterogeneous
networks (HetNets) one of the most promising communica-
tion networks for delivering high rates and extensive indoor
and cell-edge coverage in wireless networks. HetNets are
composed of low-power small cells (Microcell, Picocell, and
Femtocell) superimposed over a high-power macrocell [1].
It is anticipated that the deployment of these small cells
over the macrocell can enhance essential parameters such as
spectrum efficiency, outage probability, user performance at
the cell edge, energy consumption, and sum rate [1]. However,
interference in HetNets rises as a result of network densifica-
tion, intensive growth, and the usage of the same frequency
978-1-6654-6601-1/22/$31.00 ©2022 IEEE
Authorized licensed use limited to: MULTIMEDIA UNIVERSITY. Downloaded on December 01,2023 at 20:08:30 UTC from IEEE Xplore. Restrictions apply.
Prashant Wali
Department of Electrical and Electronics Engineering
Birla Institute of Technology and Science, Pilani
Hyderabad, India
iwalihere@gmail.com
by all cells [1]. As a result, the cellular network’s outage
and data rate performance degrades. Hence, interference is a
significant performance constraint in the network [1]. Using
Non orthogonal multiple access (NOMA) in HetNets helps
reduce cross-tier interference and improve the network’s sum
rate and outage performance [2], which we do in this paper.
The frequency band in typical OFDMA HetNets may be
split into many sub-frequency bands. The users in the high-
power macro and low-power small cells are allocated to
various sub-frequency bands to minimize cross-tier interfer-
ence [3]. However, the receivers in NOMA HetNets employ
successive interference cancellation (SIC) to enable several
users to multiplex on the same sub-frequency band, resulting
in a higher sum rate than in OMA [4]. In the NOMA SIC
process, strong users can eliminate interference from weak
users who share the same sub-frequency band. Because several
users in the NOMA network share the same frequency/time
resource, users are allocated distinct powers depending on
channel circumstances. The NOMA network gives weak users
more power and strong users less power. Power allocation (PA)
coefficients can be used to provide different levels of power to
different users. As a result, PA is a vital tool for the NOMA
network’s sum rate and outage improvement. Hence, this paper
focuses on the minimization of outage probability as well as
maximization of sum rate for downlink NOMA heterogeneous
networks obtaining optimal PA.
II. L ITERATURE R EVIEW
The downlink NOMA network’s outage probability has been
addressed in a few studies. The work [5] investigates the
optimal outage probability problem in the NOMA network
while considering PA, selection of decoding order, and group-
ing of users. In a paper [6], NOMA’s ergodic sum rate and
outage probability for randomly deployed users with optimal
PA are investigated under perfect and imperfect channel state
information (CSI). The optimum PA is explored in [7] in
respect of sum rate and outage probability under CSI to
optimize fairness between NOMA downlink network users.
Work [8] derives and investigates closed-form expressions
for network capacity and outage probability in simultaneous Table I: Illustrates the system parameters.
wireless information and power transmission to NOMA net- Notation Description
works. The outage performance and diversity order by the S Channel realization.
SIN Rm,n SINR of SCU n at SC m
cooperative NOMA method are studied in the article [9]. In P Transmission power from SBS m to SCU n
paper [10], the authors come up with a simple way to calculate dm,i Distance from SBS m to SCU n
outage probability in heterogeneous cellular networks (HCNs) hm,i Channel gain from SBS m to SCU n
gm,i Channel gain from MBS to SCU n
of downlink transmission with variable cell affiliation. The xm,i Data intended for nth user in SC m
HCN is a multi-tiered cellular network, with base stations wm,i Noise for nth user in SC m
(BSs) dispersed at random locations in each tier. In [11], a αm,i PA coefficient from SBS m to SCU n
closed-form equation for secrecy outage probability in HetNets Pk Transmission power from MBS m to MU k
B Bandwidth
and K-tier HetNets based on stochastic geometry is derived Rm,n Data rate of SBS m to SCU m
∗
with a BS association constraint. Rm,n Target rate of SCU n of SBS m
Am,n Additive white gaussian noise from SBS m to SCU n
In addition to improving outage probability, one of the pl Path loss exponent
primary objectives of this study is to improve the sum rate.
For HetNets based on NOMA, subcarrier allocation and a PA Table II: Illustrates the System parameters and it’s index set.
iterative algorithm is offered to maximize the sum rate of small Notation Description and it’s index set
cells in a limited number of iterations [2]. The subchannel M SBS, M = 1, 2, ....., M
allocation problem and the PA problem in downlink NOMA- K MCU, K = 1, 2, ....., K
N SCU, N = 1, 2, ....., N
based HetNets are investigated in the article [12]. In [13], the
authors find the best PA technique for NOMA networks to
maximize the sum rate. They do this by fulfilling minimum III. S YSTEM M ODEL
user rate requirements. In [14], determine the maximum sum
rate of a downlink two-user NOMA network under the as- This section introduces the downlink NOMA based HetNets
sumption of an imperfect SIC and minimal quality of service system model. Let us consider a downlink NOMA-based two-
(QoS) constraint. In [15], the authors look into the best PA for tier heterogeneous wireless network, as shown in figure 1.
a sub-carrier-based NOMA network to maximize its sum rate M small cells (SCs) are superimposed on a single macrocell
under total power and QoS constraint. Work [16] maximizes (MC) in the network. Each SC has N SC users (SCUs),
the total rate using optimum PA in a multiple-input multiple- while each MC has K MC users (MCUs). Let’s denote
output (MIMO) NOMA network with layered transmissions. dm,n their respective distance from SC base station (SBS) m
The authors offer an alternating maximization algorithm for to SCU n. Let’s denote hm,n their corresponding complex
instantaneous CSI and statistical CSI at the base station. [17] Rayleigh fading coefficient from SBS m to SCU n, Where
investigates the subcarrier and PA for a single-cell multicarrier hm,n = rm,n d−plm,n , rm,n is a Rayleigh distribution and pl is

downlink transmission NOMA network. The Double Iterative
Waterfilling Algorithm is introduced to optimize the NOMA
network’s sum rate. Paper [18] proposes a multi-objective
optimization framework for effectively allocating PA in a
downlink transmission NOMA network, which maximizes
the total rate while reducing transmit power and considering
the QoS, SIC, and transmit power budget. The sum-rate
performance of two and multiple users in the MIMO NOMA
network is investigated in [19].
In NOMA networks and HetNets, most of the research
looked at the problem of outage probability and the network’s
rate separately. In addition, only a few studies have examined
both outage probability and network rate in NOMA networks.
However, to the best of our knowledge, in most current work
on NOMA-based HetNets, considering both outage probability
and the date rate of the network is usually ignored. As a
result, one of the required challenges in the downlink NOMA-
based HetNets research is to optimize outage probability and
overall rate using optimal PA. Consequently, with NOMA-
based HetNets, we are motivated to improve both outage
probability and sum rate with the optimal PA. Table I shows
system parameters and it’s discription and Table II shows index
set and it’s discription used in this manuscript.
the path loss exponent. Let’s denote gm,n their corresponding
complex Rayleigh fading coefficient from MBS to SCU n in
SC m, Am,n = CN (0, σ 2 ) is the Additive white gaussian
noise (AWGN), Let’s indicate xm,n data intended for nth
user in SC m. Let’s represent wm,n interference plus noise
for nth user in SC m. Without losing generality, we arrange
channel coefficients on SBS m in a descending order as
|hm,1 | ≥ |hm,2 | ≥ · · · ≥ |hm,N |.
According to NOMA’s principles, the weakest user should
have the greatest power, while the strongest user should have
the least. As a result, the PA coefficients must be sorted as
αm,1 ≤ αm,2 ≤ αm,3 · · · ≤ αm,N , Where αm,n PA coefficient
from SBS m to SCUs n. Due to low power and intense
penetration loss in SC, we can ignore interference between
different SCUs [3], [20]. The signal to interference plus noise
ratio (SINR) of the NOMA network received at the nth SCU
of SC m can be written as [18],
2
αm,n P |hm,n |
SIN Rm,n = 2 Pn−1 2 (1)
|hm,n | P j=1 αm,n + Pk |gm,n | + σ 2
According to Shannon’s capacity formula, the data rate of SCU
n in SC m is written as [18],
Rm,n = B log(1 + SIN Rm,n ) (2)

Authorized licensed use limited to: MULTIMEDIA UNIVERSITY. Downloaded on December 01,2023 at 20:08:30 UTC from IEEE Xplore. Restrictions apply.
 Desired signal
Interference
signal
MCU 1

SCU 2
SCU 1
SCU 1
SBS 3
SCU 2
SBS 1
SCU 3
SCU 1
MBS
SCU 2
MCU 2 SBS 2

Figure 1: A downlink transmission NOMA-based Heterogeneous Network

∗
Thus, the sum rate of SCU n in SC m is given as, To obtain PA coefficient αm,2 , we can use Rm,2 ≥ Rm,2 ,
N
solving, we get,
2
X αm,n P |hm,n | 2
SR = 2Pn−1 2 (3) αm,2 P |hm,2 | ∗
n=1 |hm,n | P j=1 αm,n + Pk |gm,n | + σ
2
2 2 ≥ 2Rm,2 − 1
αm,1 P |hm,2 | + Pk |gm,1 | + σ2
For simplicity, we use B = 1 Hz. If the data rate falls below
the user’s required minimum data rate at any point, the user solving, αm,2 equation can be written as,
can experience an outage. As a result, outage probability of 2
ξ1 ξ2 (Pk |gm,1 | + σ 2 )
2
ξ2 (Pk |gm,2 | + σ 2 )
SCU n in SC m is expressed as, αm,2 ≥ 2 + 2
P |hm,1 | P |hm,2 |
∗
Outage Probability = P(Rm,n < Rm,n ) (4) Therefore, using the deduction method and taking the lower
Where, ∗
Rm,n is target rate of SCU n of SBS m. limit of αm,n , the PA coefficient for user n in SC m can
be written as shown in equation (7). This equation can be
IV. P ROPOSED M ETHOD expressed in terms of previously determined PA coefficients,
which are as shown in equation (8). Thus, we achieved the
This section determines the best PA solution that maximizes optimal αm,n in the NOMA-based HetNets while adhering to
the sum rate and minimizes the outage probability in NOMA- the QoS constraint. In the context of the preceding description,
based HetNets. To achieve optimal PA coefficient while en- we now propose an algorithm 1 for acquiring the optimum
suring QoS for user n in SC m, we choose αm,n such that, sum rate and outage probability with PA in a downlink
∗
Rm,n ≥ Rm,n . To determine the PA coefficient for user 1 NOMA-based HetNets while sticking to the QoS constraint.
(nearest user) in small cell m, we can describe the achievable We assume that all users’ target rates Rm,n∗
are the same in
rate of user 1 as, the algorithm. Lines 13 and 14 provide the outage probability
and sum rate, respectively.
!
2
αm,1 P |hm,1 |
Rm,1 = log 1 + 2 (5) We can now examine the time complexity of our proposed
Pk |gm,1 | + σ 2 algorithm. The if statement of the algorithm has no impact
∗
To obtain PA coefficient αm,1 , we use Rm,1 ≥ Rm,1 , solving, on the algorithm’s time complexity. The time complexity of
we get, the αm,n equation is N . Since the inner loop (lines 5 to 12)
2 repeats S times, the worst-case time complexity of the inner
αm,1 P |hm,1 | ∗

2 ≥ 2Rm,1 − 1 loop is S, provided S is greater than N . Let us suppose R

Pk |gm,1 | + σ 2
has a length of L. The outer loop is repeats L times, giving a
Let’s denote the target SINR for user n on SC m as, ξn = worst-case time complexity of LS. Consequently, the worst-
∗
2Rm,n − 1, solving, equation of αm,1 can be written as, case run time complexity of our proposed algorithm is O(LS).
Thus, we have developed a simple algorithm that has a low
2
ξ1 (Pk |gm,1 | + σ 2 ) time complexity and can be used with real-world wireless
αm,1 ≥ (6)
P |hm,1 |
2 communication networks.

Authorized licensed use limited to: MULTIMEDIA UNIVERSITY. Downloaded on December 01,2023 at 20:08:30 UTC from IEEE Xplore. Restrictions apply.
 n
X n−1
X
αm,n = ξi 1 + ξj +Sum of all possible two product terms of ST+
i=1 j=i+1
(7)
| {z }
ST
! !!
2
ξn (Pk |gm,i | + σ 2 )
Sum of all possible three product terms of ST + · · · 2
P |hm,i |

n−1
X n−1
X
αm,n = αm,i 1 + ξj +Sum of all possible two product terms of ST
i=1 j=i+1
(8)
| {z }
ST
!!
2 2
ξn (Pk |gm,n | + σ )
+ Sum of all possible three product terms of ST + · · · + 2
P |hm,n |

Algorithm 1 Outage performance and sum rate in a small cell Table III: System parameters required for simulation.
of NOMA based HetNets System Parameters Values
1: Requires: S,Pk ,P ,pl,M ,N , σ 2 . Channel realization (S) 105
2: x = zeros(1, S); The number of small cells (M ) 1
The number of users in a small cell (N ) 2 and 3
3: Obtain hm,n and gm,n ;
Total transmission power from MBS (Pk ) 1 Watt
4: Generate equally spaced R∗ vector. Total transmit power from SBS (P ) 0.1 Watt
5: for i = 1 to length(R∗ ) do Variance of AWGN noise (σ 2 ) 10−4
The minimum required rate for QoS (Rmin∗ ) 1 bps/Hz
6: Compute αm,n from eqn. (8);
Path loss exponent (pl) 3
7: Find data rates for each users in SCs;
8: for u = 1 to S do
9: if Rm,n (u) ≤ R∗ (i) then Table IV: Simulation results obtained for the proposed method.
10: x(u) = x(u) + 1 for each users System Obtained values
11: end if Parameters 2 Users 3 Users
α1 0.1845 0.1924
12: end for α2 0.8155 0.2420
13: OP = x/S for each user; α3 - 0.5656
14: Calculate Sum rate in each SC; R1 1.0000 bps/Hz 1.0000 bps/Hz
15: end for R2 4.1729 bps/Hz 1.0000 bps/Hz
R3 - 2.9909 bps/Hz
Sum rate 5.1729 bps/Hz 4.9909 bps/Hz

V. R ESULTS
This section introduces and discusses the simulation results
for our proposed problem. Furthermore, we compare simula-
tion results of the proposed method with OMA to demonstrate
comparable performance. Unless otherwise specified, Table III
contains the simulation parameters used to obtain the simu-
lation results. In the simulation, we obtain the performance
parameters for all users across 105 Rayleigh fading channel
gain realizations (S) and then we take the average of these
performance parameters. The system parameters obtained in
simulations for our proposed method are summarised in Table
IV. We provide the PA coefficients and the achievable rates for
each user in SC for two and three users case for the proposed
method in Table IV. Additionally, the sum rate for two and
three users in SC is shown. The PA coefficients are between
0 and 1, and their sum equal to 1 indicates that users take
power in such a way that their total power equals P , as seen
in table IV. Individual rates in table IV ensure that each user’s
minimum rate constraint is fulfilled.
Figures (2) and (3) depicts the fluctuation of outage prob-
ability with target rate for the proposed NOMA method and
OMA method for two and three users in SC of NOMA-based
HetNets. The outage probability of NOMA-based HetNets
outperforms the OMA method irrespective of target rates,
as seen in figures (2) and (3). This is obliged because PA
coefficients are dynamically modified whenever the channels
and the target rates change. It’s straightforward to notice in
figures (2) and (3) that the SC user’s performance worsens
as the target rate rises. This should happen because the SC
user’s chances of meeting the target rate get less and less as
∗
the target rate rises. Hence, Rmin must be smaller in order
for the proposed and OMA methods to perform better for two
and three users, as seen in figures (2) and (3).
In figure (2), we can observe that the strong user performs
well for the proposed method, but the weak user instantly

Authorized licensed use limited to: MULTIMEDIA UNIVERSITY. Downloaded on December 01,2023 at 20:08:30 UTC from IEEE Xplore. Restrictions apply.
 quickly approaching outage. The outage performance of user
3 gives quite a sharp transition around 0.5 to 3.5 bps/Hz and
1.0 exceeding that, user 3 is always in an outage. The outage of
all users is almost the same for the OMA method. Initially,
users show transition around values 0 to 3 bps/Hz; after that,
0.8
they are always in an outage. Thus, NOMA-based HetNets
Outage probability

outperforms the OMA method for three users in SC.

0.6 The change of sum rate with transmit power is shown in
figures (4) and (5) for the proposed NOMA and OMA methods
0.4
for two and three users in SC of NOMA-based HetNets. The

User 1 (OMA)
User 2 (OMA)
0.2
User 1 (NOMA)
User 2 (NOMA) 8.5
0.0 8.0
0 1 2 3 4 5 6 7 8
Target rate (bps/Hz) 7.5

Sum rate (bps/Hz)

7.0
Figure 2: Outage probablity Vs target Rate for proposed
NOMA method and OMA method for two users case in SC 6.5
of NOMA-based HetNets.
6.0

5.5

5.0 OMA
NOMA
1.0
4.5
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0.8 Transmit power (Watt)
Outage probability

0.6
Figure 4: Sum rate versus transmit power for proposed NOMA
method and OMA method for two users case in SC of NOMA-
User 1 (OMA) based HetNets.
0.4 User 2 (OMA)
User 3 (OMA)
User 1 (NOMA)
User 2 (NOMA)
0.2 User 3 (NOMA)

0.0 8
0 1 2 3 4 5 6 7 8
Target rate (bps/Hz)
7
Sum rate (bps/Hz)

Figure 3: Outage probablity Vs target Rate for proposed

NOMA method and OMA method for three users case in SC 6
of NOMA-based HetNets.

5
reaches outage. In the proposed method, the far user’s outage
∗ OMA
increase rapidly for the initial Rmin , and beyond 5 bps/Hz, the NOMA
4
far user is continuously in an outage. The strong and weak
user outage performance is almost the same for the OMA
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
method. The OMA method shows transition around values 1
Transmit power (Watt)
to 5 bps/Hz for both users, and beyond that, users are always in
an outage. Thus, for two users in SC, NOMA-based HetNets
performance beat the OMA method. In figure (3), we can Figure 5: Sum rate versus transmit power for proposed NOMA
observe that users 1 and 2 outage performance is satisfactory method and OMA method for three users case in SC of
∗
for all Rmin in the proposed method, but the weakest user is NOMA-based HetNets.

Authorized licensed use limited to: MULTIMEDIA UNIVERSITY. Downloaded on December 01,2023 at 20:08:30 UTC from IEEE Xplore. Restrictions apply.
figures (4) and (5) show that when transmission power rises,
the network’s sum rate of NOMA-based HetNets and OMA
methods for two and three users increases as well. Given that
we used the QoS constraint to obtain the optimum PA in the
proposed method, the proposed method’s sum rate may be
higher than the OMA method, as shown in figures (4) and
(5). The NOMA sum rate outperforms the OMA sum rate
because each user in OMA must share bandwidth with all other
users compared to NOMA, where each user shares the entire
bandwidth. Additionally, in figures (4) and (5), the separation
between NOMA and OMA sum rates increases as transmit
power increases. This happened because PA is inextricably
linked to the SIC process. The user with the small PA is
greatly affected by noise when the BS’s transmit power is
low, resulting in a sum rate degradation of NOMA compared
to OMA. For low transmit power, this is exactly what we see
in figures (4) and (5). The receiver can detect the signal more
accurately, and SIC can be carried out more efficiently if the
transmit power is sufficiently high, resulting in a significant
rate increase. As a result, at higher transmit power, the NOMA
sum rate increases more than the OMA sum rate, as illustrated
in figures (4) and (5).
VI. C ONCLUSIONS
This study aims to find the optimal PA problem of NOMA-
based HetNets for downlink transmission to optimize the sum
rate and outage. We devised a problem to enhance the sum rate
and reduce the outage while sticking to the minimum data rate
requirement with optimal PA coefficients. A novel technique
is investigated to get the solution of the proposed method in a
NOMA-based HetNets for downlink transmission. In the SC of
NOMA-based HetNets, we derived a generalized optimal PA
coefficient equation. We then offered a low-complexity, fast
algorithm for our proposed method to improve the sum rate
and decrease outage probability. It is proved that the proposed
method functions efficiently in a NOMA network for downlink
communication. In this manuscript, we illustrate simulation
results for the proposed method and compare the proposed
method results to results from the OMA scheme.
In the future, we want to enhance the sum rate and outage
probability of downlink transmission NOMA-based HetNets
while taking the SIC constraint into account.
R EFERENCES
[1] S. Trankatwar and P. Wali, “Power control algorithm to improve
coverage probability in heterogeneous networks,” Wireless Personal
Communications, vol. 121, no. 3, pp. 1821–1833, 2021.
[2] J. Zhao, Y. Liu, K. K. Chai, A. Nallanathan, Y. Chen, and Z. Han,
“Spectrum allocation and power control for non-orthogonal multiple
access in hetnets,” IEEE Transactions on Wireless Communications,
vol. 16, pp. 5825–5837, 2017.
[3] H. Zhang, C. Jiang, N. C. Beaulieu, X. Chu, X. Wang, and T. Q. S. Quek,
“Resource allocation for cognitive small cell networks: A cooperative
bargaining game theoretic approach,” IEEE Transactions on Wireless
Communications, vol. 14, no. 6, pp. 3481–3493, 2015.
[4] H. Zhang, F. Fang, J. Cheng, K. Long, W. Wang, and V. C. M. Leung,
“Energy-efficient resource allocation in noma heterogeneous networks,”
IEEE Wireless Communications, vol. 25, no. 2, pp. 48–53, 2018.
Authorized licensed use limited to: MULTIMEDIA UNIVERSITY. Downloaded on December 01,2023 at 20:08:30 UTC from IEEE Xplore. Restrictions apply.
[5] S. Shi, L. Yang, and H. Zhu, “Outage balancing in downlink nonorthog-
onal multiple access with statistical channel state information,” IEEE
Transactions on Wireless Communications, vol. 15, no. 7, pp. 4718–
4731, 2016.
[6] Z. Ding, Z. Yang, P. Fan, and H. V. Poor, “On the performance of
non-orthogonal multiple access in 5g systems with randomly deployed
users,” IEEE Signal Processing Letters, vol. 21, no. 12, pp. 1501–1505,
2014.
[7] S. Timotheou and I. Krikidis, “Fairness for non-orthogonal multiple
access in 5g systems,” IEEE Signal Processing Letters, vol. 22, no. 10,
pp. 1647–1651, 2015.
[8] Y. Liu, Z. Ding, M. Elkashlan, and H. V. Poor, “Cooperative non-
orthogonal multiple access with simultaneous wireless information and
power transfer,” IEEE Journal on Selected Areas in Communications,
vol. 34, no. 4, pp. 938–953, 2016.
[9] Z. Ding, M. Peng, and H. V. Poor, “Cooperative non-orthogonal multiple
access in 5g systems,” IEEE Communications Letters, vol. 19, no. 8,
pp. 1462–1465, 2015.
[10] H.-S. Jo, Y. J. Sang, P. Xia, and J. G. Andrews, “Outage probability for
heterogeneous cellular networks with biased cell association,” in 2011
IEEE Global Telecommunications Conference - GLOBECOM 2011,
pp. 1–5, 2011.
[11] H. Wu, X. Tao, N. Li, and J. Xu, “Secrecy outage probability in multi-rat
heterogeneous networks,” IEEE Communications Letters, vol. 20, no. 1,
pp. 53–56, 2016.
[12] X. Chu, H. Zhang, W. Huangfu, W. Liu, Y. Ren, J. Dong, and K. Long,
“Subchannel assignment and power optimization for energy-efficient
noma heterogeneous network,” in 2019 IEEE Global Communications
Conference (GLOBECOM), pp. 1–6, 2019.
[13] Z. Yang, W. Xu, C. Pan, Y. Pan, and M. Chen, “On the optimality of
power allocation for noma downlinks with individual qos constraints,”
IEEE Communications Letters, vol. 21, no. 7, pp. 1649–1652, 2017.
[14] I. A. Mahady, E. Bedeer, S. Ikki, and H. Yanikomeroglu, “Sum-rate
maximization of noma systems under imperfect successive interference
cancellation,” IEEE Communications Letters, vol. 23, no. 3, pp. 474–
477, 2019.
[15] Z. Q. Al-Abbasi and D. K. So, “Power allocation for sum rate maxi-
mization in non-orthogonal multiple access system,” in 2015 IEEE 26th
Annual International Symposium on Personal, Indoor, and Mobile Radio
Communications (PIMRC), pp. 1649–1653, IEEE, 2015.
[16] J. Choi, “On the power allocation for mimo-noma systems with layered
transmissions,” IEEE Transactions on Wireless Communications, vol. 15,
no. 5, pp. 3226–3237, 2016.
[17] Y. Fu, L. Salaün, C. W. Sung, C. S. Chen, and M. Coupechoux, “Double
iterative waterfilling for sum rate maximization in multicarrier noma
systems,” in 2017 IEEE International Conference on Communications
(ICC), pp. 1–6, IEEE, 2017.
[18] W. U. Khan, F. Jameel, T. Ristaniemi, S. Khan, G. A. S. Sidhu, and
J. Liu, “Joint spectral and energy efficiency optimization for downlink
noma networks,” IEEE Transactions on Cognitive Communications and
Networking, vol. 6, no. 2, pp. 645–656, 2019.
[19] M. Zeng, A. Yadav, O. A. Dobre, G. I. Tsiropoulos, and H. V. Poor, “On
the sum rate of mimo-noma and mimo-oma systems,” IEEE Wireless
communications letters, vol. 6, no. 4, pp. 534–537, 2017.
[20] H. Zhang, M. Feng, K. Long, G. K. Karagiannidis, V. C. Leung, and
H. V. Poor, “Energy efficient resource management in swipt enabled
heterogeneous networks with noma,” IEEE Transactions on Wireless
Communications, vol. 19, no. 2, pp. 835–845, 2019.