A Dynamic Algorithm For Interference Management in D2D Enabled Heterogeneous Cellular Networks Modelling and Analysis Rev2

Article
A Dynamic Algorithm for Interference Management in

D2D-Enabled Heterogeneous Cellular Networks: Modelling and
Analysis
Md Kamruzzaman , Nurul I. Sarkar ∗ and Jairo Gutierrez
Department of Computer Science and Software Engineering, Auckland University of Technology, Auckland 1010,
New Zealand, zaman.kamruzzaman@aut.ac.nz (M.K.); jairo.gutierrez@aut.ac.nz (J.G.)
* Corresponding: nurul.sarkar@aut.ac.nz
Abstract: For supporting wider and diverse range of applications, Device-to-Device (D2D) communi- 1
cations is a key enabler in heterogeneous cellular networks (HetCNets). It plays an important role 2
to fulfill the performance and quality of service (QoS) requirements for 5G and beyond networks. 3
D2D-enabled cellular networks enable user equipment (UE) to communicate directly without any or 4
with a partial association with base stations (eNBs). Interference management is one of the critical and 5
complex issues in D2D-enabled HetCNets. Despite of the widely adopted of D2D communications, 6
there are very few researchers address the problems of Mode Selection (MS) as well as resource allo- 7
cation for mutual interference in 3-tier cellular networks. In this paper, we first identify and analyze 8
three key factors, namely outage probability, Signal-to-Interference and Noise Ratio (SINR), and 9
cell density that influence the performance of D2D-enabled HetCNets. We then propose a dynamic 10
algorithm based on a distance-based approach to minimize the interference and to guarantee QoS 11
for both cellular and D2D communication links. Results obtained show that outage probability is 12
improved by 35% and 49% in eNB and SCeNB links respectively when compared with the traditional 13
neighbour-based methods. The findings reported in this paper provide some insights into Interference 14
management in D2D communications that can help network researchers and engineers to contribute 15
further towards developing the next-generation cellular networks. 16
Keywords: Device-to-Device (D2D) communication; heterogeneous networks; HetCNets; interference 17
management 18
1. Introduction 19
Citation: Kamruzzaman, M.; Sarkar,
In cellular networks, Device-to-Device (D2D) communications is an emerging technol- 20
N.I.; Gutierrez, J. Title. Journal Not
ogy where two nearby Users Equipments communicate with each other without any Base 21
Specified 2022, 1, 0. https://doi.org/
Station (BS) or core network support. Due to the short communication range between a 22
Received: D2D pair, D2D communication provides several advantages in terms of spectrum efficiency, 23
Accepted: throughput, latency, power management, coverage expansion, and capacity improvement 24
Published: by reusing radio resources. Furthermore, D2D communication enables new services such 25
Publisher’s Note: MDPI stays neutral as public safety, location-based commercial proximity, content sharing of files, videos or 26
with regard to jurisdictional claims in pictures, gaming, connectivity extension, and traffic offloading [1]. Owing to these benefits 27
published maps and institutional affil- in 5G and beyond networks, D2D communication is a key enabler technology [1]. 28
iations. The growing popularity of high-end user devices and the diversified content of mobile 29
multimedia has contributed, during the last decade, to an exponential growth in both 30
Copyright: © 2022 by the authors.
mobile broadband traffic and end-user demand for faster data access. In addition, the 31
Submitted to Journal Not Specified
for possible open access publication
number of mobile devices and connections are growing exponentially, and by 2022, mobile 32
under the terms and conditions

devices and connections will be 12.3 billion compared to 8.6 billion in 2017 [2]. As per 33
of the Creative Commons Attri- recently published Cisco visual networking index [2], mobile data traffic has increased 34
bution (CC BY) license (https:// 18-fold, from 400 petabytes to 7.2 exabyte per month during 2011-2016. Expecting further 35
creativecommons.org/licenses/by/ tenfold growth, reaching 77 exabyte per month by 2022. Moreover, with enhanced mobile 36
4.0/). broadband (eMBB), ultra-reliable and low latency communication (URLLC) and massive 37
Version January 28, 2022 submitted to Journal Not Specified https://www.mdpi.com/journal/notspecified

Version January 28, 2022 submitted to Journal Not Specified 2 of 20
machine type communications (mMTC), different services are surely forthcoming for 5G 38
and beyond networks [3]. Managing such a high user density and resulting immense data 39
volumes is a major concern for the cellular network operators [1]. 40
Furthermore, in the 3rd Generation Partnership Project (3GPP) Release 12 for proximity- 41
based services (ProSe) and group communication system enablers (GCSE) D2D communi- 42
cations is an integrated module in the Long Term Evolution (LTE-A) Advanced standard 43
[1]. During natural disasters such as earthquakes or hurricanes, a replacement of the 44
traditional network can be set up quickly with the help of D2D functionality. In addition, 45
multi-hop cooperation between devices can help to enhance the coverage since at those 46
times D2D may be the only mode of communication in no coverage areas. Hence D2D 47
communication in cellular networks will bring significant performance gains in terms of 48
data offload (due to direct communications), improved spectrum efficiency (due to reuse 49
cellular resources), coverage extension (by providing improved connectivity among UEs) 50
and content sharing/dissemination [4], [5], [6]. 51
However, to maximize the benefits of D2D communications, there are many open 52
challenges that need to be thoroughly addressed [6]. These challenges include mode 53
selection, neighbor discovery, interference and radio resource management, energy con- 54
sumption, coexistence of D2D with small cells, mobility management, and network security 55
etc. Among them, interference management (IM) in a heterogeneous scenario comparing 56
all tiers simultaneously is very important and complex [1]. 57
Figure 1. Interference scenarios in D2D-enabled 3-tiers cellular network.
Fig.-1 shows the possible interference scenarios in a 3-tier cellular network where eNB, 58
SCeNB and D2D pairs will reuse the cellular resources to communicate simultaneously 59
and introduce mutual interference among different tiers. Hence, to achieve the benefits 60
of D2D communication in cellular networks, it is essential to manage interference, and 61
by selecting an appropriate mode of transmission, power control, resource allocation, 62
antenna systems and location restriction, we can achieve this. Based on spectrum usage, 63
the main two categories of D2D communications are: (1) D2D overlay where a dedicated 64
orthogonal spectrum is used for D2D communications within conventional cellular users 65
in a cell [7]; and (2) D2D underlay where conventional cellular spectrum will be shared 66
with D2D communications which leads to better spectrum utilization at the cost of complex 67
interference scenarios [3]. 68
In this paper, we identify various key factors that contribute to interference in an 69
underlay heterogeneous cellular network where up link (UL) resources are shared among 70
D2D pairs and small cells. The main contribution and strength of this paper is the emphasise 71
on the fact that a dynamic algorithm is required to handle interference management in 72
D2D-enabled heterogeneous cellular networks. 73
The main contributions of the paper are as follows. 74
• We propose a dynamic algorithm called Acceptance Interference Region (AIR) to 75
provide a solution to the problem of achieving a strict QoS guaranteed for all links in 76
D2D-enabled heterogeneous cellular networks. A distance-based approach is used to 77
achieve guaranteed links quality. The proof of AIR is provided in Appendix A. 78
• We propose an efficient ON/OFF algorithm to provide a solution to the problem of 79
achieving a maximum transmission capacity in the network. 80
• We develop a mathematical model containing the network, SINR, and small cell 81
density models for system performance modelling and analysis. To this end, we derive 82
the outage probabilities of D2D link, macrocell link, and small cell link for system 83
performance analysis. We also provide as analysis and proof (see Appendix B) to show 84
how a small cell density and the number of D2D pairs affect the communication link 85
quality. We validate our analytical models using a MATLAB-based simulation. 86
The rest of this paper is organized as follows. Section 2 reviews the literature on 87
interference management in D2D-enabled heterogeneous cellular networks. Section 3 88
presents the system model, including network, Signal-to-interference-plus-noise ratio 89
(SINR), and small cell density models. The proposed AIR dynamic algorithm is also 90
presented in this section. Section 3 also presents the theoretical analysis covering the 91
coverage probability and spectral efficiency. The system performance is evaluated in 92
Section 4. The simulation results are also presented in this section. Conclusions are drawn 93
in Section 5. Table-1 lists the key mathematical notations and abbreviations used in this 94
paper. 95
Table 1. List of key notations and abbreviations used
Notation Definition
Φ PPP constituted by the macro BSs (eNB)
Φs PPP constituted by the Small Cell BSs (SCeNB)
Φd PPP constituted by DUEs
λs Intensity of SCeNBs
λd Intensity of DUES
Pc Transmission power of a typical UE operating in cellular mode
Pd Transmission power of a typical UE operating in D2D mode
Ps Transmission power of a typical SUE
NC , ND , NS The set of CUEs, DUEs and SUEs
α Path loss exponent
γth Required SINR for cellular and D2D links
N0 Thermal noise
h x,y Channel coefficient between x and y
D2D Device-to-Device
SINR Signal-to-noise-plus-interference ratio
UE User Equipment
CUE Cellular user equipment
DUE Device-to-Device user equipment
MS Mode Selection
PPP Poison point process
SUE Small cell user equipment
eNB Evolved Node B i.e LTE macro base station
SCeNB Small cell evolved Node B i.e small cell
AIR Accepted Interference Region
QoS Quality of Service
2. Related Work 96
In a D2D-enabled heterogeneous cellular network, to manage interference the state- 97
of-the-art researchers have proposed different methods and their working principles and 98
techniques were different also. Based on interference behavior all of them can be categorized 99
as interference avoidance, interference cancellation and interference reduction. 100
In interference avoidance techniques, DUEs (D2D User Equipment) are not allowed to 101
use the cellular resources up to a certain area of coverage which is defined by a predefined 102
signal-to-interference-noise ratio (SINR) and the interference limited area is normally 103
derived with a specific transmissions power level at the UEs [8], or based on the distance 104
as well as channel quality of the UEs [9]. In [3], to make a robust cellular link and to 105
reduce interference between the cellular and D2D layers, an interference aware power 106
allocation (IAPA) solution is proposed to improve spectral efficiency, throughput and 107
outage probability of cellular links. An analytical framework for MS is developed for 2-tier 108
cellular network by using Markov-chain theory in [10] for a single user scenario and finally 109
a joint MS, resource allocation, and scheduling optimization is formulated using greedy 110
heuristic algorithm. In [11] a deep reinforcement learning based dynamic spectrum access 111
scheme for D2D communication in heterogeneous cellular network is proposed using the 112
location of CUEs (Cellular User Equipment) where links QoS is guaranteed. In [12] authors 113
first derived the outage probabilities of D2D, macro cell and small cell links, and then 114
presented the lower distance boundaries from D2D transmitter to macro base station (eNB) 115
and small cell receiver, and from base stations receiver to a D2D receiver. Finally, an optimal 116
small cell deployment density that ensures the quality of service (QoS) requirements for 117
both D2D and cellular communications is proposed. In [13], the authors proposed an 118
interference limited area (ILA) based D2D management scheme where an appropriate 119
power control algorithm is used for mitigating interference. In their work, multiple DUEs 120
but single macro cell is considered and DUEs can only initiate their communication outside 121
a restricted area. Like [12], lower distance boundaries between D2D receiver and cellular 122
cells (both macro and small cell) are defined, and based on that interference management 123
scheme and upper boundary, small cell density is proposed in which D2D devices can only 124
reuse the cellular resources effectively to meet QoS requirements [14]. 125
Whereas in interference cancellation techniques- multiple antenna systems, beam- 126
forming or pre-coding are used. The major advantages of these approaches is to manage or 127
eliminate the interference without reducing the transmission power. Hence transmission 128
rate is improved but at the cost of additional computation power and communication 129
overheads. In [15] an overlay mode of D2D communication in mm-wave 5G networks 130
is proposed where alternate offer bargaining game theory is used to improve system 131
throughput without compromising SINR. 132
But in interference coordination approach, CUEs and DUEs transmit power as well as 133
channel assignment being optimized to maximize the objectives. However, it often requires 134
the centralized computation at the eNB. In addition, to reduce the system complexity, 135
D2D pairs can reuse the radio resources from only one CUE regardless of other available 136
CUEs with better channel conditions. In [16] both line-of- Sight (LoS) and non-line-of- 137
Sight (NLoS) transmissions are considered in a practical path loss model and the authors 138
proposed mode selection techniques based on the maximum received signal strength for 139
each UE to control D2D-to-cellular interference. To minimize the impact of interference in 140
D2D-enabled cellular network, in [17] the authors propose a distributed algorithm using 141
matching theory to allocate the appropriate resources. The researchers first modelled the 142
spatial distributions of UEs and BSs in HetCNets using homogeneous poison point process, 143
and finally proposed resource allocation algorithm where the QoS of D2D communications 144
is guaranteed. Similar to [13], to minimize the interference in D2D-enabled HetNets cellular 145
networks, a joint power control and mode selection scheme is proposed where power 146
control is used to adjust the interference limited area dynamically and mode selection 147
is used to maximize the spectral efficiency [18]. In addition, to improve performance 148
and mitigate in-band emission interference (IEI) for cellular links in D2D-enabled cellular 149
network, Albasry et al. [19] propose a distance-density based (DDB) frequency resources 150
grouping strategy and optimal power allocation (OPA) algorithm. DDB strategy is used to 151
give the higher priority on the QoS for D2D links whereas OPA is used when the number 152
of D2D links is more important. Albasry et al. use stochastic geometry and an analytical 153
approach for their modelling in 2-tier networks. 154
In [20] Hassan et al. propose a weighted bipartite matching algorithm to minimize 155
the interference in a 2-tier D2D-enabled cellular network and later on use local search 156
techniques to improve further their outcomes. Finally, Hassan et al. compare their pro- 157
posed algorithms with a two-phase auction based fair and interference aware resource 158
allocation algorithm to show the improvement of their algorithms. Huynh et al.in [21] 159
propose an interference management algorithm to maximize the performance of D2D 160
communication without compromising the QoS requirements for cellular link in both UL 161
and DL by optimizing admission control, power control and resource allocation. The main 162
disadvantage of their work is resource sharing with multiple D2D pairs is not considered 163
and their analysis is limited to a 2 tier network. 164
In [22] Chen et al. proposed two different resource sharing strategies for co-tier and 165
cross-tier interference separately along with their respective power control mechanisms. To 166
optimize the problem, here convex optimization and 0-1 assignment problem techniques 167
are used for power control and resource allocation respectively. But in [23] a concatenated 168
bi-partite matching (CBM) method is proposed to mitigate interference by appropriate sub- 169
band assignment (SA) and resource allocation (RA). In their proposal, user equipment (UE) 170
density, e-node-B (eNB) density, and the switching frequency of small cells are adaptively 171
determined, and the effect of UL power control is managed by full and truncated channel 172
inversion methods. Initially, CBM is developed on single matching which is eventually 173
generalized to multiple cells for SA and RA DUEs. By using Monte Carlo methods, an 174
iterative scheme is proposed where mode selection and power control are used jointly. 175
The interference on each RB (Resource Block) is measured and the transmission power is 176
adjusted to achieve the targeted throughput without affecting the CUEs performance. 177
In [24], the proposed scheme is only validated using a single cell with different radius. 178
The state-of-art-researchers proposed another smart mode and power selection based 179
distribution approach in [25] where real-time information of local traffic channels and 180
surrounding nodes information are used. Here dynamic switching is adopted to control 181
interference otherwise continue the communication using cellular mode. Moreover, to 182
resolve the issues arising from spectrum sharing in D2D enabled cellular networks, a 183
location related strategy for mode selection and a spectrum sharing algorithm is proposed 184
in [26] where devices form a coalition to share the spectrum among DUEs and CUEs. 185
Most of the aforementioned state-of-the-art techniques focus on interference manage- 186
ment in single or 2-tier cellular networks. Moreover, for interference management most of 187
the researchers consider only one D2D pair and the effects of macro and multiple small 188
cells are neglected, and very few papers focus on where DUEs in a D2D pairs are attached 189
to a different eNBs or one in eNB and another in small cell. In 5G and beyond, ultra dense 190
networks (UDNs) will be deployed, and in practice multiple CUEs may associate with 191
different cellular layers, and additional D2D layers will create the most prominent challenge 192
on minimizing interference. Despite the prominent usage of D2D communications, there 193
are very few on-going researchers who are working to address the MS as well as resource 194
allocation for mutual interference problems in 3-tier cellular networks. Nonetheless, none 195
of the authors reviewed the worst and, at the same time, the most challenging interference 196
case, where all three-tiers of the networks (i.e. eNB, SCeNB and D2D communication) 197
mutually interfere [1] and the effect of multiple SCeNBs and D2D pairs are considered. 198
This is the main difference between the existing approaches and our proposal for managing 199
interference in a 3-tier heterogeneous cellular network. 200

3. System Model 201
3.1. Network Model 202
For modelling a 3-tier D2D-enabled heterogeneous cellular network, we consider a 203
macro cell (eNB) at the centre of the coverage area with radius R which is surrounded by 204
several small cells (SCeNBs) and D2D pairs. Small cells are randomly distributed within the 205
macro cell (eNB) coverage area. Due to random and unpredictable location of small cells, the 206
spatial position of the small cells is modelled by using a homogeneous PPP ϕs with density 207
λs , and DUEs are also distributed in the network region according to another independent 208
homogeneous PPP ϕd with density λd . Here DUEs, SCeNB and evenly distributed CUEs 209
are denoted by j ∈ {1, 2, . . . , ND }, k ∈ {1, 2, . . . , NS } and i ∈ {1, 2, . . . , NC } respectively. 210
Moreover, for modeling large scale wireless networks and capturing the effects of network 211
topology on network performance stochastic geometry is more suitable[27]. In a 3-tier 212
network shown in Fig-1, each UE can communicate in any one of the following modes: 213
(1) DUEs can communicate directly without base stations using the D2D communication 214
mode, (2) CUEs can communicate with each other through the eNB, and this is known as 215
the macrocell or cellular communication mode, and (3) SUEs can communicate with each 216
other through the SCeNB, and this is known as the small cell communication mode. To 217
mitigate the intra-cell interference (between UEs within the same cell) cellular resources are 218
assigned orthogonally and each cellular UE uses separate RBs. Co-channel interference can 219
be limited by allowing only one D2D link to share the resource of a cellular link at a time. 220
We assume all channel gains are independent of each other, independent of the spatial 221
locations, symmetric, and are identically distributed (i.i.d.). For simplicity of analysis, 222
only a Rayleigh fading environment is considered and channel coefficients are assumed 223
to be exponentially distributed. In such a D2D-enabled HetCNets, the channel model is 224
composed of large-scale path loss and small Rayleigh fading, so in general the received 225
signal can be expressed as[14]: 226
Pr = Pt h xy D −α (1)
Where Pt is the transmission power, α is the path loss exponent, D is the distance between 227
the transmitter and the receiver, and h xy is channel coefficient for that particular link. 228
A receiver can decode a message successfully if and only if the SINR at the receiver is 229
greater than a specific threshold γth . If the SINR at the receiver does not meet γth , the link 230
experiences an outage. So the outage probability of the x,y link can be expressed as: 231
Pout = Pr {γy ≤ γth } (2)

Where Pr (.) is the outage probability for a minimum SNIR threshold γth . 232
Here e, s, i, j and k subscripts are used to denote the serving eNB, the serving SCeNB, 233
the ith CUE, the jth D2D pairs and the k th SUE, respectively. And the subscripts t and r are 234
used to denote the transmitter and the receiver of D2D pair respectively. In the context of 235
the above defined network where UL cellular resources are shared by D2D pairs and small 236
cells, the mutual interference at different receiver can be expressed as: 237
ND NS
Ii = ∑ Pj h j,e d−j,eα + ∑ Pk hk,e d−k,eα + N0 (3)
j =1 k =1
ND NS
−α
Ij = Pi hi,r di,r + ∑ Pj′ h j′ ,r d− α
j′ ,r + ∑ Pk hk,r d−k,rα + N0 (4)
j′ =1,j′ ̸= j k =1
ND NS
−α
Ik = Pi hi,s di,s + ∑ Pj h j,s d−j,sα + ∑ Pk′ hk′ ,s d− α
k′ ,s + N0 (5)
j =1 k′ =1,k′ ̸=k
Where Ii is the combined interference received by eNB, Ij is the same for jth D2D pairs 238
receiver other than jth transmitter and Ik is the same for all SUEs except k th to the SCeNB. 239
3.2. SINR Model 240
In wireless communication, the SNIR is measured as the ratio of the received power by 241
the receiver to the total interference including spectral noise density. Since communications 242
may take place in any of the early mentioned 3 cases, the SINR at the jth D2D receiver is 243
given by: 244
Pj ht,r d−
t,r
α
Pj ht,r d−
t,r
α
γ jD = = (6)
Ij −α
ND NS
Pi hi,r di,r + ∑ Pj′ h j′ ,r d− α
j′ ,r
+ ∑ Pk hk,r d− α
k,r + N0
j′ =1,j′ ̸= j k =1
So according to eq.(2) the outage probability of the jth D2D link can be given as 245
D
Pout,j = Pr {γ jD < γth } = 1 − Pr {γ jD ≥ γth }
where γth is the required SINR threshold at the receiver for effective D2D communication. 246
 

 

Pj ht,r d−
 

 α 

D t,r
Poutj = 1 − Pr NS
≥ γth
ND
−α
Pj′ h j′ ,r d− Pk hk,e d−
 
 Pi hi,r di,r + ∑ α
+ ∑ α
+ N0
 
j′ ,r
 
 k,e 
j′ =1,j′ ̸= j

k =1
  
 γth dαt,r ND NS 
= 1 − Pr ht,r ≥  Pi hi,r d−α + ∑ Pj′ h j′ ,r d−′ α + ∑ Pk hk,r d−α + N0 
Pj i,r j ,r k,r
j′ =1,j′ ̸= j k =1
 
Since channel coefficient is exponentially distributed, the expectation of interference 247
from the above equation can be expressed as follows 248
  
γth dαt,r ND NS
D
Poutj = 1 − Eexp−  Pi hi,r d−α + ∑ Pj′ h j′ ,r d−′ α + ∑ Pk hk,r d−α + N0 )
Pj i,r j ,r k,r
j′ =1,j′ ̸= j k =1
(7)
Where E(.) is expectation function and D
Poutj is the outage probability of of jth DUEs with 249
γth SNIR threshold. 250
Here channel quality follows the Rayleigh fading assumption which is an exponentially 251
γth dαt,r
distributed random variable. Assume z = and L Id (z) and L Is (z) are the Laplace
Pj 252
transformation of random variables Id and Is evaluated at z, respectively. Interference due 253
to same cellular resource reuses by other D2D pairs and small cell links Id and Is are defined 254
ND NS
as Id = ∑ Pj′ h j′ ,r d− α
j′ ,r
and Is = ∑ Pk hk,r d− α
k,r respectively. So eq.(7) can be written as 255
j′ =1,j′ ̸= j k =1
D
Pout,j = 1 − exp(−zN0 )L Id (z)L Is (z)
! !
− N0 γth dαt,r ) α
Pj,r di,r γth dαt,r γth dαt,r
= 1 − exp L Id ( )L Is ( )
Pj Pi γth dαt,r + Pj di,r
α Pj Pj
−κPj′ m γth m d2t,r λd

! ! !
Pj di,r
= 1 − exp exp
α Pj m
!
−κPk m γth m d2t,r λs
exp
Pj m
−κγth m d2t,r ( Pj′ m λd + Pk m λs )

! ! !
Pj di,r
= 1 − exp exp
α Pj m
= 1 − δD exp(− β D ( Pj′ m λd + Pk m λs )) (8)

− N γ dα ) α
Pj di,r

0 th t,r
Where E(.) is expectation function, δD = exp Pj Pγ α α , 256
i th dt,r + Pj di,r
κ = πmΓ(m)Γ(1 − m), β D = κγth m d2t,r /Pj m ,
λd is density of DUEs for D2D pairs, λs is 257
density of small cells and the proof of the above equation can be referred to Appendix A in 258
[28]. 259
From the above expression it’s clearly visible that the outage probability of D2D links 260
depends on various factors such as path loss coefficient, required SINR, distances between 261
UEs, transmission powers, small cell and D2D pairs density. Outage probability increases 262
with increasing required SINR but decreases while the distance between SUEs and the D2D 263
receiver is increased. 264
Similarly, in the case of macro cellular communications mode where UE is served by 265
eNB, the SINR at the receiver i can be expressed as: 266
−α
Pi hi,e di,e
γiM =
Ii
Hence the outage probability of the macro cell link can be written as 267
M
Pout,i = Pr {γiM < γth } = 1 − Pr {γiM ≥ γth } (9)
Where γth is the required SINR for ith CUE for effective cellular communication, after 268
substituting the SINR values the outage probability of macro cellular link will be as follows 269
 

 

−α
 
P h d
 
i i,e i,e
 
M
Pout,i = 1 − Pr ≥ γth
ND NS
 ∑ Pj h j,e d−α + ∑ Pk hk,e d−α + N0

 



 j,e k,e 

j =1 k =1
( !)
α
γth di,e ND NS
= 1 − Pr hi,e ≥
Pi ∑ Pj h j,e d−j,eα + ∑ Pk hk,e d−k,eα + N0
j =1 k =1
" !!#
α
γth di,e ND NS
= 1 − E exp −
Pi ∑ Pj h j,e d−j,eα + ∑ Pk hk,e d−k,eα + N0
j =1 k =1
α
! α
! α
!
N0 γth di,e γth di,e γth di,e
= 1 − exp − L Id L Is
Pi Pi Pi
 
κγth m di,e 2 Pjm λd + Pkm λs
α
!
N0 γth di,e
= 1 − exp − exp− 
Pi Pim
= 1 − δM exp(− β M ( Pj m λd + Pk m λs )) (10)
N0 γm dα κγth m d2
Where δM = exp(− Pth i,e ), and β M = P m i,e . Hence the outage probability de- 270
i i
creases with increasing distances between SUEs, and UEs as well as UEs and D2D transmit- 271
ters. 272
For small cell mode communications, UE is served by small cell SCeNB, and the link 273
between SUE and SCeNB will be interfered by other SCeNBs, D2D pairs and macro eNB. 274
Therefore, the SINR at the small cell receiver can be written as 275
γk S = ( Pk hk,s d− α
k,k ) /Ik
and the outage probability of the small cell link can be expressed as
S
Pout,k = Pr {γiS < γth } = 1 − Pr {γiS ≥ γth } (11)
So, after substituting all the values the outage probability of small cell cellular link can be 276
derived as 277
 

 

Pk hk,s d−

 α


 
S k,s
Pout,k = 1 − Pr NS
≥ γth
ND
−α −α
Pk′ hk′ ,s d−
 
 Pi hi,s di,s + ∑ Pj h j,s d j,s + ∑ α
+ N0
 
k′ ,s

 

k′ =1,k′ ̸=k

j =1
( !)
γth dαk,s ND NS
= 1 − Pr hk,s ≥ −α
Pi hi,s di,s +∑ Pj h j,s d−
j,s
α
+ ∑ Pk′ hk′ ,s d− α
k′ ,s + N0
Pk j =1 k′ =1,k′ ̸=k
" !!#
γth dαk,s ND NS
= 1 − E exp − −α
Pi hi,s di,s +∑ Pj h j,s d−
j,s
α
+ ∑ Pk′ hk′ ,s d− α
k′ ,s + N0
Pk j =1 k′ =1,k′ ̸=k
! ! ! !
N0 γth dαk,s Pk dk,e α γth dαk,s γth dαk,s
= 1 − exp − L Id L Is′
Pk Pi γth dαk,s + Pk dk,e α Pk Pk
 
κγth m dk,s 2 Pjm λd + Pkm′ λs
! !
Pk dαk,e N0 γth dαk,s
= 1− exp − exp− 
Pi γth dk,s α + Pk dk,e α Pk Pkm
= 1 − δS exp(− β S ( Pj m λd + Pk′ m λs )) (12)

Pk dαk,e N0 γth dα κγth m d2
Where δS = Pi γth d α +P d α
exp − P k,s ,and β S = P m k,s , and Is′ is the inter- 278
k,s k k,e k k
ference from all SCeNBs except respective small cell links. 279
According to the above expressions of outage probabilities it is easily visible that 280
intensity of interference (i.e probabilities of success) depends on the density of small cells 281
and D2D pairs, the distances between the receiver and transmitter of those links, required 282
SINR threhold and transmission powers. Increasing DUE receiver distances from CUEs 283
or SUEs will increase the success of the probability for D2D links. Similarly increasing 284
distances between D2D transmitter and CUEs or SUEs will decrease the outage probability 285
of SceNB links. 286
Based on the previous expressions we propose Algorithm 1 (see below) called Accept- 287
able Interference Regions (AIR) where links QoS for various communications modes will 288
be guaranteed by limiting the coexistence of different UEs under D2D link constraints as 289
follows: 290
dijD ≥ dijD min

(13a)
M
d je ≥ d jeM min
(13b)
dSjs ≥ dijS min

(13c)
Algorithm 1: The proposed AIR dynamic algorithm

1 Initialization: Cellular resources are not used by another D2D pair
2 Arbitrarily generate the numbers of UEs and determine their locations
3 Release the cellular resources if used by D2D pairs and small cells
4 Randomly select cellular resources Cij for reuse
5 eNB calculates the distances dij , d je , d js and d∗min
6 Compare distance with their corresponding thresholds
7 while if dij ≥ dminD do
8 D2D pairs reuse Cij resource

9 if d js ≥ dSmin then
10 SCeNB reuses cellular resource Cij
11 else
12 Explore for another D2D pair
13 Use macro cellular mode
Where dmin is the lower boundary which must be satisfied to maintain the QoS for various 291
links and it can be measured as shown in Appendix A. For any communications, UEs will 292
fulfill the minimum distances requirement to guarantee the link QoS. 293
3.3. Small Cell Density Model 294
Here both small cells and UEs are considered to be distributed based on separate 295
Poisson Point Processes. 296
2500
2000
1500
1000
500
0
0 500 1000 1500 2000 2500
Figure 2. Poison point process distributions of User Equipment - diamonds represent the eNB, circles
represent UEs (green circles are DUEs) and asterisks represent SCeNBs in a cellular network.
Previous analysis indicates that mutual interference is heavily dependant on small cell 297
and D2D pairs densities. Small cell density, γs beyond a certain threshold causes excessive 298
interference for a D2D communications resulted in no solution for the AIR feasible set, and 299
one or more links will fail to satisfy QoS requirements. Contrarily, lower values of γs yields 300
a higher feasible DUEs. Nevertheless, lower γs may result in a smaller overall transmission 301
rate for small cell UEs. In general, the transmission rate of small cell UE is defined as the 302
number of successful transmissions per unit area [29]. So the transmission capacity of small 303
cells can be expressed as 304
Tc = λs (1 − Po utS ) = λs δS exp(− β S ( Pj m λd + Pk′ m λs )) (14)

And an optimum problem for the above scenario can be formulated as: 305
Max Tc = λs δS exp(− β S ( Pj m λd + Pk′ m λs )) (15)

An optimum value for small cell density, γs can be obtained by maximizing the above 306
equation (17) for small cell with or without satisfying the QoS requirements for the various 307
links. So the optimum solution for the above equation without considering QoS constraints 308
is as follows: 309
1 Pkm
λes = =
Pkm′ β S m
Pk′ γth d2k,s
Therefore, under fixed transmission power conditions, a higher no. of SCeNBs can be added 310
into the network for sharing cellular resources by reducing either the distance between 311
SCeNB and SUE or the required SINR threshold. The proof of the solution for this optimal 312
problem is shown in Appendix B. 313
Similarly for D2D pairs density, we can get 314
1 Pjm
fd =
λ =
Pjm′ β D Pjm′ γth d2t,r
From previous analysis of outage probabilities for DUEs, CUEs and SUEs, it is very 315
hard to avoid the monotonically increasing nature of success probabilities with increasing 316
small cell density,λs , or D2D pairs density, λd or both. But increasing λs or λd introduces 317
additional interference. Hence to obtain an optimum value for small cell density, λs by 318
fulfilling QoS requirements for all communications mode eq (18) has to satisfy the following 319
constraints. 320
D
Pout ≤τ (16a)
M
Pout ≤τ (16b)
S
Pout ≤τ (16c)
Where τ is the maximum allowable outage probabilities for any links. As controlling no. of 321
small cell is easier compared to DUEs, according to the above constraints in eq. (16), the 322
density of small cell has to satisfy the following: 323
λs ≤ min{ f (τ ), g(τ ), h(τ )} (17)

Where 324
D
f (τ ) = argmax λs >0 { Pout,j ≤ τ}
M
g(τ ) = argmax λs >0 { Pout,i ≤ τ}
S
h(τ ) = argmax λs >0 { Pout,k ≤ τ}
Hence by considering QoS constraints in eq.(16a)-(16c), we can obtain the solution for small 325
cell density, λs for various communications modes as follows: 326

1 δ
λSs ≤ m ln S − ρmjk′ λd
β S Pk′ 1 − τ
1 δ
λsM ≤ m ln M − ρmj′k λd
β M Pk 1−τ
1 δ
λsD ≤ ln D − ρmjk λd
β D Pkm 1 − τ
The proof for this solution is shown in Appendix C. Here λSs , λSs , λSs are the small cell 327
density for small cell, macro cell and D2D mode communications, and ρ is the ratio of 328
transmission powers. Hence the arguments of eq.(17) are as follows: 329
1 δ
f (τ ) = m ln( S ) − ρmjk′ λd
β S Pk′ 1−τ
1 δ
g(τ ) = m ln ( M ) − ρmj′k λd
β M Pk 1−τ
1 δ
h(τ ) = m ln( D ) − ρmjk λd
β D Pk 1−τ
Therefore, the outage probability constraints of optimal problem in eq. (17) can be repre- 330
sented to the constraints of small cell density as follows: 331

1 δ 1 δ 1 δ
λs ≤ min { m ln( S ) − ρm
jk ′ λ d }, { m ln ( M ) − ρm
j ′ k λ d }, { m ln( D ) − ρm
jk λd } = λsMax
β S Pk′ 1−τ β M Pk 1−τ β D Pk 1−τ
Where λsMax is the maximum allowable small cell density to guarantee the links QoS. Hence 332
to maximize the transmission capacity of small cell we can propose transmission ON-OFF 333
algorithm for small cells. 334
Algorithm 2: ON-OFF Algorithm for Interference minimization

1 Initialize small cell density λs
2 Randomly select an SCeNB and calculate the distance d js
3 Calculate the maximum allowable small cell density λsMax
4 Compare the ratio of λsMax and λs
λsMax
5 if < 1 then
λs
6 Turn OFF the small cells
7 else
8 Turn ON
9 end
4. Performance Evaluation 335
4.1. Simulation environment and parameters 336
The system model is validated using MATLAB-based simulation. Table-2 lists the 337
parameters used in the simulation for the D2D-enabled cellular HetNets. In our simulation 338
setup, we consider a single eNB with a 500m cell radius located at the centre of D2D-enabled 339
HetCNets where CUEs are randomly distributed. DUEs and SUEs are realized according 340
to two independents PPP with densities of λd and λs respectively. The number of CUEs 341
is selected in such a way that the saturation condition is always satisfied. Our analytical 342
model is only valid under the assumptions that each eNB has at least one user to serve in 343
the uplink. We evaluate the coverage probability of the proposed scheme with an average 344
of 10000 independent realizations. 345
Table 2. Parameters used in the simulation
Simulation Parameters Values

Intensity of SCeNBs, λs 10−5
Intensity of DUES, λd 10−3
CUE Transmission power, Pi 23 (dBm)
Transmission power of DUEs, Pj 20 (dBm)
Transmission power of SUEs, Pk 20 (dBm)
Path loss exponent, α 3 and 4
Required SINR threshold, γth -2.6 dB
Noise power, N0 -118 dBm
Cellular UE numbers,NC 150
D2D pairs numbers, ND 50
Small cell UE numbers, NS 250
Maximum allowable outage ie. QoS, τ 0.1,0.01 and 0.001
Channel Bandwidth, B 180KHz
4.2. Results and discussion 346
The proposed model is based on the distance-based mode selection strategy without 347
power control. As DUEs are distributed based on the PPP function, and higher distances 348
between D2D receiver and transmitter are more likely to be occurred which eventually 349
increases the transmission power and associated interference. Hence the proposed scheme 350
with an added power control mechanism can be part of our future work. 351
Due to the limited related published work in the field, the proposed AIR algorithm 352
and the resulting scheme is validated by comparing it with the traditional neighbour- 353
based scheme where information is transmitted via D2D transmission to the targeted 354
neighboring UEs and the work presented in[12] where D2D transmission reuse resources 355
only if D2D link satisfy the given QoS requirements with guaranteed CUEs transmission. 356
Initially Cumulative Distribution Functions (CDFs) of outage probabilities of all three 357
communications modes are evaluated and compared with the above mentioned baseline 358
schemes. 359
0.9
0.8
0.7
0.6
CDF 0.5
0.4
0.3
0.2
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Outage Probability of D2D link
Figure 3. Outage probability of D2D links with different schemes
In Fig.-3, we plot CDF versus the outage probability of the D2D link. We observe that 360
the outage probability of the D2D link is improved up to 55% using the proposed scheme 361
when compared with the neighbour-based scheme. However, the improvement is marginal 362
(up to 3%)when contrasted with the scheme proposed in[12] where the interference is 363
considered for single SCeNB and D2D pairs. It is also clearly visible that the outage 364
probability decreases with the decrease of τ (QoS requirements for D2D links) and more 365
than 88% of the D2D links can meet the QoS requirements in this scenario. 366
0.9
0.8
0.7
0.6
CDF
0.5
0.4
0.3
0.2
0.1
0
0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
Outage Probability of macro link
Figure 4. Outage probability of macro links with various schemes

0.9
0.8
0.7
0.6
CDF 0.5
0.4
0.3
0.2
0.1
0
0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
Outage Probability of Small Cell links
Figure 5. Outage probability of small cell links with various schemes
In Figs.-4 and 5, macro and small cell outage probabilities are compared with the 367
same baseline. Similar to the D2D link outage case, outage probability improvement is 368
noticeable in our proposed scheme. However, the improvements are not as prominent as 369
those in D2D links and the reason behind this is that D2D transmitters are chosen based 370
on the AIR scheme which is designed for minimizing the D2D interference. In addition, 371
due to smaller path loss, the channel quality of D2D links is much better compared with 372
the macro and small cell links. Despite all the factors, here we can see at CDF = 0.7 the 373
outage probability of the feasible set AIR scheme is much better when compared to the 374
neighbour based scheme (35% in eNB and 69% in SCeNB links) and considerably better 375
when compared to the [12] scheme (6% in eNB and 8% in SCeNB links). 376
0.9
0.8
Outage Probability
0.7
0.6
0.5
0.4
0.3
0.2 D2D links

SCeNB links
eNB links
0.1
-10 -8 -6 -4 -2 0 2 4 6 8 10
SINR Threshold (dBm)
Figure 6. Outage probability of different links with various SINR threshold
From Fig.-6 we observe that the outage probability increases with increase in SINR 377
threshold requirements, and at lower SINR threshold values, the cellular coverage is nearly 378
perfect i.e. cellular outage is almost zero. By controlling interference we can minimize the 379
required threshold for decoding the message and hence improve the performance in terms 380
of links availability. Figs.-7 and 8 also reveal the impact of densities of D2D pairs and small 381
cells on the links availability. In both cases outage probability increases with increasing 382
D2D pairs and small cell densities. 383
1
D2D links
Cellular links
0.95 SCeNB links
0.9
0.85
Success Probability (%)
0.8
0.75
0.7
0.65
0.6
0.55
0.5
0 10 20 30 40 50 60 70 80 90 100
# of small cell
Figure 7. Success probability with small cell density.
SCeNB links
0.9 D2D links
Cellular links
0.8
0.7
Success Probability (%)
0.6
0.5
0.4
0.3
0.2
0.1
0
1 2 3 4 5 6 7 8 9
# of D2D pairs
Figure 8. Success probability with D2D pairs density.
In Fig.-9, transmission capacity increases with path loss coefficient, α due to increase 384
in the fading of interference. By looking at Fig.-9, we observe that for α = 4 almost a similar 385
transmission capacity is achieved with AIR scheme at 4.06e-05 of optimal small cell density 386
compare to 5.31e-05 for the same in the scheme presented in [12]. For α = 3, an equivalent 387
transmission capacity is achieved for optimal small cell density of 3.7e-05 and 4.23e-5 in 388
AIR and the [12] scheme, respectively. 389

10-5
3
2.5
Transmission Capacity
2
1.5
0.5
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Small Cell Density 10-4
Figure 9. Transmission capacities for optimal deployment of various schemes.
5. Conclusions 390
In this paper, we developed mathematical models for network, SNIR, and small cell 391
density for system performance modelling and analysis. In addition, we derive the outage 392
probabilities of D2D, macrocell, and small cell links. Finally, we proposed a dynamic 393
algorithm called acceptance interference region (AIR) to provide a solution to the problem 394
of achieving a strict QoS guarantee to all links in D2D-enabled HetCNets. Our analytical 395
models are validated using a MATLAB-based simulation. The simulation results show that 396
the proposed AIR scheme achieved an improved outage probability of 35% and 49% in 397
eNB and SCeNB links, respectively when compared with the traditional neighbor-based 398
methods. We also proposed an efficient ON/OFF algorithm to achieve a better transmission 399
capacity in the network than the existing methods. We found that the transmission capacity 400
is maximized at lower small cell densities. Developing a test-bed measurement system for 401
further validating the system performance is suggested as future research work. 402
Author Contributions: Conceptualization, M.K.; investigation, M.K. and N.I.S.; methodology, M.K. 403
and N.I.S.; project administration, N.I.S.; resources, N.I.S. and J.G.; supervision, N.I.S. and J.G.; 404
validation, M.K. and N.I.S.; writing—original draft, M.K.; writing—review and editing, N.I.S. and 405
J.G. Corresponding author, N.I.S.; Responded to reviews, M.K., N.I.S., and J.G. All authors have read 406
and agreed to the published version of the manuscript. 407
Funding: Auckland University of Technology, Auckland, New Zealand. 408
Institutional Review Board Statement: Not Applicable 409
Informed Consent Statement: Not Applicable 410
Data Availability Statement: Not Applicable 411
Conflicts of Interest: Not Applicable 412
Appendix A. Proof of AIR algorithm 413
Let us consider a the maximum allowable outage probabilities for any link is τ. So to 414
guarantee the QoS requirements all communications links must satisfy the following as: 415
Pout ≤ τ
! ! !
Pj di,r
⇒ 1 − exp exp ≤τ
α Pj m

! ! !
Pj di,r
⇒ exp exp ≥ 1−τ
α Pj m

! !
− N0 γth dαt,r )
⇒ exp α
Pj di,r exp ≥ (1 − τ )( Pi γth dαt,r + Pj di,r
α
)
Pj Pj m

" ! ! #
− N0 γth dαt,r )
⇒ α
Pj di,r exp exp − 1 + τ ≥ (1 − τ ) Pi γth dαt,r
Pj Pj m
 1
α
(1 − τ ) Pi γth 1
⇒ di,r ≥ dt,r  ∗ 

− N γ dα
exp − β D ( Pjm′ λd + Pk m λs ) − 1 + τ Pj
0 th t,r
exp Pj
 1
α
min (1 − τ ) Pi γth 1
⇒ di,r = dt,r  ∗ 

− N γ dα
exp
0 th t,r m m
exp − β D ( Pj′ λd + Pk λs ) − 1 + τ Pj
Pj
−N α α
Pj di,r
0 γth dt,r )

Where β D = κγth m d2t,r /Pj m , δD = exp Pj Pi γth dαt,r + Pj di,r
α , and m = 2/α. 416
417
Appendix B. Proof of small cell density and D2D pairs affecting communication link 418
quality 419
The transmission capacity of small cell can be expressed as:

S
Tc = λs (1 − Pout )
= λs δS exp(− β S ( Pj m λd + Pk′ m λs ))
As transmission capacity depends on small cell density, λs , so we can have 420
dTc
= δS exp(− β S ( Pj m λd + Pk′ m λs )) − λs δS exp(− β S ( Pj m λd + Pk′ m λs )
dλs
d
X {exp(− β S ( Pj m λd + Pk′ m λs ))}
dλs
= δS (1 − β S Pkm′ λs ) exp(− β S ( Pj m λd + Pk′ m λs ))
d2 T c
As λs ̸= 0 and here < 0, so for optimal values we have
dλ2s
dTc
=0
dλs
=> δS (1 − β S Pk ′m λs ) exp(− β S ( Pj m λd + Pk′ m λs )) = 0
=> (1 − β S Pkm′ λs ) = 0, as exp(− β S ( Pj m λd + Pk′ m λs )) ̸= 0
1 Pkm
=> λs = = S d2
Pk′m β S Pkm′ κγth k,s
Appendix C. Proof of small cell density of communications mode 421
To guarantee link QoS for small cell links we have, 422

Pout,k ≤ τ
=> 1 − δS exp(− β S ( Pj m λd + Pk′ m λs )) ≤ τ
1−τ
=> exp(− β S ( Pj m λd + Pk′ m λs )) ≥
δS
1 δ
=> ( Pj m λd + Pk′ m λs ) ≤ ln ( S )
βS 1−τ
1 δ
=> λs ≤ ln ( S ) − ρm jk′ λd
β S Pk′ 1−τ
Where ρm m m
jk′ = Pj /Pk ′ is the ratio of transmission power of the participating UEs. 423
References 424
1. Kamruzzaman, M.; Sarkar, N.I.; Gutierrez, J.; Ray, S.K. A mode selection algorithm for mitigating interference in D2D enabled 425
next-generation heterogeneous cellular networks. 2019 International Conference on Information Networking (ICOIN). IEEE, 426
2019, pp. 131–135. 427
2. Forecast, G. Cisco visual networking index: global mobile data traffic forecast update, 2017–2022 ; 2019. 428
3. Kuruvatti, N.P.; Hernandez, R.; Schotten, H.D. Interference Aware Power Management in D2D Underlay Cellular Networks. 429
2019 IEEE AFRICON. IEEE, 2019, pp. 1–5. 430
4. Kazmi, S.A.; Tran, N.H.; Saad, W.; Han, Z.; Ho, T.M.; Oo, T.Z.; Hong, C.S. Mode selection and resource allocation in device-to- 431
device communications: A matching game approach. IEEE Transactions on Mobile Computing 2017, 16, 3126–3141. 432
5. Araniti, G.; Raschellà, A.; Orsino, A.; Militano, L.; Condoluci, M. Device-to-device communications over 5G systems: Standard- 433
ization, challenges and open issues. In 5G mobile communications; 2017; pp. 337–360. 434
6. Lin, X.; Andrews, J.G.; Ghosh, A.; Ratasuk, R. An overview of 3GPP device-to-device proximity services. IEEE Communications 435
Magazine 2014, 52, 40–48. 436
7. Attaul Mustafa, H.; Imran, M.A.; Zeeshan Shakir, M.; Imran, A.; Tafazolli, R. Separation Framework: An Enabler for Cooperative 437
and D2D Communication for Future 5G Networks. arXiv e-prints 2016, pp. arXiv–1604. 438
8. Cai, X.; Zheng, J.; Zhang, Y. A graph-coloring based resource allocation algorithm for D2D communication in cellular networks. 439
2015 IEEE International Conference on Communications (ICC). IEEE, 2015, pp. 5429–5434. 440
9. Xu, S.; Kwak, K.S.; Rao, R. Interference-aware resource sharing in D2D underlaying LTE-A networks. Transactions on Emerging 441
Telecommunications Technologies 2015, 26, 1306–1322. 442
10. Bithas, P.S.; Maliatsos, K.; Foukalas, F. An SINR-aware joint mode selection, scheduling, and resource allocation scheme for D2D 443
communications. IEEE Transactions on Vehicular Technology 2019, 68, 4949–4963. 444
11. Zhi, Y.; Tian, J.; Deng, X.; Qiao, J.; Lu, D. Deep reinforcement learning-based resource allocation for D2D communications in 445
heterogeneous cellular networks. Digital Communications and Networks 2021. 446
12. Xu, Y.; Liu, F.; Wu, P. Interference management for D2D communications in heterogeneous cellular networks. Pervasive and Mobile 447
Computing 2018, 51, 138–149. 448
13. Sun, J.; Zhang, Z.; Xiao, H.; Xing, C. Uplink interference coordination management with power control for D2D underlaying 449
cellular networks: Modeling, algorithms, and analysis. IEEE Transactions on Vehicular Technology 2018, 67, 8582–8594. 450
14. Alzoubi, K.H.; Roslee, M.B.; Elgamati, M.A.A. Interference Management of D2D Communication in 5G Cellular Network. 2019 451
Symposium on Future Telecommunication Technologies (SOFTT). IEEE, 2019, Vol. 1, pp. 1–7. 452
15. Sarma, S.S.; Hazra, R. Interference management for D2D communication in mmWave 5G network: An Alternate Offer Bargaining 453
Game theory approach. 2020 7th International Conference on Signal Processing and Integrated Networks (SPIN). IEEE, 2020, pp. 454
202–207. 455
16. Yang, J.; Ding, M.; Mao, G.; Lin, Z. Interference Management in In-Band D2D Underlaid Cellular Networks. IEEE Transactions on 456
Cognitive Communications and Networking 2019, 5, 873–885. 457
17. Shamaei, S.; Bayat, S.; Hemmatyar, A.M.A. Interference management in D2D-enabled heterogeneous cellular networks using 458
matching theory. IEEE Transactions on mobile computing 2018, 18, 2091–2102. 459
18. Liu, X.; Xiao, H.; Chronopoulos, A.T. Joint Mode Selection and Power Control for Interference Management in D2D-Enabled 460
Heterogeneous Cellular Networks. IEEE Transactions on Vehicular Technology 2020, 69, 9707–9719. 461
19. Albasry, H.; Zhu, H.; Wang, J. In-Band Emission Interference in D2D-Enabled Cellular Networks: Modeling, Analysis, and 462
Mitigation. IEEE Transactions on Wireless Communications 2018, 17, 7395–7410. 463
20. Hassan, Y.; Hussain, F.; Hossen, S.; Choudhury, S.; Alam, M.M. Interference minimization in D2D communication underlaying 464
cellular networks. IEEE Access 2017, 5, 22471–22484. 465
21. Huynh, T.; Onuma, T.; Kuroda, K.; Hasegawa, M.; Hwang, W.J. Joint downlink and uplink interference management for device to 466
device communication underlaying cellular networks. IEEE Access 2016, 4, 4420–4430. 467
22. Chen, W.; Li, T.; Xiao, Z.; Wang, D. On mitigating interference under device-to-device communication in macro-small cell 468
networks. 2016 International Conference on Computer, Information and Telecommunication Systems (CITS). IEEE, 2016, pp. 1–5. 469
23. Celik, A.; Radaydeh, R.M.; Al-Qahtani, F.S.; Alouini, M.S. Resource allocation and interference management for D2D-enabled 470
DL/UL decoupled Het-Nets. IEEE Access 2017, 5, 22735–22749. 471
24. Jiang, F.; Wang, B.; Sun, C.; Liu, Y.; Wang, R. Mode selection and resource allocation for device-to-device communications in 5G 472
cellular networks. China Communications 2016, 13, 32–47. 473
25. Librino, F.; Quer, G. Distributed mode and power selection for non-orthogonal D2D communications: a stochastic approach. 474
IEEE Transactions on Cognitive Communications and Networking 2018, 4, 232–243. 475
26. Chen, H.; Deng, X.; Gao, M.; Yang, L.; Guo, L.; Chi, M. Location related communication mode selection and spectrum sharing for 476
D2D communications in cellular networks. 2018 International Conference on Intelligent Transportation, Big Data & Smart City 477
(ICITBS). IEEE, 2018, pp. 169–173. 478
27. Haneggi, M. Stochastic geometry for wireless networks Cambridge University Press 2012 , 2012. 479
28. Xu, Y. On the performance of device-to-device communications with delay constraint. IEEE Transactions on Vehicular Technology 480
2016, 65, 9330–9344. 481
29. Gupta, P.; Kumar, P.R. The capacity of wireless networks. IEEE Transactions on information theory 2000, 46, 388–404. 482

A Dynamic Algorithm For Interference Management in D2D Enabled Heterogeneous Cellular Networks Modelling and Analysis Rev2

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A Dynamic Algorithm For Interference Management in D2D Enabled Heterogeneous Cellular Networks Modelling and Analysis Rev2

Uploaded by

Copyright:

Available Formats

Article

A Dynamic Algorithm for Interference Management in

further towards developing the next-generation cellular networks. 16

Keywords: Device-to-Device (D2D) communication; heterogeneous networks; HetCNets; interference 17

under the terms and conditions

Version January 28, 2022 submitted to Journal Not Specified https://www.mdpi.com/journal/notspecified

volumes is a major concern for the cellular network operators [1]. 40

[1]. During natural disasters such as earthquakes or hurricanes, a replacement of the 44

communication in cellular networks will bring significant performance gains in terms of 48

and content sharing/dissemination [4], [5], [6]. 51

etc. Among them, interference management (IM) in a heterogeneous scenario comparing 56

all tiers simultaneously is very important and complex [1]. 57

Figure 1. Interference scenarios in D2D-enabled 3-tiers cellular network.

of D2D communication in cellular networks, it is essential to manage interference, and 61

by selecting an appropriate mode of transmission, power control, resource allocation, 62

interference scenarios [3]. 68

In this paper, we identify various key factors that contribute to interference in an 69

on the fact that a dynamic algorithm is required to handle interference management in 72

D2D-enabled heterogeneous cellular networks. 73

The main contributions of the paper are as follows. 74

• We propose a dynamic algorithm called Acceptance Interference Region (AIR) to 75

D2D-enabled heterogeneous cellular networks. A distance-based approach is used to 77

achieve guaranteed links quality. The proof of AIR is provided in Appendix A. 78

• We propose an efficient ON/OFF algorithm to provide a solution to the problem of 79

achieving a maximum transmission capacity in the network. 80

quality. We validate our analytical models using a MATLAB-based simulation. 86

interference management in D2D-enabled heterogeneous cellular networks. Section 3 88

presents the system model, including network, Signal-to-interference-plus-noise ratio 89

coverage probability and spectral efficiency. The system performance is evaluated in 92

Table 1. List of key notations and abbreviations used

In a D2D-enabled heterogeneous cellular network, to manage interference the state- 97

as interference avoidance, interference cancellation and interference reduction. 100

Whereas in interference cancellation techniques- multiple antenna systems, beam- 126

overheads. In [15] an overlay mode of D2D communication in mm-wave 5G networks 130

throughput without compromising SINR. 132

each UE to control D2D-to-cellular interference. To minimize the impact of interference in 140

approach for their modelling in 2-tier networks. 154

propose an interference management algorithm to maximize the performance of D2D 160

and their analysis is limited to a 2 tier network. 164

Most of the aforementioned state-of-the-art techniques focus on interference manage- 186

interference in a 3-tier heterogeneous cellular network. 200

3. System Model 201

3.1. Network Model 202

For modelling a 3-tier D2D-enabled heterogeneous cellular network, we consider a 203

are denoted by j ∈ {1, 2, . . . , ND }, k ∈ {1, 2, . . . , NS } and i ∈ {1, 2, . . . , NC } respectively. 210

topology on network performance stochastic geometry is more suitable[27]. In a 3-tier 212

to be exponentially distributed. In such a D2D-enabled HetCNets, the channel model is 224

signal can be expressed as[14]: 226

Pout = Pr {γy ≤ γth } (2)

3.2. SINR Model 240

given by: 244

Since channel coefficient is exponentially distributed, the expectation of interference 247

from the above equation can be expressed as follows 248

γth SNIR threshold. 250

transformation of random variables Id and Is evaluated at z, respectively. Interference due 253

−κPj′ m γth m d2t,r λd

−κγth m d2t,r ( Pj′ m λd + Pk m λs )

= 1 − δD exp(− β D ( Pj′ m λd + Pk m λs )) (8)

receiver is increased. 264

eNB, the SINR at the receiver i can be expressed as: 266

= 1 − δS exp(− β S ( Pj m λd + Pk′ m λs )) (12)

of SceNB links. 286

dijD ≥ dijD min