Content uploaded by Adedotun Ajibare
Author content
All content in this area was uploaded by Adedotun Ajibare on Oct 29, 2021
Content may be subject to copyright.
Content uploaded by Adedotun Ajibare
Author content
All content in this area was uploaded by Adedotun Ajibare on Oct 29, 2021
Content may be subject to copyright.
Dynamic Slicing and Scheduling for 5G Networks Using Joint
Power and Sub-Carrier Allocation
Adedotun T. Ajibare#, Olabisi E. Falowo#
#Department of Electrical Engineering, University of Cape Town, Cape Town, South Africa.
1ajbade004@myuct.ac.za
2olabisi.falowo@uct.ac.za
Abstract—This paper investigates the joint power and sub-
channel allocation scheme in 5G Sliced networks, where through
Network Slicing (NS), an Infrastructure Provider (InP) provides
different services to Multiple Virtual Network Operators
(MVNOs) through slices that have different QoS requirements.
With the aim of minimizing the co-users interference among users
in the 5G networks, we propose a novel slicing carrier assignment
scheme (SCAS) where an optimization problem is formulated to
minimize the total transmit power while guaranteeing a minimum
data rate requirement for each slice. The proposed SCAS
minimizes the downlink transmit power subject to QoS
constraints, interference thresholds, gNB power budget and the
sub-channel orthogonality constraints. The scheme assigns sub-
channels to users considering the transmit power level of the
neighboring sub-channel before allocating the sub-channel to a
user by comparing the transmit power threshold of the slice the
user belongs to. Simulation results show that the proposed scheme
reduces the effect of tolerable interference on the users, increase
the number of admitted users and provide the target data rate of
the slice for its users.
Keywords— 5G, network slicing, resource allocation, radio
frequency interference, power control, spectrum resource
sharing.
I. INTRODUCTION
It is expected that the Fifth Generation (5G) networks will
solve the problem of envisaged data increase of which the
global monthly data traffic is estimated to be around 48
exabytes by 2022. [1]. Also, the 5G Network is anticipated to
support massive connections of more than 100 times today’s
traffic volume[2] [3] and besides, support diverse quality of
services (QoS) requirements such as low latency, high data
rate, etc.
Network Slicing as an inherent concept in 5G networks
allows the composition of multiple virtual networks operating
on a partitioned physical network. Each slice is designed
differently depending on the QoS requirements of the specific
use case. The use cases as defined by the International Mobile
Telecommunications (IMT) 2020 and beyond are the enhanced
Mobile Broadband (eMBB), Ultra-Reliable Low-Latency
Communication (URLLC), and the massive Machine Type
Communication (mMTC) which are broadly the slice types in
5G networks [3].
Network slicing is made possible by virtualizing the
wireless network infrastructures and radio resource into
different logical networks. Virtualization as one of the key
technology in 5G sliced network brings the benefits of making
the network more flexible, efficient and manageable. It also
provides the idea of multitenancy where different Mobile
Virtual Network Operators (MVNOs) share the underlay
physical network of the Infrastructure Providers (InP) [4].
Other benefits include; maximizing the profit of the MVNOs
and InP, maximizing the resource efficiently etc.
However, for 5G Sliced Network, one of the most important
challenges is how to flexibly, efficiently and dynamically
assign both radio and core network resources to an isolated
virtualised network. Also, the problem of interference caused
by both spectrum resources sharing and the high transmit
power of some slice users as a result of some QoS requirements
of the slice such as high data rate allocated to these users is a
non-trivial task.
Based on the QoS requirement of the use cases, it is
expected that the various slices will require different transmit
power. For example, an eMBB slice refers to as slice 2 in Fig.
1, is envisaged to require the highest data rate compared to
other slices, hence more power. Slice 3, an average URLLC
user will require more data rate than an average slice 1
(mMTC) user, hence slice 2 require more power than slice 1.
Using other 3GPP use case requirements such as latency and
connection rate, these requirements are categorised alongside
the power required in gaining access and requesting resource
from the 5G gNodeB (gNB). For instance, a slice that has a
high data rate, a high connection rate and a high latency is
expected to use more power and therefore, causes higher co-
user interference among the users of the channel.
Figure 1: 5G Use case (Slice’s) QoS requirement analogy with transmit
Power.
In this paper, to meet the adaptive resource allocation
problem and reduce the interference challenge, we address
these issues by introducing a dynamic slicing and scheduling
scheme for virtualised 5G sliced network. In this scheme, we
propose an efficient joint power and subcarrier allocation
algorithm that assigns sub-channels to slice users based on
their transmit power and subject to the QoS constraints and the
Channel State Information (CSI) of the sub-channel. Using
orthogonality of the sub-channels, Physical Resource Blocks
(PRBs) are allocated to users based on the slice they belong to.
For example, if a User Equipment (UE) U11 that belongs to a
slice that uses high power is allocated a sub-channel, UE U12
Page 32 Southern Africa Telecommunication Networks and Applications Conference (SATNAC) 2019
belonging to the same slice (with equal or more power usage)
as U11 will not be allocated the next sub-channel or PRB.
Rather, UE from another slice with less power consumption
will be assigned the successive sub-channel.
The orthogonality constraint reduces interference, hence
increases the Signal to Interference Noise Ratio (SINR) of the
channel, increases the throughput and therefore increases the
overall network efficiency. This proposed dynamic scheme is
made possible using a slice scheduler.
The main contributions of the paper are summarised as
follows:
We formulate an optimization problem for Resource
Allocation (RA) problem in a 5G sliced Network to
maximise utilization and minimize the total downlink
transmit power subject to QoS constraints,
interference limit and the Channel State Information
(CSI)
We incorporate the idea of orthogonality in the sub-
channel assignment and apply it to sliced network in
allocating Physical Resource Blocks (PRBs) to slice
Users (UEs) to minimise the co-user interference in
the 5G network. This is based on the different QoS
requirement of these slices. Hence, improve the
overall target rate of the users.
We propose an algorithm that uses a centralised
scheduler that dynamically solves the resource
allocation problem.
Through simulation analysis, we evaluate and
validate the proposed algorithms by comparing it with
other sub-optimal solutions such as the scheme that
allocates transmit power equally among its sub-
carriers and without orthogonality constraint.
The rest of the paper is organised as follows. In Section II,
we briefly review the related works in the literature. After
discussing the proposed system model and assumptions in
Section III, the problem formulation is defined in Section IV.
The proposed scheduling mechanism and the performance
evaluation are presented in Sections V and VI respectively.
Finally, Section VII concludes the paper.
II. RELATED WORK
In this section, we briefly review some of the relevant and
recent studies on resource allocation and flexible scheduling in
virtual wireless networks.
The authors in [5] discuss major works in the field of
wireless network virtualization. They review important topics
such as their challenges and performance metrics in their
survey. In [6], the authors propose a component called Slice
Optimizer which communicates and receives information
about the network slices from an SDN controller. The Slice
Optimizer which is an extension to the LTE’s Radio Access
Networks uses the received information and adapts the slices
according to the network state. The authors rely on SDN
architecture, the slice Optimizer along with the SDN controller
is implemented in NS-3.
The authors in [7] propose a framework that leverages on
QoS Class Identifier (QCI) and security requirements in
negotiating, selecting and assigning virtualized networks to the
requesting applications or users in 5G Networks. In [8], A.
Devlic, et al describe a network slicing management and
orchestration framework that automates, configure and
activate multiple network infrastructure resource domain.
However, the authors of [7] and [8] have not validated their
proposed frameworks with either simulation or experiment.
In [9], the authors present a slice scheme where a Wireless
Virtualized Network (WVN) is sliced into two service-based
slices, with resource-based and rate-based reservations. The
scheme aims to maximize the total rate of WVN by
guaranteeing a minimum Resource Blocks for each slice.
However, the authors considered equal chain gain and assumed
equal noise power over all sub-carriers. The authors in [10]
propose a framework and formulated an RA problem in small
cells that minimizes the total downlink transmit power subject
to their power budget, QoS requirements and interference
threshold of macrocell users. In [11], the authors present a
slicing and scheduling scheme to meet the different rate of
virtual Networks (VNs). Each VN is allocated certain numbers
of sub-channels to provide services to its users. This scheme
focuses on isolation among the VNs. A. Pratap, et al in [12]
propose a joint resource allocation that minimizes interference
and maximizes spectrum reuse in the 5G heterogeneous small
cell networks keeping user-level fairness into consideration.
However, none of the authors has focused on the transmit
power required for the various slices which are different as a
result of their specific QoS requirements in minimizing the co-
users interference.
In this paper, we focus on 5G RAN where PRBs are
assigned to users based on their slice target rates and sub-
channel CSI. We propose a dynamic joint power and subcarrier
allocation algorithm that assigns sub-channels to slice users
based on their transmit power and subject to the QoS
constraints and Channel State Information (CSI) of the sub-
channels. We validate our scheme using a different number of
users, varying the transmit power for the users with different
target rates. We analyse the performance of the proposed
scheme through simulations, using data rate and connection
density QoS metrics.
III. SYSTEM MODEL AND ASSUMPTIONS
We consider a system model which is illustrated by Fig 2. It
shows the downlink of a single cell scenario of a 5G Slice
Networks. This base station which in 5G terms is also known
as gNodeB (gNB) is owned and managed by a single
infrastructure provider (InP). There are several Mobile Virtual
Network Operators (MVNOs) sharing the physical network
and the wireless spectrum. Each MVNO provides service to its
set of users (UEs). We considered isolation by assuming a
scheme that place restriction on the assigned minimum
required resources to the different tenants. Each MVNO is
allocated a certain minimum share of the sub-channels which
is required to efficiently serve their users and it is also based
on the contract with the InP. This changes dynamically with
target rate and CSI.
Figure 2: System Model with single gNB and multiple MVNOs
Southern Africa Telecommunication Networks and Applications Conference (SATNAC) 2019 Page 33
The network services are virtualized into a specific set of
slices, S = {1,2,…,s}. The network also has a set of MVNOs
denoted by M = {1,2,…,m}. Each MVNO M serves a set of
users, U = {1,2,…,u} that are assigned to a particular slice Sm.
(Sm S) from the virtual network resources based on its
demand and provide various services for their customers. We
assume three slices based on the 5G use cases which are
mMTC, eMBB and URLLC. Also, there are c C sub-
channels in the frequency domain where the system bandwidth
B is divided into bandwidth
of size 180 kHz each [13].
The T sub-frames in the time domain refers to as Transmission
Time Interval (TTI) of 1ms which ten of it makes up the frame
structure of the downlink air interface [14]. This total of TC
PRBs is available for the network Scheduler in each duration
of T. The minimum amount of the available PRBs of the 5G
network assigned to the MVNOs is based on the agreed
contract with the InP and this will be updated after each
duration of scheduling round.
The gNB has a total transmit power of and let
denotes the power allocated to user u in the sub-channel c
belonging to slice s. The channel gains depend on the distance
(in m) between gNB and the user u and it is expressed
as
. The path loss between the gNB and
the users [15] is model as
, where
is given in GHz and is the distance (in m) between gNB
and the user u.
In the system model, slices are differentiated with the QoS
demand (the data rate) of the use cases, while we
differentiate users based on the SLA or contract they
subscribed with their MVNOs. The Signal to Interference
Noise Ratio (SINR) of the user u occupying the cth
{1,2,…,C} sub-channel of the cell in slice sth {1,2,…,S} can
be expressed as:
Where is the noise power and the interference is caused by
the transmissions of gNB to users belonging to other slices
with more transmit power. To protect the user u from this
interference, a maximum tolerable interference limit is set
as.
IV. PROBLEM FORMULATION
Based on Shannon’s capacity, the achievable instantaneous
data rate (i.e. the capacity) on the c C sub-channel of the cell
in slice s S can be given as:
Where is denoted as the constant SINR gap as it is expressed
as and is the bandwidth of the sub-
channel
Also,
is an assignment binary variable which indicates
the allocation of the sub-channel to the user u in slice s during
a scheduling round T.
Also, a PRB must be allocated to exactly one user belonging to
one slice. i.e.
. And each slice is assigned a
minimum number of PRBs according to the agreed SLA
in order to provide services for its users such that .
Where the user’s transmit power should not be more than the
maximum transmit power of its slice. i.e.
.
Therefore, the power metric is set as
.
The total achievable instantaneous data rate of the network
is:
To establish this result and assuming that the optimal
solution is
. Hence, for a given power allocation (
) can
be expressed as follows:
Where is the interference experienced in the sub-channel
is expressed by:
The objective of the joint power and sub-carrier allocation
optimization problem is to minimize the total downlink
transmit power subject to UEs’ QoS requirements, slice UE’s
interference limits and sub-channels orthogonality constraints.
The optimization problem is formulated and derived in (7),
where the objective function is to minimize the total downlink
transmit power and hence the interference of the users in the
cell.
Subject to
Constraint C1 is the data rate constraint for each slice
user. Constraint C2 makes the transmit power on sub-
channel that is not allocated to be zero. C3 limits the number
of assigned sub-channels to the same users that belong to the
same slice. This is to prevent the scheduler from allocating
neighbouring sub-channels to a user of the same slice.
Constraint C4 indicates that the power budget to the user is less
than the maximum power threshold. C5 ensures that the
interference threshold is not exceeded. It puts a limit on the
total interference created by users on sub-channel c, it is only
active if same slice users are allocated neighbouring sub-
Page 34 Southern Africa Telecommunication Networks and Applications Conference (SATNAC) 2019
channels. Lastly in constraint C6,
has been explained
earlier as an assignment binary variable.
The problem (7) is formulated as a Mixed Integer Non-
Linear Program (MINLP) that solves the resource allocation
problem by optimizing the transmit power and as such limits
interference among slice users.
V. PROPOSED ALGORITHM
A. Slicing and Scheduling Scheme
In the proposed scheme, we exploit the frequency diversity
offered by the multipath channel since a deep fade might hit a
substantial number of subcarriers of a given slice. More
importantly, we use a more flexible allocation strategy where
a given slice can select the best subcarriers, which is one with
the highest SINRs. We considered a novel Slicing Carrier
Assignment Scheme (SCAS) where the neighbour sub-
channels are not allocated to users of the same slice when the
transmit power required to achieve the target rate is high
enough to cause inter-user interference. However, there is no
rigid association between the subcarriers and the users. As
shown in Fig 3, a neighbour sub-channels are allocated to slice
1 (mMTC) because the users in the slice only need low transmit
power based on their required data rate. Since the sub-carriers
are at lower data rate and therefore have longer symbol
duration which is more robust against Inter-Symbol
Interference (ISI). Therefore, our proposed SCAS allows
dynamic resource allocation and provides more flexibility by
considering alternating sub-channel allocation for the user
on sub-channel Ci. .i.e. each sub-channel is assigned
exclusively to one user. Mathematically constraints C3 in (7)
assign the neighbour the sub-channel Cj to different slice s if
the power threshold is exceeded in C4 also in (7).
Figure 3: Network Slicing and Scheduling Scheme with multiple MVNOs
This also shows that the Access Point (AP) communicate
with a user using the sub-channel with the least transmit power
possible. Finally, the scheduler and in some cases called the
controller (SDN controller), located at the MAC layer of the
Access Point gNB is to decide on how to assign the PRBs
among users taking into account the QoS requirements and the
channel conditions by solving for Algorithm 2 to determine the
feasible power solution so as to allocate sub-channel(s)
according to Algorithm 1. The assignment of these sub-
channels is updated in each scheduling period.
B. Joint-Power and Sub-carrier Allocation Algorithm
This subsection illustrates the joint power and sub-carrier
allocation algorithm.
Algorithm 1 Joint-Power and Sub-carrier Allocation Algorithm
1. Given the set ´ where the set ´={
}
as the set of sub-channels not allocated to
2. for s = 1:S do
u = 1:U do
3. Initialize =, as the set of sub-channel allocated to
´=
4. Sort sub-channel in the set in descending order
according to metric
5. Sort QoS requirement of in descending order
according to metric
where = (
6. if and
(i.e. 1) then
7. Allocate the highest sub-channels in the set to
(i.e.
)
8. until The QoS request of is met.
9. else
10. Allocate the next sub-channel in the set to the user of
the slice with the next lowest and or till all sub-
channels in the set are allocated
11. end if
12. Update
13. end for
14. if All UEs have their QoS requirement satisfied then
15 Terminate
16. else
17. Given the set ´
18. for s = 1:S do, u = 1:U do
19. Initialize
20. Sort sub-channel in the set in descending order
according to metric
21. Sort QoS requirement of in descending
order according to metric
22. Assign sub-channels such that the next sub-channel is
allocated to UE with least transmit power and QoS
requirement are satisfied.
23. Update
24. end for
25. end if
Algorithm 2 Optimal Interference and Power Allocation
Algorithm
1. Sort the set of available sub-channel in
descending order channel gain
such that
2. Initialize
3.Assume and find
using equation (5)
4. if
or
then
5. Set
and
6. Terminate as the optimal solution is found
7. else
8. Increase
9. Set
10. repeat
11. until A feasible solution is found or in else increase
12. end if
Southern Africa Telecommunication Networks and Applications Conference (SATNAC) 2019 Page 35
VI. PERFORMANCE EVALUATION
In this section, we simulate and evaluate the performance of
our proposed Slicing Carrier Assignment Scheme (SCAS)
using Joint power and Sub-carrier Allocation algorithms.
A. Simulation Environment
TABLE I
SIMULATION PARAMETERS AND VALUES
Parameter
Value
Number of Users/Slice
6 (varies)
Number of Slices
3
Number of MVNOs
2
Number of gNodeB
1
Spectrum allocation
20 MHz
Carrier frequency
2 GHz
Number of subcarriers per RB
12
Size of subcarrier
15KHz
RB bandwidth
12*15kHz = 180 kHz
Number of available RBs
100
Set of sub-channels C’
5 sub-channels (i.e. C’ = 20)
Max gNodeB Tx power
20 W [43dBm]
(Slice 1, 2 & 3)
1Mbps, 2.5Mbps, 0.5Mbps
respectively
(Slice 1, 2 & 3)
0.05W [17dBm], 0.2W
[23dBm], 0.1W
[20dBm] respectively
(Slice 1, 2 & 3)
103N0, 105N0, 104N0, respectively
(Slice 1, 2 & 3)
N0, N0, N0, respectively
Slot duration
0.5ms
Scheduling frame
10ms
Cell-level UEs distribution
Uniform
Antenna height
35m
UE thermal noise density (N0)
10-13W [-100 dBm]
Cell coverage radius
500m
The simulation parameters used for the experiments are as
shown in Table I. Using Matlab [16], we simulate a wireless
5G network with a gNodeB as InP deployed at (0,0) having
several numbers of user uniformly distributed in its coverage
area of radius 500m. These users request for resources
according to the slice they belong to and the assignment of sub-
channels is done based on slice thresholds set for data rate,
transmit power and interference as shown in Table I. The
assignment is regularly updated.
B. Simulation Results.
In this sub-section, we discuss the performance of the
proposed scheme. For comparison, we use equal power
allocation algorithm that shares total signal power equally
among the sub-carriers [17] and this algorithm does not assign
sub-channel in respect to the target power of the slice. We
present the simulation results with different performance
metrics below.
1) Tolerable Interference Level: In Fig 4, we study the sum
of the tolerable interference on the sub-channels with non-zero
power, which is the allocated sub-channels versus the number
of admitted users. Fig 4 shows that as the number of users in
the network increases, the interference also increases.
However, result shows that in the three slices (S1, S2, and S3),
our proposed algorithm performs better than the traditional
algorithm as the level of increase in the interference is minimal
compared to the benchmark algorithm.
Fig 5 shows that effect of our scheduling algorithm. In the
traditional scheme, slices 2 and 3 with the highest and higher
transmit power respectively have significant interference level
on their users. Compared with our proposed scheme, the
interference levels of users in slices with high transmit power
are low especially the slice2.
Figure 4: Sum of tolerable Interference of UEs in Slice S vs Number of
UEs/Slice/MVNO
With few numbers of admitted users, the tradition scheme
provides less interference than our proposed scheme in Slice1,
this is as a result of low transmit power of the few UEs. But as
the number of admitted UEs increases, our proposed scheme
outperformed the traditional method as shown in Fig 4.
M1Slice1 stands for Slice1 in MVNO1.
Figure 5: Sum of tolerable Interference of UEs in Slices (S1, S2, and S3) in
different slices and MVNOs.
2) Transmit Power: In Fig 6, we study the total transmit
power of users in each slice against the total data rate of the
slice. With lower (minimised) transmit power, the proposed
scheme gives the user’s required QoS in terms of minimum
data rate for each slice compared with the benchmarked
algorithm that gives more transmit power across the slices and
little above the required data rate.
3) Number of admitted UEs: Fig 4 and Fig 7 show that as
the number of users increases in the slice and network, the
interference increases and average data rate reduces
respectively. Though at few users (expectedly at low level of
interference), the data rate of the traditional scheme outweigh
our proposed scheme due to high transmit power of the users,
but as the number of admitted UEs increases, our proposed
scheme data rate outrun that of the traditional scheme in all the
slices (S1, S2 and S3).
Page 36 Southern Africa Telecommunication Networks and Applications Conference (SATNAC) 2019
4) QoS of admitted UEs: Fig 8 shows the effect of
interference on the average data rate of the network. Despite
having minimised transmit power, in all the slices (Proposed
S1, S2 and S3), our scheme still allocates resources with a data
rate that is quite more than the target requirement for each user
according to its slice.
Figure 6: Total downlink transmission power vs Total slice data rate.
Figure 7: Number of admitted UEs vs average data rate.
Figure 8: Sum of UEs Interference vs average data rate.
VII. CONCLUSION
In this paper, we have proposed a dynamic resource allocation
(power and sub-channel) and slicing scheduling scheme for
users in a multitenancy 5G sliced networks. The algorithm is
done by formulating an MINLP optimization problem, which
minimizes the total downlink transmit power of the network
subject to the QoS constraints of users, capacity constraints and
interference thresholds. The slicing scheduler accounts for sub-
channels allocation and admission control which takes
infeasibility into account. Through numerical simulation, the
performance of the proposed scheme has been investigated.
Results have shown the improvement of the proposed scheme
over the traditional scheme.
ACKNOWLEDGMENT
The authors acknowledge the support received from Telkom
SA via the Telkom Centre of Excellence (CoE) in Broadband
Networks and application at UCT.
REFERENCES
[1] Cisco Whitepaper, “Cisco Visual Network Index: Forecast and
Methodology, 2016-2021," pp. 1-17, June 5, 2017.
[2] 5G-PPP whitepaper “Use Cases and Performance Evaluation Models,”
pp.1-39, April 2016.
[3] ITU-R, “IMT Vision — Framework and Overall Objectives of the
Future Development of IMT for 2020 and Beyond,” Sept. 2015.
[4] X. Foukas, G. Patounas, A. Elmokashfi, and M. K. Marina, “Network
Slicing in 5G: Survey and Challenges," IEEE Comms. Mag., vol. 55,
no. 5, pp. 94-100, May 2017.
[5] M. Richart, J. Baliosian, J. Serrat and J. Gorricho, "Resource Slicing in
Virtual Wireless Networks: A Survey," in IEEE Transactions on
Network and Service Management, vol. 13, no. 3, pp. 462-476, Sept.
2016.
[6] P. H. A. Rezende and E. R. M. Madeira, "An adaptive network slicing
for LTE radio access networks," 2018 Wireless Days (WD), Dubai,
2018, pp. 68-73.
[7] V. K. Choyi, A. Abdel-Hamid, Y. Shah, S. Ferdi and A. Brusilovsky,
"Network slice selection, assignment and routing within 5G
Networks," 2016 IEEE Conference on Standards for Communications
and Networking (CSCN), Berlin, 2016, pp. 1-7.
[8] A. Devlic, A. Hamidian, D. Liang, M. Eriksson, A. Consoli and J.
Lundstedt, "NESMO: Network slicing management and orchestration
framework," 2017 IEEE International Conference on Communications
Workshops (ICC Workshops), Paris, 2017, pp. 1202-1208.
[9] S. Parsaeefard, V. Jumba, M. Derakhshani and T. Le-Ngoc, "Joint
resource provisioning and admission control in wireless virtualized
networks," 2015 IEEE Wireless Communications and Networking
Conference (WCNC), New Orleans, LA, 2015, pp. 2020-2025.
[10] A. Abdelnasser and E. Hossain, "Resource Allocation for an OFDMA
Cloud-RAN of Small Cells Underlaying a Macrocell," in IEEE
Transactions on Mobile Computing, vol. 15, no. 11, pp. 2837-2850, 1
Nov. 2016.
[11] M. Hu, Y. Chang, Y. Sun and H. Li, "Dynamic slicing and scheduling
for wireless network virtualization in downlink LTE system," 2016 19th
International Symposium on Wireless Personal Multimedia
Communications (WPMC), Shenzhen, 2016, pp. 153-158.
[12] A. Pratap, R. Misra and S. K. Das, "Resource Allocation to Maximize
Fairness and Minimize Interference for Maximum Spectrum Reuse in
5G Cellular Networks," 2018 IEEE 19th International Symposium on
"A World of Wireless, Mobile and Multimedia Networks" (WoWMoM),
Chania, 2018, pp. 1-9.
[13] S. Chadchan and C. Akki, “A fair downlink scheduling algorithm for
3gpp lte networks,” International Journal of Computer Network and
Information Security (IJCNIS), vol. 5, no. 6, p. 34, 2013.
[14] 3GPP, “Further advancements for E-UTRA physical layer aspects
(Release 9),” 3GPP, Nice, France, TR 36.814, Mar. 2010.
[15] O. Grøndalen, A. Zanella, K. Mahmood, M. Carpin, J. Rasool and O.
N. Østerbø, "Scheduling Policies in Time and Frequency Domains for
LTE Downlink Channel: A Performance Comparison," in IEEE
Transactions on Vehicular Technology, vol. 66, no. 4, pp. 3345-3360,
April 2017.
[16] MATLAB, version 9.6.0.1072779 (R2019a). The Mathworks Inc.,
2019.
[17] A. Giovanidis, T. Haustein, E. Jorswieck, and D. Kim, “Maximization
of the single user rate in OFDMA assuming equal power on allocated
subcarriers,” in Vehicular Technology Conference, 2007. VTC2007-
spring. IEEE 65th, April 2007, pp. 2751–2755.
Adedotun T. Ajibare is an MSc Eng. candidate with the Department
of Electrical Engineering in the University of Cape Town.
Southern Africa Telecommunication Networks and Applications Conference (SATNAC) 2019 Page 37
