Author’s Accepted Manuscript

A Survey on Interference Management for Device-

to-Device (D2D) Communication and its
Challenges in 5G Networks

Mahda Noura, Rosdiadee Nordin

www.elsevier.com/locate/jnca

PII: S1084-8045(16)30075-3
DOI: http://dx.doi.org/10.1016/j.jnca.2016.04.021
Reference: YJNCA1638
To appear in: Journal of Network and Computer Applications
Received date: 24 November 2015
Revised date: 26 April 2016
Accepted date: 27 April 2016
Cite this article as: Mahda Noura and Rosdiadee Nordin, A Survey on
Interference Management for Device-to-Device (D2D) Communication and its
Challenges in 5G Networks, Journal of Network and Computer Applications,
http://dx.doi.org/10.1016/j.jnca.2016.04.021
This is a PDF file of an unedited manuscript that has been accepted for
publication. As a service to our customers we are providing this early version of
the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting galley proof before it is published in its final citable form.
Please note that during the production process errors may be discovered which
could affect the content, and all legal disclaimers that apply to the journal pertain.
 1

A Survey on Interference Management for Device-to-

Device (D2D) Communication and its Challenges in 5G
Networks
Mahda Noura*, Rosdiadee Nordin
Department of Electrical, Electronic and Systems and Engineering, Faculty of Engineering and Built Environment Building,
Universiti Kebangsaan Malaysia, 43600 UKM, Bangi, Slenagor, Malaysia

Abstract

Device-to-Device (D2D) communication is a promising concept to enhance the performance of devices by allowing direct
transmission between closely located user pairs. The initial studies have proven that, direct communication will improve spectrum
reuse, throughput, energy consumption, coverage, and reduce end to end latency. Additionally, it will enable the creation of new peer-
to-peer services and location-based applications. Therefore, current research trends have revealed that D2D will be one of the
technologies in next generation cellular network, i.e. 5G. However, introducing D2D to cellular network imposes various technical
challenges. Interference management between cellular users and D2D users is considered to be one of the most critical issues when
D2D is introduced to cellular network because D2D users share the same licensed spectrum with cellular users. In this paper, we
provide a comprehensive survey of the various state-of-the-art approaches for interference management in D2D communication
enabled in cellular networks. Furthermore, we classify these interference management techniques based on their underlying
approaches. Qualitative comparison between the various interference management techniques found that the existing approaches do
not satisfy 5G requirements. To this end, the open challenge in introducing D2D to 5G cellular networks is provided at the end of this
paper.

Keywords: Device-to-device communication, D2D, interference management, cellular network, 5G, survey

1. INTRODUCTION

Continuously evolving mobile network capabilities from 2G to 3G and 4G has fundamentally changed the world and how
mobile devices are used. 2G introduced a harmonized digital standard for voice and enabled roaming, and SMS messaging
gained popularity later. Evolution to 3G delivered the first experience of mobile broadband and improvements in stages. 4G
ushered in the era of superfast mobile broadband driving massive adoption by Smartphone users. Users worldwide have
embraced social networking and are increasingly mobile. In 2020 around 90% of the world’s population will be covered by
mobile broadband networks. Mobile data traffic in Q1 2015 was 55% higher than in Q1 2014. By 2020, 80% of mobile data
traffic will be from Smartphones with consumption of video-based content the main driver [1]. 5G will become the dominant
mobile communications technology during 2020 in subscription numbers, amassing 3.6 billion users at that time [2].

D2D communication represents a new type of wireless communication paradigm technology which allows direct
communication between nearby wireless devices while remaining controlled under macro base stations [3] [4]. With D2D
communication, the data between a UE pair does not need to traverse through the core network such as access points (APs) or
base stations (BS) as long as they are in proximity. Figure 1 illustrates D2D communication in future dense small cell networks
with macro-cells, micro-cells, pico-cells and femto-cell layers. In particular, D2D communication has recently attracted interest
from academia and industry due to the proximity, reuse and hop gains [5].
Although D2D communication provides many advantages to LTE/LTE-A systems, several challenges arise in terms of
interference mitigation, device discovery and synchronization, mode selection, security, and QoS, all of which will be detailed in
Section II. To realize the potential of D2D communication in cellular networks, intensive research has been carried out by both
academia and industry to address these issues.
To the best of our knowledge, quite a few survey papers related to D2D communication in the cellular network has been
published recently. In [6], the authors categorize D2D communication based on spectrum reuse and provide the- state-of-art
based on this classification in terms of the performance metrics studied. Furthermore, it discusses the existing D2D protocols for
D2D communication. The paper is concluded with the advantages and disadvantages of each of the spectrum sharing schemes,
common assumptions and the maturity of D2D communication in the real world. However, this work does not consider
interference management in its review. In [7], the research issues for D2D including interference management and radio resource
management are briefly discussed. At the end of the paper, it presents the performance metrics and available prototypes and
experiments developed for D2D studies. In [8], the authors performed a classification for D2D communication based on the level
of involvement of the cellular operator and then introduced a two-tier 5G cellular network model for D2D communication. Then
research challenges such as security, interference management, resource allocation and pricing are discussed briefly. None of the
 2

surveys presented in [6]–[8] provide an in-depth discussion or an analysis in terms of interference

Coverage area of Macrocell

Coverage area of Microcell

Coverage area of Picocell

Traditional
Coverage area of Femtocell Traditional
Link
Link
Cellular Mode
D2D Model

D2D Link
D2D Link

Figure 1: D2D communications in multi-tier cells in HetNets

mitigation. To be precise, only a small section of their work briefly discusses the interference management issue. However, this
paper is totally dedicated to interference management providing a comprehensive survey. In addition to surveying the current
issues and approaches, the authors of [9] provide a classification on D2D communication from different perspectives. Then, the-
state-of-art in D2D communication including: mode selection, interference management, power consumption and advanced
topologies for OFDMA cellular networks are discussed. The authors classify the current approaches in interference management
based on interference scenarios and provide a qualitative comparison among the surveyed approaches. However, the scope of
this survey is limited to general OFDMA-based cellular networks.
In this paper, we aim to fill the research gaps found in the previous surveys by focusing on interference management
techniques proposed in recent years for LTE/LTE-A HetNets. In particular, our paper includes a comprehensive classification
specifically for interference management for D2D communication in terms of cross-tier and co-tier interference which has not
been presented in previous works. Moreover, this survey classifies the different interference management approaches proposed
in the literature based on the underlying technique and then provides an in-depth explanation of the related work for each
technique, and the current challenges that exist to incorporate D2D communication to the future 5G cellular network. In our
survey, the more primitive existing interference management schemes are first reviewed, followed by the more intricate
approaches, to reflect the knowledge advancement in the field to date. Furthermore, a comprehensive qualitative comparison
between the D2D interference management techniques is provided. This comparison enables us to identify the strength and
weaknesses of existing interference management techniques. This comparison ultimately leads to the discussion of the technical
challenges and potential research directions for future 5G cellular networks. Finally, we elaborate on the limitations that exist in
the current literature to include D2D in future 5G cellular networks. We believe that this survey will aid researchers to gain a
deep understanding on interference management in D2D communications and its addition to 5G networks.
The rest of the article is organized as follows. Section 2 briefly discusses the technical challenges faced by D2D
communication when introduced to the cellular network. Section 3 elaborates on the types of interference appearing in D2D
communication and the main interference mitigation schemes are provided in Section 4. Finally Section 5 explains the main
challenges which exist to incorporate D2D in 5G networks with regards to the interference management and Section 6 draws our
conclusion.

2. TECHNICAL CHALLENGES OF D2D COMMUNICATION IN CELLULAR NETWORKS

Although D2D communication is being studied in 3GPP [10], D2D is still immature and faces many technical challenges and
issues regarding aspects such as device discovery, mode selection, security and interference mitigation. The main focus of this
paper is interference management, but other challenges faced by D2D communication are also discussed briefly.
 3

2.1. Device discovery

Before two devices can directly communicate with one another, they must first know that they are nearby each other. During
the device discovery phase, the devices search for the presence of their peer in range for D2D communication [5].
The discovery is made possible by sending a discovery signal to identify the presence of possible devices in proximity, and
then the identity of devices can be exchanged between the new pair. When two UEs find each other in the device discovery
phase they are considered as D2D candidates. Finally, a series of messages about link quality is passed between devices and the
BS. This information serve as the basic input to the mode selection and D2D candidates cannot communicate directly until mode
selection criterion is satisfied.

2.2. Mode Selection

Mode selection is performed after the pair of D2D candidates find each other for potential future communication. Although
the D2D candidates are within the direct communication range with each other, it may not be optimal for them to work in the
D2D mode from the performance perspective. Mode selection means that the network and/or the D2D candidates decide whether
the D2D candidates should communicate directly or via the network as conventional cellular network. The communication mode
is categorized into the following modes:

 Dedicated Mode/Overlay Mode: In this mode, the cellular network has abundant channel resources so that the DUEs can use
dedicated resources that are orthogonal to CUEs.
 Reuse Mode/Underlay Mode: In this mode, the D2D communication will share the same resources with existing CUEs and
hence may cause interference to CUEs.
 Cellular Mode: The two UEs will communicate as traditional CUEs, that is, communicate with each other through the BS.

The reuse/underlay mode achieves higher spectrum efficiency in compared to the other communication modes, but D2D
communication in this resource sharing mode may interfere with cellular UEs and other D2D UEs using the cellular radio
resources. On the other hand, the dedicated/overlay mode can completely avoid interference since some resources are dedicated
for D2D communications. However, the spectrum utilization can be very poor in this resource sharing mode. D2D
communication modes have a direct impact on the interference in the network and therefore, should be carefully selected after
much analysis.

2.3. Security

Providing efficient security is a major issue in D2D communication. The D2D communication network is prone to many
security risks because of the routing of user data through other users’ devices. This data can be hacked, which would breach
privacy and confidentiality. Since D2D communication could be vulnerable to malicious attacks (e.g., masquerading,
eavesdropping, man-in-the-middle attack etc.), enhanced authentication and key agreement mechanisms are required to secure
D2D communication in cellular networks. The security of devices can be ensured if closed access is applied to devices. In close
access, a device has a list of certain reliable devices, like the users in the close vicinity or office to whom you are familiar with,
otherwise the users that have been legitimated through a reliable party like an association, can unswervingly communicate with
each other, sustaining a level of discretion, whereas the devices not on this list need to use the macro cell level to communicate
with it. Instead of this, in open access, each device can turn in to relay for other devices deprived of any limits. Meanwhile, in
such an instance security is an open research problem. Interference exploitation can be used as an aid to provide secret
communication in D2D communication, which is explained in [11], [12].

2.4. Interference Management

Interference management is one of the most important challenges for D2D communication enabled in cellular networks. As
described before, the operators prefer the sharing mode to increase the spectral efficiency, but this causes the problem of
interference. As many D2D users and cellular users are using the same portion of spectrum, these can cause interruptions to each
other. Interference management in D2D communication and the different techniques used to mitigate interference are explained
in detail in the next section.

3. INTERFERENCE MANAGEMENT IN D2D

 4

Due to the introduction of D2D communication to the cellular network, the cellular architecture changes and now includes two
tier [7], [8]. The first tier is the conventional macrocell layer, which involves the communication between BS and device. The
new tier, called the device tier involves D2D communication. Thus, such system is called two-tier or cellular system architecture.
The device tier is an unplanned and random distribution of D2D user equipment (DUE). The new architecture has significant
improvement in terms of throughput, coverage, end to end latency if designed carefully [8]. However, it introduces several
technical challenges and issues for both DUE and cellular user equipment (CUE). Among these challenges interference
management between CUE and DUEs becomes one of the most critical issues for D2D communication in sharing mode; where
the same radio resources are used for both cellular and D2D communication. It is preferred to deploy D2D communication in
sharing mode to enhance the spectral efficiency. However this in return, gives rise to severe interference management challenges
since relative to cellular communication scenarios, the system requires managing new interference situations. If the generated
interference is not well controlled it would deteriorate the potential benefits of D2D communication since the overall cellular
capacity and efficiency is degraded.
This section of the study explains interference management in the two-tier network architecture in terms of interference type,
interference control level, and interference management technique.

3.1. Types of Interference in Two-Tiered Network Architecture

In two-tiered network architecture two types of interference: co-tier and cross-tier are introduced as shown in Fig. 2.
All the possible interference scenarios in Fig. 2 are explained in this section and are illustrated in Fig. 3 as below:

3.1.1. Co-tier interference

This type of interference is produced between network elements which belong to the same tier in the network. In the case of D2D
enabled cellular network, co-tier interference occurs between a D2D users and another D2D user in the same tier. The D2D users
causing interference to each other are immediate neighboring D2D users, since they are located near each other. To set up a
direct link between D2D users, the signal to interference plus noise ratio (SINR) value must be higher than a predefined
threshold parameter. Otherwise, if the DUE SINR falls below the defined threshold parameter due to co-tier interference a
communication link cannot be established.
In OFDMA systems, the co-tier interference is caused when the same set of resource blocks are allocated to multiple DUEs. In
this case the interference is always generated from the D2D transmitter to D2D receiver in a D2D pairs which are assigned the
same cellular resources regardless of the resource reuse direction (UL/DL). Furthermore, the co-tier interference incurred at a
D2D receiver from neighboring D2D transmitter can be mitigated through proper user pairing and frequency assignment
techniques.

3.1.2. Cross-tier interference

This type of interference is produced between network elements which belong to different tiers, i.e., interference between DUE
and CUE. The cross-tier interference can be between (i) a CUE a DUE, and between (ii) a CUE and multiple DUEs. This
interference scenario occurs when the resource blocks allocated to a cellular user are reused by one (or more) D2D users. In this
type of interference, the aggressor (or the source of interference) and the victim of interference are different depending on the
resource reuse direction (UL/DL).

Case 1: Interference from D2D to cellular network: When D2D links use the same frequency resources as CUEs in the uplink
direction, D2D transmitter is the aggressor interfering with eNB and also the cellular uplink user is the aggressor interfering with
D2D receiver.

Case 2: Interference from cellular network to D2D user: On

 5

Figure 2: Types of interference in a two tier D2D enabled cellular network

the other hand, when downlink resources of licensed band are reused in for D2D links, eNB is the aggressor interfering with
D2D receivers and D2D transmitter is the aggressor interfering with cellular downlink user.

3.2. Interference Control Level

In general, interference management schemes can be classified into centralized, distributed and semi-distributed approaches
depending on the algorithm operation as shown in Fig. 4.

3.2.1. Centralized
In the centralized approach, the eNB fully manages interference between cellular and D2D users. This central entity collects
information such as the channel state information (CSI), channel quality, interference level for each user in the network, decides
on the channels to assign to each single user with the suitable format and power level. Based on the information obtained, the
central entity allocates the resources to each CUE or DUE. The main problem with centralized schemes is the huge amount of
signaling overhead required for exchanging CSI and feedback. Moreover, the interference management complexity increases
exponentially with the number of users in the network, since the operation is performed by a single entity, which has to process
large amounts of data. Thus, centralized schemes are only possible for small-sized D2D networks.
3.2.2. Distributed
In a distributed scheme, the operation of interference management does not require a central entity and is performed
autonomously by DUEs themselves. The distributed scheme reduces the control and computational overhead, due to limited CSI
and feedback. However, interference is difficult to coordinate; this approach is more appropriate for large size D2D networks.
3.2.3. Semi-distributed
Although both centralized and distributed schemes have their advantages and disadvantages,
tradeoffs can be achieved between them. Such interference management schemes are said to be “semi-distributed” or “hybrid”.
In the semi-distributed interference management
 6

CUE1 CUE1

Index 2

D2D Rx
f1 Index 4

D2D Rx
f1

f1
Index 1

Index 3
f1
BS1

D2D Tx
D2D Rx Index 5 BS1
Index 5

D2D Link D2D Tx

Cellular Link D2D Rx

D2D Tx
Co-tier interference D2D Tx

Cross-tier interference

For example, D2D UEs and CUEs act as a source of uplink cross-tier interference to the serving base station and the nearby D2D
UEs, respectively. On the other hand, the serving base station and D2D UEs cause downlink cross-tier interference to the D2D
UEs and nearby CUEs, respectively.

Figure 3: Interference scenario for different reuse resource

schemes, different levels of involvement can be defined. Such schemes could be suitable for moderately large networks.

4. COMPUTATIONAL MATHEMATICS

For the analysis and design of interference management schemes in D2D enabled cellular network extensive research has been
performed using mathematical theories including stochastic geometry, game theory, graph theory, evolution theory, queuing
theory and optimization theory. Table 1 summarizes the mathematical theories used in D2D interference management literature.
This Section of the study aims to describe each of these tools briefly and in the next Section their related literature is studied in
detailed.

Table 1: Computational Mathematics used in the literature

Mathematical Tool Related Literature
Stochastic Geometry [13], [14], [15], [16], [17], [18]
Game Theory [19], [20], [21], [22], [23]
Graph Theory [24], [25], [26], [27]
Evolution Theory [28], [29], [30]
Queuing Theory [31]

4.1. Stochastic Geometry

Stochastic geometry is a rich branch of applied probability which allows the study of random phenomena on the plane or in
higher dimensions. It is intrinsically related to the theory of point processes. It has been used as a tool for characterizing
interference in wireless networks for a long period of time [32]. Powerful tools from stochastic geometry have been successfully
applied to the spatial modeling and performance analysis of both wireless and ad hoc and cellular networks [33]. More recently,
 7

these tools have been used to characterize various aspects of D2D networks, such as mode selection in D2D communication
underlaying cellular networks [34], [35], D2D interference management [36], [37], multicast transmission [38], and distributed
caching in D2D networks [39].

Control Level

Centralized Distributed Semi-

Central Distributed
DUEs perform
controller interference Different levels
performs management of network
interference involvement
management can be defined

Figure 4: Interference control level in D2D communication

4.2. Game Theory

Game theory offers a set of mathematical tools that can be used to model, design, and analyze a competitive situation between
intelligent rational decision makers. In recent years, game theory has been widely used for networking and wireless
communication to resolve competitive problems. A game is characterized by a set of rational players, and a utility function for
each player. There are different game theoretic models to solve the interference problem for D2D communication such as non-
cooperative static games, stakelberg game and combinatorial auction. Two broad categories of a game theoretic model for D2D
resource allocation are non-cooperative and cooperative. In the former type, DUEs are commonly viewed as players competing
for the resources.

4.3. Graph Theory

Graph theory is an efficient tool for modeling and analyzing various types of interactions, relations, and dynamics in different
networks. A graph can be used to represent the interference relationships among different D2D communication links and cellular
communication links, and the resource sharing problem can be solved using graph theory. Graph theory has been used for
interference management in D2D underlay in several recent works [24]–[26].

4.4. Evolution Theory

In [30], the theory of evolutionary game is applied for a cognitive cellular network with D2D communication. A distributed
resource allocation and mode selection scheme is proposed for secondary users to maximize the spectrum utility. The user can
select to operate in either cellular mode (BS mode) or D2D mode. The authors attempt to optimize the MS criterion with a set of
utilities considering the data rate, transmit power, and cross-tier interference for the above mentioned modes, and then use the
evolutionary game to obtain the MS.
The authors of [29] consider the MS and RA in D2D communications underlay cellular networks, where several pairs of D2D
links co-exist with several cellular users. They formulate the problem of maximizing the system throughput with minimum data
rate requirements, and use the particle swarm optimization method to obtain the solutions. Through simulation, the authors show
that the proposed scheme improves system performance compared to overlay D2D communication.

4.5. Queuing Theory

Queuing theory is a useful analytical tool to model a wide range of problems and scenarios in communication networks [40].
Traditionally, queuing theory based models have extensively been used in predicting the QoS of access networks. In previous
studies, queuing models were used to evaluate QoS parameters such as packet blocking probability, average packet delay and
throughput. The authors of [31] formulate a centralized delay-aware resource allocation and power control framework using
queuing model to optimize the three resource control action.
 8

5. THE-STATE-OF-ART IN INTERFERENCE MANAGEMENT IN D2D-ENABLED CELLULAR NETWORK

The classification of interference management in two-tiered network due to the addition of a new device tier has been
explained from different perspectives in Section 3. The research work in interference management in D2D enabled cellular
network is still ongoing and many schemes have been proposed. In this section the main interference management schemes
would be reviewed.
There are various schemes for the management of cross-tier and co-tier interference generated by D2D communications but their
working principle or technologies are very different. Thus, a different classification can be made for interference management
schemes based on the underlying working principles or the key enabling technologies illustrated in Fig.5.
Based on the classifications provided in this study, we carefully marshal the available interference management related works
in Table 2, where the research works are classified depending on their considered type of interference. In particular, the relevant
research works on interference management in D2D-enabled cellular networks are divided into rows based on the considered
interference scenario. The Radio Resource Management (RRM) field in Table 1 specifies which radio resource management
technique the mentioned approach considers whether it is power control, mode selection or resource allocation. Network
complexity field indicates the number of cells, i.e. single/multiple cell(s), the number of D2D pairs and CUE(s), i.e. one/multiple
D2D pair(s), one/multiple CUE(s). In addition, for each research article we provide a comparative analysis in terms of the main
target, solution, the resource type (i.e, UL/DL) reused for D2D communication, and the interference control level.
In the first sub-section an overview of the main mathematical tools which have been used in the literature are provided. The
second sub-section provides an overview of the main interference management techniques that are shown in Fig. 5 is introduced
and will discuss in detail the recent schemes under each of the approaches to deal with cross-tier and co-tier interference in D2D
enabled cellular networks. The D2D devices should be able to cope with most of the situations. Intelligent interference
coordination technique is thus of great importance and more research has been dedicated to this part. In this section, several D2D
interference coordination schemes from the literature will be reviewed.

INTERFERENCE MANAGEMENT TECHNIQUES

FOR D2D COMMUNICATION

Interference Coordination Spatial Diversity Techniques

MIMO Interference
Spectrum Radio Resource Joint MS, PC, Interference
Power Control Mode Selection Cancellation & Beamforming Coding CoMP
Splitting Allocation RRA Alignment
Regeneration

41 [14] [41]
Fractional Frequency [23] [18] [100] [107]-[109]
Reuse: [58],[59],[60] Pre-Coding: Superposition [105]
[16] [52] [30]-[31] [85],[87],[88] [96] [101]] Coding: [102]
Time-Frequency
[43] [59] hopping: [13],[61] [49] [99]
Relay: [89]-[91] Rate Splitting: [103]
[46] [60] [54]
Graph-Theory: [24]-[26],
[48] [62], [64]
[78]-[80]
Cognitive Radio Networks:
[15],[65],[66],[68]
Clustering: [69]
Game-Theory: [19]-[23], [26],
[30], [71]-[73], [76]-[77]

Figure 5: Classification of interference management techniques in D2D enabled cellular network based on the underlying working
principle or key enabling technologies
5.1. Spectrum Splitting

The easiest way to coordinate the cross-tier interference between the cellular and device tier is to use spectrum splitting, which
will simplify the interference between DUEs and CUEs. The authors of [41] use spectrum splitting, where it is suggested to
divide the spectrum band into two parts, which is shown in Figure 6. One part would be dedicated to CUEs and the other part
would be assigned to DUEs. This would only leave the co-tier interference unsolved. However, dedicated channels for D2D
communication will lead to inefficient use of the available channels depending on the number of D2D terminals and the
proportion of available spectrum for them.
It is worth noting that the spectrum splitting scheme only aims to address the cross-layer interference between D2D users and
cellular users. Therefore, an additional mechanism for co-tier interference mitigation is required in spectrum splitting strategy.
 9

D2D mode

Cellular mode

Cellular Spectrum
Figure 6: spectrum splitting
5.2. Power Control

Although higher transmit power of D2D users can provide wider coverage and better signal quality, it can, at the same time,
cause tremendous interference to the cellular network. The proper power control (PC) mechanism is one approach to deal with
cross-tier interference generated from DUEs to the cellular network for both the UL and the DL case, as well as the co-tier
interference between DUEs in a hybrid cellular network with D2D communication [16], [42]–[45]. It coordinates the
interference imposed by DUEs to the cellular network and the interference from a DUE to a neighboring DUE by controlling the
transmit power levels of DUEs to improve system capacity, coverage, and reduce power consumption. To meet these goals, PC
schemes aim at maximizing the transmit power and at the same time limiting the generated interference. The benefit of these
schemes lies in that the DUEs and CUEs can exploit the whole bandwidth with interference coordination. Note that, the BS could
also adjust the transmitting power of the cellular network to reduce the interference from the cellular network (eNB and CUEs)
to the D2D receiver; however since the PC of the cellular network causes performance degradation of the cellular networks, it
may not be the best solution to manage the cross-tier interference between the D2D pair and the cellular network.
Power settings can be performed either in a fixed or dynamic fashion. The static settings include fixed power scheme or fixed
signal-to-noise ratio (SNR) target scheme, while dynamic power settings are preferred and include open loop fraction power
control scheme (OLPS) or close loop power control scheme (CLPS) [45]. In the OLPS setting, the D2D user adjusts its
transmission power based on its measurement results or predetermined system parameters. In the CLPS setting, the D2D user
adjusts its transmission power based on the coordination with eNB. Based on the investigations performed in [45], it is
concluded that the close loop power control scheme can be a suitable power control scheme for DEUs.
The PC algorithms are designed for different objectives, such as maximizing the rate [44], minimizing total power with respect
to a target rate [46], or maximizing a utility function characterized by the trade-off between spectrum and energy efficiency [47].
Several power control algorithms have been proposed for D2D communication underlaid cellular networks. A simple
centralized power control mechanism was proposed in [48] in a single cell scenario with one D2D pair and one CUE to restrict
the intra-cell interference from DUEs to cellular network (both BS and CUE). This is accomplished by reducing the power level
of D2D transmitters from its maximum power such that the SINR of the cellular link does not degrade below a certain level. To
achieve this, first the SINR distribution of the cellular user is formulated and then power control is applied to the D2D
communication to limit too much interference to cellular communication. The performance has been performed based on
different values of D (distance of the D2D pair from the BS) and L (distance between the D2D pairs). They conclude that, there
exists a threshold value D according to the position of the D2D pair for deciding when UL or DL resources should be reused for
the D2D communication. If D is less than the threshold value, DL resources sharing are favored for the D2D communication. On
the other hand, if D is larger than the threshold value, UL resources are beneficial for the D2D communication. However, the
main limitation of such approach for power control is that it would lead to inefficient use of resources since it is designed for the
worst case scenario.
The authors of [44] consider two fractional power control scheme for the sharing mode. In the first scheme, the BS applies
greedy sum-rate maximization without considering priority between cellular and D2D communication. In the second scheme, a
rate constrained power control is considered, where the CUEs are given priority. Similar to [48], the performance is analyzed for
different distance values of D and L. It is shown that the greedy sum rate maximization scheme is significant for small values of
D (D2D connection located close to the BS). On the other hand, for the scheme prioritizing the cellular communication, the
performance is worse than the fixed transmit power, yet it still achieves a considerably large gain in the case of a single CUE
without proper PC. However, this approach assumes that all the CSI is available at the BS to control the transmit powers, which
is not realistic.
Another paper [43] considers a multi-cell scenario focusing on a centralized dynamic power control mechanism for a single
D2D link to improve the cellular system performance by mitigating the interference generated by D2D communication to CUEs.
The transmission power of the DUE is adjusted by eNB as follows. At first, the received signal strength from the CUEs is
measured in the UL by DUEs in a pair, which shares the same set of resources. Next, the channel gain between D2D pairs is
calculated. Then, the eNB determines the channel gain between itself and the D2D pair. This excludes both CUEs using the same
resource blocks with the DUEs and the eNB from the coverage areas of DUEs. Finally, the transmission powers of DUEs located
in a pair is adjusted based on an indication from the eNB. The main limitation of this work is that power control is performed in a
centralized manner by the BS.
 10

A more sophisticated PC method for D2D connection that can share the UL resources is presented in [49][3]. The scheme
limits the maximum D2D transmit power, by utilizing the cellular PC information of the devices in D2D communication. To
control the interference when the DUE reuses UL cellular resources, the serving BS can reduce the D2D transmitter power by a
back-off parameter. If a DUE is assigned dedicated resources back-off value is not needed. For large values of the back-off
parameter B, D2D transmissions cause very low interference to UL transmissions. However, a large B implies a reduced range
for the D2D link itself. To avoid this limitation, they incorporate a power boosting factor β to the transmit power of the UL
transmitter that ensures that the SINR of the cellular uplink meets the target SINR. The boosting factor is dependent on the
backoff value.
Different from the works proposed for PC up to now, the authors of [46] consider PC in a scenario where multiple DUEs share
the same resources of one CUE to control interference. This interference management scheme is investigated under the
availability of either instantaneous CSI or average CSI using the heuristic algorithm. Their results demonstrate that multiple
DUEs can share the spectrum with one CUE without degrading its performance.
Furthermore, a distributed on-off PC algorithm for D2D underlaid cellular networks to mitigate cross-tier and co-tier
interference is proposed in [16] by limiting the DUE and CUE power levels in a random network scenario. In the presented on-
off scheme each D2D transmitter selects its transmitting power towards its intended receiver without the need of a central
control. The D2D transmit power value is selected from a range { }depending on (i) channel gain and (ii) a nonnegative
fixed threshold known by all users. The transmission power of a D2D pair is chosen to be if the channel quality is good.
Selecting a large threshold value reduces the interference from DUE, while larger threshold value leads to a smaller number
DUE to be scheduled. The coverage probability of CUE has been derived using stochastic geometry. The transmit power of the
CUE is determined according to the location of the user. When the user is located around the cell center, the CUE uses the
constant transmit power. Alternatively, when the user is located at mid range of the cell edge, the on-off power control strategy is
activated where the CUE increases it’s transmit power. Further, the uplink user is required to increase the transmit power linearly
according to the random number of DUE in the coverage of the BS. If the user is located at around the cell edge, the CUE sends
a signal with its maximum transmit power, thus the coverage probability decrease linearly. For performance analysis, the
coverage probability has been investigated under sparse DUE and dense DUE. In the sparse scenario, the proposed algorithm
improves the CUE coverage probability. In the dense scenario the DUE coverage probability decreases, yet still the total number
of successful D2D transmission is large than that of the sparse DUE deployment scenario.
To ensure the reliability of cellular users, the authors of [14] propose a scheme that does not cause an outage for CUEs. They
state that assuming DUEs have knowledge of the location and channel state of CUE is not feasible in a real system. Therefore,
they design a distributed power control scheme without any coordination from a central controller that uses statistical estimates
of the channel gains between itself and the eNB to leverage a specified interference margin of CUE. Initially, DUES set their
transmit power level in such a way that their transmission does not exceed the interference bound of CUE. D2D power
adjustment can be done if the interference margin and estimating the channel gain between DUE and the BS are known. Then,
dynamic source routing (DSR) protocol is used between the D2D source and destination to discover either a single-hop or multi-
hop route between them.
PC on its own is not an effective scheme to handle the interference imposed to DUEs and it should be considered jointly with
mode selection, resource scheduling and link adaption [45], which will be explained in the next sections.

5.3. Mode Selection

It is possible to avoid the effect of cross-tier interference between the cellular and D2D user or co-tier interference among DUE
with a proper mode selection (MS) algorithm. Although D2D candidates may be in range for direct communication with each
other, it may not be optimal for them to work in D2D mode because of the interference imposed on DUE or CUE. In this sense,
D2D MS algorithm decides on the optimal communication mode so that the overall network throughput is maximized and the
QoS requirements of the communication links are satisfied. As discussed in Section 2 of this paper each of the communication
modes affects the amount of interference between cellular users and D2D users or between multiple DUE.
There are many works, which focus on MS in the literature with different MS criteria. Simple MS can be performed based on
the path loss [45], received signal strength over the D2D link [50] or the distance between the terminals [34], [35], [51].
However, these schemes do not reflect exact channel quality or interference issues. In [44], [49] MS has been performed based
on the channel quality. In particular [49] considers sum rate of the connection between D2D pair and of the cellular connection
between BS and CUE as the MS criterion. Four communication modes are defined as: (i) DL resource sharing, (ii) UL resource
sharing, (iii) separate resource sharing and (iv) cellular mode sharing. The mode with the maximum sum rate is selected. The
authors evaluate the proposed scheme with different distance values of D and L. It is concluded that when D is large, it is more
beneficial to use sharing mode as the percentage area where the cellular user experiences strong interference reduces, and
alternatively when D is small, it is more beneficial to use either uplink sharing mode or dedicated mode depending on the value
of L parameter.
 11

A more sophisticated MS strategy is proposed in [52], which takes the link quality of both D2D and cellular users, the
interference situation (cross-tier interference from DUE to cellular network) for each possible mode and the load situation of the
cell into account for a multi-cell scenario. The MS strategy proposed in this work is as follows. Initially, the D2D terminals send
probing signals to each other and estimate the received signal powers. Then, the D2D terminal estimate interference plus noise
power in both uplink and downlink. Next, the obtained information is sent to the eNB, when it can decide about the amount of
resources it would allocate to the DUE in UL/DL based on cellular load as well as the maximum transmit power of DUE for the
different direct modes. Then, eNB estimates the expected SINR for each communication mode and the expected throughput
based on SINR and available amount of resources for each communication mode. Finally, the communication mode with the
highest throughput is selected. The result of this study provides an improvement of 50% in sum-rate with limited interference to
the cellular network. However, PC was not considered in this scheme and it was assumed that the BS has all the CSI available to
choose the best resource sharing mode.
Furthermore, in [41] MS has been exploited to avoid the co-tier interference between DUE and as discussed earlier in this
Section, the cross-tier interference has been solved using spectrum splitting. In the proposed scheme DUEs can only exploit
cellular mode or dedicated mode. Since, the spectrum in the dedicated mode is shared by multiple DUEs, co-tier interference is
generated between DUEs. Therefore, to solve this issue the DUEs can measure the activity in the D2D spectrum and employ a
carrier sensing threshold to determine their communication mode in a distributed manner. If the measured energy is smaller than
the threshold, D2D mode is selected since it indicates that there are not many ongoing D2D communications close by. If not, the
DUE select cellular mode.
Different from the other works in this section the authors of [53], consider a dynamic MS procedure to limit the cross-tier
interference between cellular and D2D users. They proposed three routing modes (D2D, cellular and hybrid) for D2D
communications underlaying cellular networks and then combined them with different resource allocation restrictions to results
in seven communication modes to model both the semi-static and dynamic mode selection using Discrete Time Markov Chain
(DTMC).
Another mean to further reduce the cross-tier interference and co-tier interference is to consider MS jointly with PC
and/resource allocation. References [54], [55],[56], [57] study the mode selection algorithm jointly with power control in D2D
underlaying a cellular network in a single cell scenario.

5.4. Radio Resource Allocation

Radio resource allocation (RRA) mainly addresses the research issues of how to optimally assign frequency resources to a
group of or all D2D pairs so as to optimize some performance metrics. The available works under this category can be further
classified based on their underlying algorithm.

5.4.1. Fractional Frequency Reuse (FFR)

Fractional frequency reuse (FFR) is a cross-tier interference management technique to mitigate the mutual interference
between CUEs and DUEs in the D2D-enabled cellular network. The basic idea of this scheme partitions the cell regions into
spatial regions and it also divides the entire frequency spectrum into several sub-bands such that (i) cell-edge users of adjacent
cells do not interfere with each other and (ii) interference received by (and created by) cell-interior users is reduced, while (iii)
use total spectrum than conventional frequency reuse. Then, each sub-band is differently assigned to each region. Since the
frequency bands for the cell-interior are not overlapped with neighboring cells, interference from neighboring cells can be
mitigated. In [58], the authors propose a frequency sharing mechanism that uses four resource groups (f1-f4) to reduce cross-tier
interference between CUEs and DUEs. In this scheme, each cell is split into two zones: cell inner and cell outer and different
frequency bands are allocated to CUEs and DUEs according to their locations. At first the location of the DUE is identified,
where they are located in the inner region or outer region. Afterwards, if the DUEs are located in the inner region of the eNB,
they can use the frequency bands that the eNB relaying UEs do not use in the current cell. Otherwise, if the DUEs are located in
the outer region, they can reuse the frequency bands except for the sub-bands used by eNB relaying UEs in identical cell outer
region.
In [59], another interference management scheme for D2D enabled cellular network is presented based on FFR to avoid cross-
tier interference between cellular and D2D users. In this scheme, the available frequency band is allocated to each cell based on
3-sectored FFR. The cell coverage is portioned into “cell center region” (corresponding to 63 percent of the total cell coverage
area) and “cell edge region” including three sectors per each region. The cell center region has a reuse factor of one, while the
cell edge region uses the reuse factor of three. The whole frequency band is divided into four parts and one of them is assigned
to the cell center region, and the rest of the band is assigned to the three edge regions. Figure 7 shows the frequency allocation.
The sub-band X is used in the cell center region (C1,C2, and C3), and sub-bands Y,Z, and D are used in region R1,R2 and R3,
respectively. To reduce interference between CUEs and DUEs, the DUE chooses the sub-bands which are not used in the cellular
user sub-area. For CUE, different sub-bands are allocated to each LTE-A cell sub-area according to the sectored-FFR. In
addition, If DUE is located in cell center region then it excludes the resided sub-bands which are used already by CUE in cell
center as well as the one that is used by CUE in the edge sub-areas of the current sector. Their idea is that the signal strength of
the two related sub-bands is so high that can seriously interfere with the DUE in the inner sub-area.
 12

Since both of the above mentioned approaches exploit a fixed portioning, they would potentially suffer from low throughput
due to inefficient use of the bandwidth resources. To reduce cross-tier interference for bandwidth sharing, a dynamic resource
portioning approach in frequency and/or time can be used. Conversely, in [60] they state that utilizing static FFR limits the
benefits of D2D reuse. So, they propose a dynamic FFR approach to alleviate the co-tier and cross-tier interference. This scheme
finds the FFR patterns among interfering sectors; and includes a scheduling approach which jointly schedules cellular and D2D
traffic. The scheduling problem is NP-hard and solved using two different low complexity algorithms. The results show that the
proposed scheme can provide 2.79 times better improvement in terms of throughput compared to the scheme in [49].

5.4.2. Time -frequency hopping

Time-frequency hopping is another technique used to reduce co-layer interference and cross-layer interference in D2D enabled
cellular network. The basic mechanism of time-frequency hopping is to partition the entire transmission period into small periods
during which a user does not transmits over the entire spectrum but only during short periods and remains silent for the other
periods.
In [61], the authors propose time hopping (TH) to limit the co-layer interference between DUEs produced by the imperfect
spatial reuse. This approach intends to randomize the near-far interference from nearby transmitting device
R2
CU: Z
DUE: X, Y, D

C2
D2D: Y, D
R1 C1
CUE: Y D2D: Z, D
DUE: X, Z, D C3
D2D: Y, Z

R3
CUE: D
DUE: X, Y, Z

Figure 7: FFR-based frequency band allocation in [59]

pairs/clusters and CUEs as shown in Figure 8. The clusters formation algorithm is performed by eNB based on maximum D2D
transmitting power of UEs and an estimation of the geographical position of devices, for example in Fig. 8 pair 1 and 4 create
one cluster and pair 2 and 3 makes another one. Then, the eNB assigns orthogonal resources in time/frequency domain to solve
the mutual interference of nearby DUEs in one cluster. However, to solve the interference among the different groups, a TH
approach is utilized. The TH procedure is as follows: the radio resources are divided into several orthogonal groups in the time
domain, in which each group has a predefined periodical transmission timeslots. Next, the eNB applies random sequence offset
at regular time intervals to semi-persistent scheduled resources. This prevents one pair from excessively suffering near-far
interference by another pair for a longer duration.
In addition, a time-frequency hopping based method is also carried out in [13] to randomize the co-tier interference between
DUEs for a large D2D enabled cellular network. The time axis is divided into consecutive operation slots, and at each slot the
potential DUEs can either be active (i.e., in D2D mode) or silent (i.e., in cellular mode). In this scheme, each DUE chooses its
operation mode (i.e., D2D or cellular mode) at each time slot independently according to a predefined TH probability, and access
each subband independently with a predefined frequency hopping (FH) probability. The FH probability, results in a tradeoff
between frequency efficiency and additional interference.

5.4.3. Graph Theory

Graph theory has been used for interference management in D2D underlay in several recent works [24]–[26]. An interference-
aware graph based resource sharing scheme has been developed in [26] to mitigate the cross-layer interference in the DL.
 13

Pair 1
Pair 4
Pair 2
Pair 3
Pair 5

Figure 8: Time Hopping technique based in [61]

The scheme includes two phases; in the first phase, the BS first constructs an interference graph which corresponds to the
network topology. In the graph each vertex has three attributes: link attribute, the resource attribute, and the cluster attribute. The
weight of edges of the graph (the link between two vertices) is calculated as the sum of the mutual interference between two
vertices. In the second phase, a resource assignment algorithm is applied to allocate RB to different cellular and D2D
communication links. The algorithm operates through an exhaustive search and thus involves high complexity. Therefore, an
) )
optimal algorithm is proposed with the complexity ) (M represents the number of the traditional CUEs in the
network, N is the number of the D2D pairs in the network, and K denotes the number of RBs). The proposed algorithm has been
compared with two resource assignment algorithms (greedy orthogonal approach and the exhaustive search approach) in terms of
sum rate. The results verify the low complexity of the proposed algorithm. The authors further extend their work in [27].
In [24] the authors propose to use graph coloring approach based resource allocation to mitigate the cross-tier interference
between CUEs and DUEs in the UL and increase the spectrum utilization of the system. In the proposed scheme, extra
functionalities are added for UEs; in which D2D transmitters use a simple interference information mechanism to detect the
interference caused by cellular transmissions to D2D communication and send this information to the eNB notifying it of the
interference in their vicinity. Then the eNB uses this information to construct the network topology using node contention graph.
The graph of the topology is drawn as follows: vertices represent either UEs with allocated resources (cell UEs) or pairs of intra-
cell communication requests, edges represent interferences between vertices implying that connected vertices cannot use the
same resources simultaneously and extra edges connect the cell UEs with each other, visualizing the primary resource allocation.
After creating the interference graph, the resource allocation problem among CUEs and DUEs is solved by performing an
interference-free secondary resource allocation using graph coloring such to maximize the spectrum utilization. To this end, three
coloring methods are presented: greedy algorithm, random sequential algorithm and repeat random sequential algorithm to reuse
the resources allocated to cellular users.
In [25] and [62], a resource allocation scheme has been proposed using bipartite graph strategy. The authors of [25], consider
resource allocation in D2D underlay where multiple DUEs coexist with multiple CUEs to mitigate the mutual cross-layer
interference between CUEs and DUEs. This algorithm groups DUEs, and each group can use the resources of only one CUE. An
optimal resource allocation scheme is proposed using Kuhn-Munkres (KM) algorithm; an efficient bipartite matching method in
the graph theory. The proposed algorithm is compared with random resource allocation and the results show that the
computational complexity is reduced from )to ) and the KM based approach achieves 30 Mbit/s gain on average
capacity compared with random allocation and 90 Mbit/s gain with the traditional cellular network without D2D mode.
Similarly, in [62], the authors propose a two step scheme to reduce the cross-tier uplink interference for intra-cluster D2D
scenario. In the presented scheme a D2D pair can only reuse the resources of one CUE only if the mutual interference between
D2D and cellular users is within an acceptable range. The resource allocation process is as follows: (i) A weighted bipartite
graph [63] is constructed to show the pairing relationship of D2D pairs and cellular users that share the same resource pool. (ii)
The resource allocation problem is then transformed into maximum weighted matching. Then an improved Kuhn-Munkres
algorithm is introduced to solve the mixed-integer nonlinear programming problem of the constructed bipartite graph, by which
it solves the resource allocation problem for the cluster. Their proposed scheme has been compared with the interference
avoiding method and greedy matching algorithm in terms of transmission capacity. The simulation results show that the
transmission capacity is increased with the proposed scheme. However, in this scheme the paper does not discuss how to form
clusters of DUEs.
 14

In [64], the authors proposed a dynamic graph framework for multihop D2D communications in a scenario with multiple nodes
and D2D pairs to implement interference management. The global network graph model contains all the probable transmission
states of cellular direct transmission, and a graph-based optimal routing algorithm is proposed for two multihop D2D
communications: connected and opportunistic transmissions.

5.4.4. Cognitive Radio Networks

Exploiting cognitive radio based on distributed licensed channel sensing and making interference-aware decisions can be used
in D2D enabled cellular networks for interference mitigation. In [65], an efficient cross-tier interference management scheme to
avoid both interference from CUE to DUEs and interference from DUE to CUE in a scenario where one DUE share the resources
of only one CUE is proposed. The DUE is capable of sensing the radio spectrum and measuring the received interference level of
CUE during UL transmission, and passing these values to the BS. The BS then uses this information in an interference aware
resource allocation scheme. The results show a three-fold increase in cell capacity.
A centralized cognitive based resource allocation algorithm between CUE and multiple DUE is presented in [66] to mitigate
cross-tier interference between CUE and DUEs. The authors first formulate the optimal resource allocation policy as an integer
programming problem and since the optimal resource allocation solution is NP-hard they instead propose a cognitive based two-
phase low-complexity suboptimal solution approach. In the first phase, they extend the max-min fairness algorithm proposed in
[67] to perform fair resource allocation for cellular UL and DL flows. Then, in the second phase, a heuristic subchannel
allocation approach for D2D flows was conducted with rate protection for CUE.
In [68], another interference management scheme is presented based on cognitive radio, where a semi-distributed cognitive
spectrum access is proposed to make interference-aware decisions. The scheme avoids UL and DL cross-tier interference and co-
tier interference. In the proposed scheme, initially the BS assigns one of the free channels for D2D communication and
exchanges the ID of this channel with the DUEs in its coverage area. Once the DUEs are informed of the D2D channel, they are
responsible for initiating the communication session with no further supervision from the BS. This channel can also be used by
cellular transmission in that the BS schedules CUEs in any of the available channels expect the one assigned to D2D
transmission as long as the number of CUEs is less than the total number of available channels. The main idea of the scheme is
that, each D2D transmitter sends a request to its intended receiver on the channel assigned, and if the maximum interference
level received from any neighboring transmitter is sensed to be lower than a predefined sensing threshold, the intended D2D
receiver sends a signal to its corresponding transmitter. That is, the D2D communication link cannot be established if there is at
least one transmitter (i.e., CUE in UL or BS in DL) using the same channel inside this region. In addition, to avoid co-tier
interference between DUEs, after sensing the channel each D2D transmitter sets a random backoff timer upon expiration of
which the D2D transmitter can use the channel if the channel is still available.
In [15], the authors investigate cognitive radio assisted D2D communication for improved spectral efficiency, in which devices
access the network with mixed overlay-underlay spectrum sharing without the need of assistance. The devices sense the
spectrum to identify white spaces. A collision probability has been considered as the spectrum sharing criterion. Performance
metrics such as transmission capacity, isolation probability and connectivity have been derived using stochastic geometry.

5.4.5. Clustering of D2D users

In [69], a two-phase scheme is presented to eliminate co-tier interference between D2D users and enhance the reliability of
D2D communication network in dense D2D deployment scenario. To avoid the cross-tier interference between DUE and CUE
orthogonal resources are used. To mitigate co-tier interference, the scheme encompasses a combination of frequency band
allocation and clustering of DUEs based on their geographical locations. In the first phase, the interference management problem
is modeled as a MAX k-CUT problem in graph theory to group strong interferers into different clusters under the constraint that
a target outage probability of the kth D2D pair in a specific cluster is met and then the interference graph is formed. To solve this
NP-hard problem a heuristic clustering algorithm is applied. The heuristic algorithm derives the maximum number of DUEs in a
cluster under the constraint that the target outage probability of each D2D pair in the cluster is met. Also, a stepwise removal
algorithm (SRA) is proposed to reduce the number of clusters to . In the second phase, a heuristic channel assignment
algorithm is applied. The performance of the proposed scheme has been compared with static clustering in [70] and with a non-
clustering scheme. The results demonstrate that the smaller D2D distance, the higher is the throughput that can be achieved.
Also, the larger the size of cluster, the more the number of D2D pairs will be assigned to channel and finally, the higher the
average throughput of the concerned D2D network. The proposed scheme achieves 290% increase in throughput in compared to
[70] and 475% increase in compared to the scheme with no clustering.

5.4.6. Game Theory

Game theoretic models can be used to design and analyze distributed resource allocation schemes since it can provide an
 15

insightful understanding to the complex interactions among independent rational players. Two broad categories of a game
theoretic model for D2D resource allocation are non-cooperative and cooperative. In the former type, DUEs are commonly
viewed as players competing for the resources. Reference [19] investigated the resource allocation issue for intercell interference
scenarios where a D2D link is located in the overlapping area of two neighboring cells using non-cooperative static Cournot
game. The aim is to mitigate the mutual inter-cell interference between DUEs and CUEs. In this work, BSs are considered as
players competing for resource allocation quota from D2D demand. Three scenarios are considered: (i) CUEs in both cell use UL
resources (ii) CUEs in one cell use DL resource, while the cellular users in the other cell use UL resources and (iii) CUEs in both
cells use DL resources, and the corresponding interference is modeled in each scenario. References [20][71] and [72] addressed
the D2D resource allocation using auction games. In particular, [71] introduced a sequential second price auction as the
allocation mechanism for D2D underlay communication to mitigate the mutual cross-tier interference between DUEs and CUEs.
In the proposed scheme, all the spectrum resources were considered as a set of resource units auctioned off by groups of D2D
pairs in sequence. The authors of [71] further extend their work in [20] in which they propose a resource allocation scheme for
D2D communication based on the reverse iterative combinatorial auctions (I-CA) approach to allocate the DL resources between
the cellular and D2D UEs in a distributed manner. In I-CAs, the bidders submit multiple bids iteratively, and the auctioneer
computes provisional allocations and asks prices in each auction round. The resources as the bidders compete to obtain business,
while D2D links as the goods or services wait to be sold. And then a non-monotonic descending price auction algorithm is
explained depending on the utility function that accounts for the channel gain from D2D and the costs for the system.
Additionally, in [72] a reverse iterative combinatorial auction was introduced as the allocation scheme to optimize the system
sum rate as an underlay in the DL cellular networks. In [21] authors developed a Stackelberg game to solve the optimization
problem, in which a CUE and a DUE to form a leader-follower pair to control the interference from DUE to cellular networks. In
this game, the CUEs are leaders of the game and who own the channel resources, and the D2D pairs are viewed as the followers
and charged with some fee for using the radio resources. Two optimization problems are solved in this work. Then a joint
scheduling and resource allocation scheme which considers the fairness of DUEs is proposed. The authors extend their work
presented in [21], by taking the D2D rate in the Stackelberg game into account in [22].
Furthermore, the authors of [73] social aware resource allocation to mitigate the cross-tier interference and co-tier interference
using cooperative game theory. The social community aware resource allocation is formulated as a social group utility
maximization problem using the social graph and interference graph, and solved using social group utility maximization game
(SGUM) [74], [75]. The results show that the proposed SGUM game based solution increases the sum utility rate of overall
social groups about 16% to 50% without loss of fairness compared to coalition game [76], nearest first, furthest first and random
selection algorithms.
In addition to the above non-cooperative games, a cooperative game for D2D resource allocation has also been explored.
References [77][76] [23] addressed RA problem using coalition game. In [77] a distributed resource sharing allocation scheme in
D2D communications underlaying cellular networks based on a coalitional game with transferable utility is developed. In this
game, each user intends to maximize own utility and has the incentive to cooperate with other users to form a strengthened user
group that can increase the opportunity to win its preferred spectrum resources. In [76] they investigate large scale networks, and
the problem of resource allocation by considering the cross-tier interferences between the DUE and CUE. Specifically, they
address the resource allocation problem for multiple D2D pairs and CUEs using coalition formation games. The results show that
the system sum rate is increased by about 20% to 65% compared to other strategies. They focus on the scenario of a single cell
involving all the CUEs under its coverage, and only consider the intra-cell interference caused by the coexistence of the cellular
and D2D communications. Moreover, in reference [23], the problem of energy-efficient uplink resource sharing scheme based
on a coalition formation game for mobile D2D multimedia communications underlaying cellular networks with multiple D2D
pairs and cellular users has been addressed. The nontransferable coalition formation game accounts for the gains in terms of
energy efficiency and the costs in terms of mutual interference. Moreover, a distributed coalition formation algorithm based on
the merge-and-split rule and the Pareto order is presented.

5.4.7. Joint Mode Selection, Power Control and Radio Resource Allocation

To realize the full potential of D2D communications, the RA should be done jointly with the MS and PC. Several works have
investigated this joint problem, such as [54], [78], [31], [23], [79], and [80]. They mainly focus on the interference control and
management between D2D links and cellular links such that they can efficiently reuse the radio resources whenever the
interference is small. Related works mainly develop mixed-integer programming models and deal with NP-hard formulations. In
particular, the work presented in [54] is an enhanced work of [44]. In [78], a centralized joint MS, PC and RA scheme for D2D
enabled in the cellular network has been formulated as an optimization problem and limit the cross-tier interference from DUEs
to CUEs and vice versa . The optimization problem has been decomposed into two sub-problems: power control for both DUEs
and CUEs and joint MS and RA for each DUE and solved individually. The joint MS and RA is NP-hard, thus the authors
propose two low complexity algorithms to solve the problem according to the load situation of the network: heavy, medium, and
light. In the heavy load scenario the DUEs can only select the sharing mode since there are no free channels in the system. In this
case, the RA problem is solved using Hungarian algorithm. Alternatively, in the medium load case, some DUEs select the
dedicated channels while other must work in the reuse mode and share channels with CUEs. In the light load scenarios the DUEs
can either select the cellular mode or dedicated mode and don’t require to reuse the cellular resources since the number of free
 16

channels is larger than that of D2D pairs. Then, two heuristic algorithms have been proposed for the light load and medium load
scenarios, Algorithm 1 and Algorithm 2 respectively. The complexity of Algorithm1 is ), while the complexity of
Algorithm 2 is ). For evaluation purposes the proximity gain, the hop gain and the reuse gain of the two proposed
algorithms are compared with the optimal branch-and-bound (BB) scheme under light, medium and heavy load conditions. The
results show that the two algorithms perform very close to the equivalent prime algorithms.
The authors of [31] formulate a centralized delay-aware resource allocation and power control framework using queuing
model to optimize the three resource control action. Different from related works in the literature [54],[81],[77],[82] and [56], in
which they assume that the cellular and D2D users are saturated with infinite backlog traffic and constantly cause interference to
the other D2D links or cellular link; this study provides an optimization framework for the delay-aware resource control policies
with more realistic bursty traffic.
In [79], the target is to maximize the overall sum rate of all DUEs and CUEs by jointly considering centralized PC, RA and
MS. The interference from DUEs to cellular link is limited by using PC, and the cellular interference to DUE is reduced through
pairing optimization.
The authors of [80], proposed the joint optimization of MS and RA such that the system capacity is maximized. This work
considers resource sharing between CUEs and multiple D2D pairs in an underlay communication. The cross-tier interference
between multiple DUEs and a single CUE is considered. The problem is to select a set of users which currently operate in
cellular mode to switch to a D2D mode such that the overall sum rate increases. They first formulate the problem as a mixed-
integer non linear programming (MINLP) problem, and then propose a two-phase algorithm for solving the NP-hard problem
with low computation complexity. For each UE pair a parameter is defined which is the channel gain between the UE
transmitter and the receiver divided by the channel gain between the transmitter of UE pair and the BS. The larger the value of
alpha the more preferable UE pair is to operate in D2D mode. The algorithm operates as follows:
Phase 1: Each pair calculates the maximum achievable rate if it is switched to D2D mode, and if the rate is high it is switched to
D2D mode. After the mode of operation for all UE pairs have been updated, sum rate of all UE pairs is calculated and designated
as the maximum sum rate of the system so far.
Phase 2: The power allocation of each D2D pair is updated one-by-one based on the values of .
As it is shown in the left of Fig. 9, initially three pairs of UEs with intra-cell traffic are operated in the cellular mode. After the
BS collects pertinent CSI data, it can perform the proposed joint optimization of MS and RA for determining the optimal modes
of operations for each pair. The right of Fig.9 shows the changes after the proposed algorithm has been performed. If only Pair 3
stays in the cellular mode, then for Pair 1 and Pair 2 the powers to allocate for transmissions in individual resources are also
decided for them to operate in the D2D mode. For evaluation purposes, the proposed algorithm has been solved using Basic
Open-source Nonlinear Mixed Integer Programming (BONMIN) and compared with the proposed scheme. The results show that
the proposed algorithm can achieve 82%-88% of the optimal values obtained by BONMIN for various scenarios while incurring
a lower computation cost. Also, compared to the conventional mode of operation (cellular mode only), the performance gain
increases beyond 200% to 400% as the number of UE pairs increases from 5 to 15.
CUE 1 CUE 1

Pair 1

Pair 1
Pair 2 Pair 2

CUE 2 CUE 2

Pair 3 Pair 3

Pair 1 Pair 2
Pair 1 Pair 2 Pair 3 CUE 1 CUE 2 Pair 3 CUE 1 CUE 2
Pair 2 Pair 1
Figure 9: Mode selection based on [80]

5.5. Advanced Antenna and Receiver Schemes

Interference management using advanced antenna and receiver is a technique which will assist in interference management. It
will detect and even try to decode the symbols of the interference signal. Then based on a detector output, the
interference signal can be reconstructed and cancelled from the received signal so as to improve the anticipated signal decoding
performance. In the following each of the techniques using advanced receiver is discussed. Based on this, a qualitative
comparison of each of the approaches described in this section is summarized in Table 3.

5.5.1. Multi-input Multiple Output transmission schemes

 17

Muli-input multiple-output (MIMO) antenna systems have become a key feature of many advanced cellular wireless network
standards [83], [84]. In MIMO systems, each BS uses an antenna array to serve multiple users in each time-frequency resource
block. MIMO transmission techniques can be incorporated in D2D communications underlaying cellular networks to further
avoid interference among different UEs. If the number of antennas at a BS is significantly larger than the number of served users,
the channel of each user to/from the BS is nearly orthogonal to that of any other user. This implies that, with a large antenna
array at a BS, D2D signals possibly result in close-to-zero interference at the uplink.
All the previous techniques explained for interference management on D2D underlay cellular network until now focused on a
single antenna, while research performed in [17], [85]–[89] exploit MIMO scheme. To mitigate or avoid mutual interference
between cellular and D2D transmission, [85], [87], [88] considered precoding. In particular, the authors of [85] utilized MIMO
schemes particularly heuristic precoders at cellular downlink (eNB) that avoid generating cross-tier interference from an eNB to
a D2D receiver underlaying the same resources by aligning the transmitted signal from the eNB to the null space of the eNB-
D2D interference channel. To implement this scheme the BS should know the interference CSI which causes increased
complexity. The D2D link quality in term of SINR was improved up to 15 dB and 10% increase in the total cell capacity.
Table 2: A qualitative summary of the related works in interference coordination technique
RRM Functionsb Interference
Interference Reuse Network
Ref Target Solution Control
Typea PC MS RA Resourcesc Complexityd
Level
Maximize the rate of DUEs under
[41] √ Optimization --- SC, 1 CUE, 1 DUE Centralized
target rate constraint for CUEs
Semi-
[61] √ Improve SINR Time Hopping UL MC
Distributed
DUE-DUE Stochastic geometry, Time MC, N DUE, N
[13] √ √ SINR, average rate DL Distributed
Hopping CUE
Maximize throughput, guarantee Cluster-based heuristic SC, N DUD, 1
[69] √ UL.DL Centralized
reliability algorithm CUE
Deviation of interference, SINR --- UL SC, 1 CUE, 1 DUE Centralized
[45] √ √
Spectrum utility maximization Evolutionary game UL & DL SC, N DUEs, N Distributed
DUE-CUE
[30] √ √ √ CUEs

[46] √ Maximize throughput while Heuristic UL SC, 1 CUE, N Centralized

preserving SINR for the CU DUD
[14] √ Reduce outage probability Stochastic geometry UL SC Distributed
DUE-eNB [19] √ Sum rate, sum rate gain Static cournot game UL & DL 2C, 1 DUE, 2 CUE Centralized
[21] √ √ Improve sum rate of D2D Stackelberg game UL SC, 2 DUE, 1 CUE Centralized
[22] √ √ Maximize network throughput Stackelberg game, pricing UL SC, N DUE, 1 CUE Distributed
algorithm
[43] √ Throughput under SINR MC , N DUE, N
Basic Math derivations UL & DL Centralized
constraint CUE
[44] √ √ Linear Optimization
Maximize Throughput UL & DL SC, 1CU, 1 DUE Centralized
(Heuristics)
[48] √ PC, SINR Statistical UL/DL SC, 1 CUE, 1 DUE Centralized
DUE-eNB ,
[49] √ √ √ Linear Optimization MC, 1 CUE, 1
DUE-CUE Maximize Throughput
(Heuristics)
UL & DL
DUE
Centralized
[59] √ Increase throughput and reduce
Fractional Frequency Reuse UL/DL MC Centralized
overall energy cost
[66] √ Max-min fairness and a
Fair resource allocation UL & DL N D2D, N CUE Centralized
cognitive based approach
[56] √ √ Maximize energy efficiency of Exhaustive Search (Brute SC, N DUE, N
UL Centralized
DUE-CUE, system Force) CUE
CUE-DUE [15] √ Coverage probability, Interference Poisson Point Process and MC, N DUE, N
DL Distributed
probability, connectivity Stochastic geometry CUE.
[26] √ Maximize network sum rate Non linear programming DL SC, 1 DUE, 1 CUE Centralized
problem solved by Graph
theory
[20] √ Optimize system sum rate Reverse Iterative DL SC, N DUE, N Distributed
combinatorial auction game CUE
theory
eNB-DUE, [71] √ Optimize the system sum rate Sequential Second Price DL SC, N DUE, N Centralized
auction CUE
DUE-CUE [72] √ Optimize the system sum rate Reverse Iterative DL SC, N DUE, N Distributed
combinatorial auction game CUE
theory
[29] √ √ Maximize system throughput with Particle Swarm Optimization DL SC, N CUE, N Centralized
minimum rate guaranteed DUE

[24] √ SC, N DUE, N

Increase data rates Graph Coloring UL Centralized
CUE
[62] √ SC, N DUE, N
Increase capacity Kuhn-Munkres algorithm UL Centralized
CUE
DUE-eNB, [23] √ √ √ Coalition formation SC, N DUE, N
improve spectral efficiency UL Distributed
CUE-DUE algorithm CUE
[78] √ √ √ SC, N CUE, N
To maximize overall throughput Heuristic UL Centralized
DUE
[31] √ √ √ Mean throughput, Average delay, Queuing, Decision Process SC, N DUE, N
UL Distributed
dropping probability Model CUE
 18

[79] √ √ √ Maximize Sum rate while Linear Optimization UL N DUE, N CUE Distributed
guaranteeing the QoS of both
cellular and D2D users
[80] √ √ √ Maximize sum rate Mixed Integer Non-Linear UL SC, N CUE, N Distributed
Programming and a two DUE
phase algorithm
CUE-DUE, [52] √ Maximize sum rate Linear Optimization UL & DL SC & MC, 1 CUE, Centralized
eNB-DUE (Heuristic) 1 DUE
[16] √ Maximize sum rate and increase Stochastic geometry SC, 1 CUE, N DUE Centralized &
DUE-DUE, coverage probability UL/DL Distributed
DUE-eNB,
[60] √ Increase throughput Dynamic Fractional UL/DL MC Distributed
DUE-CUE Frequency Reuse

RRM Functions Interference

Interference Reuse Network
Ref Target Solution Control
Type PC MS RA Resources Complexity
Level
DUE-DUE, [68] √ Channel access probability, Cognitive spectrum access UL & DL -- Semi-
eNB-DUE, spectral efficiency Distributed
CUE-DUE
DUE-CUE, [65] √ √ Increase capacity --- UL &DL MC, N DUE, N Centralized
DUE-eNB, CUE
CUE-DUE
CUE-DUE, [54] √ √ √ Maximize Sum-rate Linear optimization UL & DL SC & MC, 1 CUE, Distributed
eNB-DUE, (Heuristic) 1 DUE
DUE-eNB, [25] √ Increase system capacity Kuhn-Munkres Algorithm UL/DL N DUE, N CUE Distributed
DUE-CUE
a
Interference type: interference between D2D pairs (DUE-DUE), interference from a D2D pair to a CUE (DUE-CUE), interference from a
CUE to a D2D pair (CUE-DUE), interference from a D2D pair to eNB (DUE-eNB), interference from eNB to D2D pair (eNB-DUE),
b
RRM functions: power control (PC), mode selection (MS), resource allocation (RA),
c
Reuse resources: uplink (UL), downlink (DL), uplink and downlink (UL & DL), uplink or downlink (UL/DL),
d
Network Complexity: single/multi-cell (SC/MC), single/multiple D2D pairs (1/N DUE), single/ multiple CUEs (1/N CUE),

In [87] they propose several new algorithms for cooperative MIMO precoder pair selection for D2D DL to mitigate intra-
cell cross-tier interference (eNB-DUE and DUE-CUE) where each cellular user shares resources with only one D2D link. The
precoding is used on both eNB and D2D transmitter. One class of precoding schemes decouples the precoder pair selection
through maximizing SLNR by considering signal and interference strengths from the perspective of the transmitter. The second
scheme imposes a restriction on SINR at the other co-channel receiver.
In [88], a joint channel allocation, power control and precoding scheme of the DUEs is considered in a D2D-enabled uplink
MIMO cellular network to mitigate the cross-tier interference between the CUE and multiple DUEs. In this scheme, multiple
DUE pairs share the radio resources of multiple CUEs. Optimal RA, PC and precoding are obtained by using a random search.
Moreover, a distributed non-cooperative RA game for the joint channel allocation, PC, and precoding of the D2D users is
formulated. The feasibility and existence of the pure strategy Nash equilibrium are then established by developing a self-
optimizing algorithm, whichshows that Nash equilibrium is achieved under a special condition. Finally, a distributed RA
algorithm based on best response dynamic is proposed, which can also achieve Nash equilibrium.
Papers [89]–[91] studied various relaying strategies to mitigate the cross-tier interference between cellular and D2D
communication. A TDD single cell with multi-user MIMO architecture is considered in [90], in which there are multiple CUEs
and multiple DUEs and the BS is equipped with multi-antennas. The authors formulate an optimization problem to maximize the
overall rate of the CUEs and D2D pairs, which is NP-hard and non-convex. Two steps are taken to solve the optimization
problem heuristically.
At first, three precoder techniques based on conventional beamforming (BF) method [92] and zero-forcing (ZF) methods in [93],
[94] is applied in multi-user MIMO architecture. The conventional BF method serves each CUE without performing interference
cancelation for other CUEs and DUEs. The ZF beamforming method cancels out inter-UE interference by transmitting the
information data of a cellular user in the null space of all other CUEs. For the ZF beamforming method, to cancel the
interference from eNB to DUE, the CUE’s data is transmitted in the null space of all DUE channels. Next, a binary D2D pair
association vector search algorithm is proposed to maintain the interference from DUEs to CUEs less than a certain level so that
the SINR of CUEs is met. The results show that ZF beamforming method gives the best performance in terms of transmission
sum rate and thus is more protective for DUEs.
The authors in [89], proposed a joint precoder-decoder schemes for MIMO D2D communications using physical layer network
coding (PNC) mapping to enhance the network performance and mitigate the interference. The relay station (RS) performs PNC,
and bi-directional communication is completed in two time slots. In the first time slot, the D2D communication is in multiple
access (MA) stage and both D1 and D2 transmit to RS, and BS transmits to mobile station. In the second time slot, the D2D
communication is in the broadcasting stage (BS) and the RS broadcasts the PNC mapped signal to devices as well as the MS
transmission to the BS. The optimization problem for joint precoder-decoder is formulated for each time slots in terms of
minimizing mean
square error (MSE) and enhancing the accuracy of estimation. It is assumed that the available powers at nodes are limited, and
 19

interference thresholds are used to reduce the interference to the other link. Distributed algorithm is proposed for D2D
communication and BS-MS in each time slot to solve the optimization problem which is non-convex. The results verify that the
proposed scheme is fast enough to be applicable in wireless networks.
In [91], the authors extend their previous research presented in [89] by considering two transmit modes for D2D as two-way
relaying mode which uses the PNC technique introduced in [89] and direct D2D mode. The authors propose a distributed scheme
based on a greedy algorithm for PNC based D2D mode and a centralized multi-criterion scheme for precoder-decoder
optimization which is solved using the scalarization method.
In contrast, [86] proposed not to schedule UL users that may generate excessive interference to DUEs to improve the overall
network throughput. In this work the interference from DUE to CUE is solved using PC and the UL interference from CUE to
DUE is handled via an interference limited area (ILA) control technique. The authors consider a single-cell uplink network
where multiple CUEs and one D2D pair can share the same radio spectrum when the BS is equipped with multiple antennas. It is
assumed that the interference from D2D transmitter to the CUE is negligible, since PC is applied to D2D transmitter. However,
to alleviate the interference from CUEs to D2D receiver, the - interference limited area (ILA) control scheme is determined
which is the area where the resulting interference-to-signal-ratio (ISR) at the D2D receiver becomes higher than a predefined
threshold, . In other words, this scheme does not allow CUEs located in to use the same resources as the DUEs.

5.5.2. Interference Regeneration and Cancellation

Interference cancellation (IC) is regarded as a promising technique to cancel intra-cell and inter-cell interference and improve
the network capacity. The basic mechanism of IC technique is to regenerate the interfering signals and subsequently cancel them
from the desired signal after the signal is received to enhance SIR of the received signal. IC techniques include successive
interference cancellation (SIC), parallel interference cancellation (PIC) and iterative interference cancellation (IIC) [95]. PIC
detects all users simultaneously. This initial rough estimate can then be used to cancel some interference, and then parallel
detection can be repeated. SIC detects just one user per stage. The strongest received signal is detected first, then the next
strongest, and so on. After each user’s transmitted data is estimated, the received signal for that user can be reconstructed by
recreating the transmit signal and applying an estimate of the channel to it. This can be subtracted from the composite signal,
which then allows subsequent users to experience a cleaner signal.
In this scheme, the SIC receiver first decodes the strongest interfering signal by treating other signals as noise. Then it
regenerates the analog signal from the decoded signal, and cancels it from the received composite signal. After this stage, the
remaining signal is free from the interference of the strongest signal. Then, the SIC receiver proceeds to decode, regenerate and
cancel the second strongest interfering signaling from the remaining signal and so forth, until the desired signal can be decoded.
In [18], the authors consider SIC technique for D2D enabled cellular network to model the cross-tier interference between
cellular and D2D users using the tools from stochastic geometry. The SIC scheme is utilized at both cellular receivers (BSs) and
D2D receivers to identify the performance of SIC on large-scale D2D enabled cellular networks. In this work, the authors have
assumed that the two-tier interference (interference from device tier and interference from cellular tier) can be represented by an
equivalent single tier-interference. The successful transmission probabilities of cellular and D2D links has been derived from
infinite and finite SIC capabilities.
In [96] the authors propose an IC technique to which exploits retransmission of the interference from the BS to reduce intra-
cell interference from CUE to DUEs and thus improve the reliability of the D2D communication in the UL period without
reducing the power of CUEs. They introduce two conventional receive modes in cellular networks with underlay D2D
communication: (i) MODE 1: the D2D receiver decodes the desired signal while treating the interference as noise [97], (ii)
MODE2: the D2D receiver first demodulates interference and then cancels it before demodulating the desired signal [98]. Mode
1 is suitable in low interference regime, while Mode 2 is superior in handling very strong interference. However, these two
receive modes suffer in scenarios with a middle level of interference. Therefore, they proposed a new received solution named
MODE 3 to overcome the performance degradation in the middle area interference. The proposed scheme retransmits the
interference from the BS to generate the desired interference as strong interference, and then exploit IC process to obtain the
desired signal. Maximum Radio Transmission (MRT) is used for the retransmission of the interference signal. The outage
probabilities of the three modes are derived in close-forms. Based on the outage probability the authors suggest MS scheme:
Mode 1 used in low interference regime, Mode 3 used for middle interference regime, and Mode 2 used for very strong
interference regime.
In [99] an IC scheme is designed to mitigate the interference from CUE to DUEs based on the location of users. In this method
a dedicated control channel is allocated to DUEs. The scheme is as follows:
(i) At the beginning of D2D transmission, eNB broadcasts the reserved time-frequency resources of DUEs in the broadcasting
channel so that all the CUEs and DUEs know the information of this channel.
(ii) CUEs periodically listen to this channel to ensure whether DUEs are near to them. Accordingly, CUEs measure the SINR to
identify the adjacent DUEs.
(iii) If the sensed SINR is higher than a pre-defined threshold, a report of the measurement result is sent to the eNB.
 20

(iv) According to the received measurement results, the eNB stops scheduling CUEs on the resource blocks that are currently
occupied by DUEs. eNB also broadcasts the position information and allocated resources of near far risk CUEs. Thus, DUEs can
use such information to avoid reusing those resources at the specific location.
(v) When DUEs receive the broadcast message they distributivly perform RRM to mitigate interference from CUE.

5.5.3. Beamforming

In [100] the authors compare BF and IC schemes for D2D communications underlaying cellular networks in a single cell
scenario with single CUE and a single DUEs, where the BS is equipped with multi-antennas. It is observed that the BF strategy
outperforms IC strategy at the low SNR region due to its capability to direct available power to increasing cellular signal
strength. However, the IC strategy is preferred at medium to high SNRs because it can alleviate the dominating degrading
influence on the D2D receiver, i.e., substantial cellular interference. Thus, based on this observation, the authors propose an
adaptive scheme to switch between BF and IC strategies to maximize entire system throughput.
In [101], the authors exploit BF technique to design a D2D interference avoiding approach. In the proposed scheme, the BS can
facilitate the DUE to obtain the interfering CSI by linear minimum mean-square error (LMMSE) channel estimation method.
Then the transmit BF is applied at the DUE transmitter to direct the signals only towards the null space of the estimated channel
between the transmitting DUE and the BS. Meanwhile, the receive BF is applied at DUE receiver to receive the signals only
through the nullspace of the estimated channel between CUE and DUE-CUE. In this way the interference to BS and from CUE is
minimized.

5.5.4. Coding Schemes

A. Superposition Coding

In [102] the authors investigate cooperative relaying schemes for D2D communication underlaying cellular networks, and
propose two superposition coding-based schemes to mitigate the intra-cell interference. In the first scheme, the D2D transmitter
receives and decodes the signal broadcast by the cellular transmitter, and then regenerates the cellular signal and superposes it
with the D2D signal. DT broadcasts the composite signal to the cellular receiver and the D2D receiver. In the first scheme, all the
information of the cellular pair is relayed by the D2D transmitter, leaving a relatively small amount of radio resource for D2D
transmission. Thus, they propose a scheme to further exploit the transmission opportunity for the D2D pair. In the second
scheme, CT splits its signal into two parts and broadcasts these two parts in a superposition signal. DT relay only one part of the
superposition signal. The analytical results show that the second scheme outperforms the first scheme in terms of the achievable
rate of the D2D link.

B. Rate Splitting
In [103], the authors applied Han-Kobayashi rate splitting technique [104] in single cell scenario to mitigate the mutual
interference between D2D and cellular communication. In rate splitting, the transmitted message is divided into private and
public parts. The private part is only decodable by the intended receiver, while the public part can be decoded by both receivers
which are subject to the interference. This approach helps the DUE and CUE interference victims to cancel the interference
caused from the public part of the message by performing a best-effort SIC algorithm. The authors also derive the rate-splitting
factors analytically in a scenario with two interfering links and compares the rate splitting scheme with other resource sharing
methods in [45], [54] as well as the conventional cellular mode. The results show that the proposed scheme outperforms the other
schemes in terms of throughput.

5.5.5. Cooperative Multi-Point

Cooperative Multi-Point (CoMP) transmission and reception techniques utilize multiple transmit and receive antennas from
multiple antenna site locations, which may or may not belong to the same physical cell, to enhance the received signal quality as
well as decrease the received spatial interference. In [105], CoMP paradigm has been investigated for D2D communication to
mitigate intra-cell cross-tier interference from DUE to CUE and inter-cell interference. The interference from DUE to CUE is
mitigated using ZF precoding, which steers a beam towards CUEs direction and null in the direction of the DUEs. The inter-cell
interference is originated from eNBs of other cells. Inter-cell interference is mitigated using the coordination between the eNB.
The proposed scheme has been evaluated in the D2D offloading scenario.

5.5.6. Interference Alignment

 21

Interference Alignment (IA) is a linear precoding scheme which aligns the interfering signals at each receiver. It is shown that
all the interference can be concentrated roughly into one half of the signal space at each receiver, leaving the other half available
for the desired signal and free of interference [106].
In [107], the authors try to enhance the sum rate of the D2D communication underlay cellular network through IA process to
reduce co-tier interference between DUE. IA is used as a mechanism to align interference between potential interfering users.
The overall process of the proposed is scheme is as follows: (i) the D2D pairs which will be grouped together to share resources
are defined, (ii) each group is assigned multiple RBs, and (iii) IA is exploited to precode transmissions. The authors propose
three different grouping mechanisms and investigate their effect on the total D2D sum rate in the cell. It is assumed that an IA
group is limited to a number of six users or three D2D pairs to reduce complexity of precoding vectors calculations which
increases as the number of users correspondingly the required symbol extensions increase. The three grouping schemes are (i)
position-based grouping algorithm, (ii) channel-based grouping algorithm, and (iii) distanced-based grouping algorithm. The first
algorithm is based on the D2D transmitting nodes positions, and the DUEs are group in such a way that they have the minimum
containing area. In the second algorithm the users are selected based on the gains of the direct channel between each D2D pair
transmitter and receiver. In the last algorithm, the grouping is performed based on the distance between D2D transmitter and
receivers as an approximate measure for the channel path loss, and it takes the channel condition between IA group main pairs
into account. The criterion for group construction is such that D2D pairs in a group are those that would cause large mutual
interference on each other. When IA is applied this mutual interference should be eliminated and throughput gains can be
achieved. The proposed schemes have been compared with point-to-point (P2P) transmission in terms of Bit Error Rate (BER)
and sum rate. The results show that IA schemes have worse performance in terms of BER in compared to P2P, but IA schemes
are still able to achieve higher throughputs due to the excess degrees of freedom available in IA transmission schemes compared
to those available in the P2P transmission scheme.
Similarly, in [108] IA technique is exploited together with clustering of D2D pairs to fully reuse the available resource blocks
over these clusters and achieve a tighter reuse than [107]. Within each cluster they assemble the D2D pairs into IA groups where
each group is comprised of three D2D pairs. The overall procedure of their scheme is as follows: (i) clusters of D2D pairs are
formed based on transmitting nodes positions, based on Fuzzy C-Means Clustering (ii) IA groups (comprised of three D2D pairs)
within each cluster are formed between transmitting and receiving nodes using two grouping schemes; channel-based and
distance-based as in [107].
In [109], the authors propose two IA based schemes to manage the cross-tier interference between cellular and D2D users in a
D2D LAN underlaying a cellular UL, where multiple DUEs want to communicate with a D2D receiver. The first scheme is
referred to as interference-free IA scheme, which can be applied in the scenarios where some sub-channels of eNB are not
occupied by CUEs. In this scheme, the interference signals from DUEs are aligned in the orthogonal space of cellular links at the
eNB. Hence, the links of CUEs are completely free from interference. Note that if all sub-channels of eNB are used by CUEs, the
orthogonal space of cellular links may not exist. In such scenarios, they propose another scheme which is referred to as
interference-limiting IA scheme. In this scheme, the DUEs signals are allowed to occupy some links of CUEs, but the peak
interference power on each of the interfered links is kept under a certain threshold. It is shown that the interference limiting IA
scheme is most efficient for the scenarios where there are a large number of DUEs.

6. INTERFERENCE MANAGEMENT CHALLENGES IN D2D-ENABLED 5G NETWORKS

There are several open challenges and issues that need to be considered in D2D to become as a key enabler for 5G technology.
Although there has been extensive work performed on interference management in D2D communication based on previous
legacy networks (WiMAX, LTE and LTE-A), we believe interference management in D2D communication on future 5G cellular
networks is a research topic still in its infancy. We support this belief by looking at the research challenges which exists in the
recent literature to fulfill the requirements of 5G systems. The open challenges are explained in the following.

6.1. D2D in mmWave communication

mmWave band communication has recently received significant attention as one of essential features of 5G cellular networks
[110], [111][112]. In future 5G cellular networks, it is expected that the traditional copper/fiber infrastructure will be replaced
with mmWave mesh networks to allow 5G to offer rapid deployment and mesh-like connectivity. A mmWave communication
operates on a larger frequency band (30-300GHZ) and thus it has the potential to provide extremely high data rates (multi-Gbps)
for mobile devices, which can provide overwhelming network capacity. However, mmWave band communication has several
major propagation characteristic which are different from the microwave band, and leads to challenges in terms of interference
management.
In D2D enabled in mmWave 5G cellular networks different interference scenarios are introduced within each cell. Most previous
works aim to manage interference by optimizing resource sharing algorithms. Also, based on the above survey recent studies on
interference management in D2D communication consider the mutual interference of omnidirectional antennas. But, new
 22

interference management schemes should be proposed accounting for directional interference in mmWave 5G cellular networks
to enable multiple D2D communication.

6.2. Cell Densification and Offloading

A simple but prominent instrument to increase the network capacity in 5G cellular networks is network densification [113]. This
approach to enhancing network capacity has been demonstrated over several cellular generations. In general, the idea of network
densification for performance enhancement suggest the deployment of small coverage cells e.g., picocels (range under 100
meters) and femtocells (WiFi like range) into a close proximity of an intended terminal/devices resulting in more favorable
channel conditions between transmitters and receivers. Consequently, a reduced amount of transmission powers can be used
leading to a reduced interference towards other co-existing network elements. However, Interference management is one of the
issues when integrating D2D and small cells technology in underlay inband D2D communication. In this multi-tier
heterogeneous network, interference management and resource allocation issues for underlay spectrum sharing is more
challenging compared to the existing ones proposed in the literature for traditional single-tier systems, because the transmit
power of each BS is different. In addition, interference levels in different tiers are different due to the various access restrictions
(i.e., public, private, hybrid etc.). Besides, to improve spectrum efficiency, the interference among macro-cell cellular links,
small-cell cellular links, and D2D links all should be considered and efficiently managed. Moreover, the mode selection
algorithms need to be adapted to this heterogeneous environment to make a dynamic decision based on the network condition.
Thus, it is essential to consider how to obtain effective interference management.

7. CONCLUSIONS

D2D communication can provide many benefits to cellular networks such as the reuse gain, hop gain and proximity gain.
Therefore, D2D communication could be viewed as a promising paradigm for next generation 5G networks. This article has
provided a comprehensive survey on the existing interference management techniques that can be used to mitigate interference in
cellular networks with D2D communication. In particular, we first classified the types of interference scenarios in a two-tier
cellular network. We classified the existing interference management techniques based on the underlaying algorithm and
presented the state-of-the-art research works on D2D communication according to this classification. These researches
demonstrate the potential benefit of D2D for future 5G networks. However, existing
Table 3: A qualitative summary of the related works using advanced antenna and receiver techniques for interference management
Interference Ref Technique Target Solution Reuse Network Complexity Solution
Scenario Direction Type
DUE-DUE [107] IA Bit Error rate, -- - SC Centralized
sum rate
[108] IA Improve sum -- - SC Centralized
rate
DUE-CUE [101] BF Maximize Linear minimum mean- UL SC, 1 CUE, 1 DUE Centralized
throughput square error
[102] Superposition Improve Cooperative relaying scheme NA SC, 1CUE, 1 DUE Centralized
coding throughput
CUE-DUE [86] RRA, MIMO Enhance Interference Limited Area UL SC, N CUE, 1 DUE Centralized
capacity
[96] IC Improve Retransmission of UL SC, 1 CUE, 1 DUE Centralized
reliability interference signal
[99] IC Avoid near-far -- UL SC, N CUE, N DUE Distributed
interference
eNB-DUE [85] MIMO Improve cell Heuristic DL N CUE, N DUE Centralized
capacity
eNB-DUE, [87] MIMO Improve Signal processing DL 1 CUE, 1 DUE Centralized
DUE-CUE throughput
[100] BF and IC Improve -- DL SC, 1 CUE, 1 DUE Centralized
channel
capacity
[105] CoMP Improve Zero-forcing UL/DL MC, N CUE, 1 DUE Centralized
throughput
DUE-eNB, [88] Joint Power Network rate Non cooperative game theory UL SC, N CUE, N DUE Centralized
CUE-DUE control, Resource maximization and a self-optimizing
scheduling and with target rate algorithm
precoding constraints for
both D2D and
CUEs
[89] MIMO Mean square Physical Layer Network UL & DL SC, 1 DUE, 1 CUE, 1 Distributed
error coding Relay Node
[90] MIMO Maximize cell Heuristic UL SC, 1 DUE, N CUE Centralized
throughput
DUE-CUE, [18] SIC Successful Stochastic geometry UL MC, N DUE, N CUE Distributed
CUE-DUE transmission
probability
 23

DUE-DUE, [17] MIMO Increase Geometry, Poisson Point UL MS, N CUE, N DUE distributed
DUE-eNB, spectral Process
CUE-DUE efficiency
CUE-DUE, [91] MIMO Mean square Physical Layer Network UL & DL SC, 1 DUE, 1 CUE, 1 Centralized
eNB-DUE, error coding , Greedy algorithm, Relay Node & Distributed
DUE-eNB, [103] Rate Splitting Maximize Sum Han-Kobayashi based DL SC, 1 CUE, 1 DUE Centralized
DUE-CUE rate derivation

research studies also show new issues that must be solved to incorporate D2D communication in future 5G networks. The main
limitation of the existing studies is not considering interference management for D2D communication in potential 5G scenarios,
such as mmWave and cell densification.

References

[1] L. Qureshi, “Ericsson Mobility Report: On the pulse of the networked society,” no. June, 2015.

[2] GSA, “LTE Subscriptions forecast (worldwide) to 2020,” http://www.gsacom.com/news/gsa_434.php, 2015.

[3] K. Doppler, M. Rinne, C. Wijting, C. B. Ribeiro, and K. Hugl, “Device-to-Device Communication as an Underlay to
LTE-Advanced Networks,” Commun. Mag. IEEE, no. December, pp. 42–49, 2009.

[4] D. Feng, L. Lu, Y. W. Yi, G. Y. Li, S. Li, and G. Feng, “Device-to-device communications in cellular networks,” IEEE
Commun. Mag., vol. 52, no. April, pp. 49–55, 2014.

[5] G. Fodor, E. Dahlman, G. Mildh, S. Parkvall, N. Reider, G. Miklós, and Z. Turányi, “Design aspects of network assisted
device-to-device communications,” IEEE Commun. Mag., vol. 50, no. March, pp. 170–177, 2012.

[6] A. Asadi, Q. Wang, and V. Mancuso, “A Survey on Device-to-Device Communication in Cellular Networks,” IEEE
Commun. Surv. Tutorials, vol. PP, no. 4, pp. 1–1, 2014.

[7] J. Liu, N. Kato, J. Ma, and N. Kadowaki, “Device-to-Device Communication in LTE-Advanced Networks: A Survey,”
IEEE Commun. Surv. Tutorials, 2014.

[8] M. Tehrani, M. Uysal, and H. Yanikomeroglu, “Device-to-device communication in 5G cellular networks: Challenges,
solutions, and future directions,” IEEE Commun. Mag., vol. 52, no. May, pp. 86–92, 2014.

[9] P. Mach, Z. Becvar, and T. Vanek, “In-Band Device-to-Device Communication in OFDMA Cellular Networks: A
Survey and Challenges,” IEEE Commun. Surv. Tutorials, pp. 1–1, 2015.

[10] D. Astely, E. Dahlman, G. Fodor, S. Parkvall, and J. Sachs, “LTE release 12 and beyond [accepted from open call],”
Commun. Mag. IEEE, vol. 51, no. 7, pp. 154–160, 2013.

[11] J. Yue, C. Ma, H. Yu, and W. Zhou, “Secrecy-based access control for device-to-device communication underlaying
cellular networks,” IEEE Commun. Lett., vol. 17, no. 11, pp. 2068–2071, 2013.

[12] C. Ma, J. Liu, X. Tian, H. Yu, Y. Cui, and X. Wang, “Interference exploitation in D2D-enabled cellular networks: A
secrecy perspective,” IEEE Trans. Commun., vol. 63, no. 1, pp. 229–242, 2015.

[13] Q. Ye, M. Al-Shalash, C. Caramanis, and J. G. Andrews, “Resource Optimization in Device-to-Device Cellular Systems
Using Time-Frequency Hopping,” arXiv Prepr. arXiv …, vol. 13, no. 10, pp. 1–13, 2013.

[14] B. Kaufman, S. Member, J. Lilleberg, and S. Member, “Spectrum sharing scheme between Cellular Users and Ad-hoc
Device-to-Device Users,” IEEE Trans. Wirel. Commun., vol. 12, no. 3, pp. 1038–1049, 2013.

[15] K. G. Shin, L. Fellow, S. Gjessing, and a D. D. D. Technologies, “Connectivity of Cognitive Device-to-Device

Communications Underlying Cellular Networks,” IEEE J. Sel. Areas Commun., vol. 33, no. 1, pp. 81–99, 2015.
 24

[16] N. Lee, X. Lin, J. Andrews, and R. H. Jr, “Power Control for D2D Underlaid Cellular Networks: Modeling, Algorithms
and Analysis,” arXiv Prepr. arXiv1305.6161, vol. 33, no. 1, pp. 1–12, 2013.

[17] X. Lin, R. W. H. Jr, and J. G. Andrews, “The Interplay between Massive MIMO and Underlaid D2D Networking,” pp.
1–35, 2014.

[18] C. Ma, W. Wu, Y. Cui, and X. Wang, “On the Performance of Successive Interference Cancellation in D2D-enabled
Cellular Network,” pp. 1177–1182, 2015.

[19] J. Huang, Y. Yin, Y. Zhao, Q. Duan, W. Wang, and S. Yu, “A Game-Theoretic Resource Allocation Approach for
IntercellDevice-to-Device Communications in Cellular Networks,” IEEE Trans. Emerg. Top. Comput., vol. 6750, 2014.

[20] C. Xu, L. Song, Z. Han, Q. Zhao, X. Wang, X. Cheng, and B. Jiao, “Efficiency Resource Allocation for Device-to-
Device Underlay Communication Systems: A Reverse Iterative Combinatorial Auction Based Approach,” IEEE J. Sel.
Areas Commun., vol. 31, no. 9, pp. 348–358, 2013.

[21] F. Wang, L. Song, Z. Han, Q. Zhao, and X. Wang, “Joint Scheduling and Resource Allocation for Device-to-Device
Underlay Communication,” Wirel. Commun. Netw. Conf. (WCNC), 2013 IEEE, pp. 134–139, 2013.

[22] Q. Ye, M. Al-shalash, C. Caramanis, J. G. Andrews, and I. T. May, “Distributed Resource Allocation in Device-to-
Device Enhanced Cellular Networks,” Commun. IEEE Trans., vol. 63, no. 2, pp. 441–454, 2015.

[23] D. Wu, J. Wang, R. Q. Hu, Y. Cai, and L. Zhou, “Energy-efficient resource sharing for mobile device-to-device
multimedia communications,” IEEE Trans. Veh. Technol., vol. 63, no. 5, pp. 2093–2103, 2014.

[24] D. Tsolkas, E. Liotou, N. Passas, and L. Merakos, “A graph-coloring secondary resource allocation for D2D
communications in LTE networks,” 2012 IEEE 17th Int. Work. Comput. Aided Model. Des. Commun. Links Networks,
CAMAD 2012, pp. 56–60, 2012.

[25] C. Yang, Q. Cui, J. Han, and X. Tao, “Bipartite matching approach to optimal resource allocation in device to device
underlaying cellular network,” Electron. Lett., vol. 50, no. 3, pp. 212–214, 2014.

[26] R. Zhang, X. Cheng, L. Yang, and B. Jiao, “Interference-aware graph based resource sharing for device-to-device
communications underlaying cellular networks,” IEEE Wirel. Commun. Netw. Conf. WCNC, pp. 140–145, 2013.

[27] R. Zhang, X. Cheng, L. Yang, and B. Jiao, “Interference Graph-Based Resource Allocation ( InGRA ) for D2D
Communications Underlaying Cellular Networks,” vol. 64, no. 8, pp. 3844–3850, 2014.

[28] S. Sun and Y. Shin, “Resource Allocation for D2D Communication Using Particle Swarm Optimization in LTE
Networks,” 2014 Int. Conf. Inf. Commun. Technol. Converg., pp. 371–376, 2014.

[29] L. Su, Y. Ji, P. Wang, and F. Liu, “Resource Allocation Using Particle Swarm Optimization for D2D Communication
Underlay of Cellular Networks,” 2013 IEEE Wirel. Commun. Netw. Conf., pp. 129–133, 2013.

[30] P. Cheng, L. Deng, H. Yu, Y. Xu, and H. Wang, “Resource allocation for cognitive networks with D2D communication:
An evolutionary approach,” 2012 IEEE Wirel. Commun. Netw. Conf., pp. 2671–2676, 2012.

[31] L. Lei, Y. Kuang, X. Sherman Shen, C. Lin, and Z. Zhong, “Resource control in network assisted device-to-device
communications: Solutions and challenges,” IEEE Commun. Mag., vol. 52, no. June, pp. 108–117, 2014.

[32] E. S. Sousa and J. A. Silvester, “Optimum transmission ranges in a direct-sequence spread-spectrum multihop packet
radio network,” Sel. Areas Commun. IEEE J., vol. 8, no. 5, pp. 762–771, 1990.

[33] M. Haenggi, Stochastic geometry for wireless networks. Cambridge University Press, 2012.
 25

[34] H. Elsawy and E. Hossain, “Analytical Modeling of Mode Selection and Power Control for Underlay D2D
Communication in Cellular Networks,” Commun. IEEE Trans., vol. 62, no. 11, pp. 4147–4161, 2014.

[35] J. G. A. and A. G. Xingqin Lin, “Spectrum Sharing for Device-to-Device Communication in Cellular Networks,” vol. 13,
no. 12, pp. 6727–6740, 2014.

[36] H. Sun, M. Wildemeersch, M. Sheng, and T. Q. S. Quek, “D2D Enhanced Heterogeneous Cellular Networks With
Dynamic TDD,” IEEE Trans. Wirel. Commun., vol. 14, pp. 4204–4218, 2015.

[37] A. H. Sakr, E. Hossain, and N. I. May, “Cognitive and Energy Harvesting-Based D2D Communication in Cellular
Networks : Stochastic Geometry Modeling and Analysis,” arXiv, vol. 63, pp. 1–13, 2014.

[38] X. Lin, R. Ratasuk, A. Ghosh, and J. G. Andrews, “Modeling, analysis, and optimization of multicast device-to-device
transmissions,” IEEE Trans. Wirel. Commun., vol. 13, no. 8, pp. 4346–4359, 2014.

[39] S. Krishnan and H. S. Dhillon, “Distributed caching in device-todevice networks: A stochastic geometry perspective,”
Proc. Asilomar, Pacific Grove, CA, 2015.

[40] G. Giambene, “Queuing Theory And Telecommunications Networks And Applications, 2005 Springer Science+
Business Media.” Inc.

[41] B. Cho, K. Koufos, and J. Riku, “Spectrum allocation and mode selection for overlay D2D using carrier sensing
threshold,” Cogn. Radio Oriented Wirel. Networks Commun., pp. 26–31, 2014.

[42] X. Xiao, X. Tao, and J. Lu, “A QoS-aware power optimization scheme in OFDMA systems with integrated device-to-
device (D2D) communications,” IEEE Veh. Technol. Conf. 2011 IEEE, pp. 1–5, 2011.

[43] J. Gu, S. J. Bae, B.-G. Choi, and M. Y. Chung, “Dynamic power control mechanism for interference coordination of
device-to-device communication in cellular networks,” 2011 Third Int. Conf. Ubiquitous Futur. Networks, pp. 71–75,
2011.

[44] C. H. Yu, O. Tirkkonen, K. Doppler, and C. Ribeiro, “Power optimization of device-to-device communication
underlaying cellular communication,” IEEE Int. Conf. Commun., pp. 1–5, 2009.

[45] H. Xing and S. Hakola, “The investigation of power control schemes for a device - To-device communication integrated
into OFDMA cellular system,” IEEE Int. Symp. Pers. Indoor Mob. Radio Commun. PIMRC, pp. 1775–1780, 2010.

[46] S. Shalmashi, G. Miao, and S. Ben Slimane, “Interference management for Multiple Device-to-Device Communications
Underlaying Cellular Networks,” Pers. Indoor Mob. Radio Commun. (PIMRC), 2013 IEEE 24th Int. Symp. on. IEEE, pp.
223–227, 2013.

[47] G. Fodor, M. Johansson, P. Demia, B. Marco, and a Abrardo, “A joint power control and resource allocation algorithm
for D2D communications,” 2012.

[48] C. H. Yu, O. Tirkkonen, K. Doppler, and C. Ribeiro, “On the performance of device-to-device underlay communication
with simple power control,” IEEE Veh. Technol. Conf., pp. 4–8, 2009.

[49] J. Pekka, Y. Chia-Hao, R. Cassio, W. Carl, H. Klaus, T. Olav, and K. Visa, “Device-to-Device Communication
Underlaying Cellular Communications Systems,” Int’l J. Commun. Netw. Syst. Sci., vol. 2009, no. June, pp. 169–178,
2009.

[50] Z. Liu, T. Peng, S. Xiang, and W. Wang, “Mode selection for Device-to-Device (D2D) communication under LTE-
advanced networks,” IEEE Int. Conf. Commun., pp. 5563–5567, 2012.

[51] E. Frlan, “Direct communication wireless radio system.” Google Patents, 2000.
 26

[52] K. Doppler, C. H. Yu, C. B. Ribeiro, and P. Jänis, “Mode selection for device-to-device communication underlaying an
LTE-advanced network,” IEEE Wirel. Commun. Netw. Conf. WCNC, 2010.

[53] L. Lei, X. S. Shen, M. Dohler, C. Lin, S. Member, and Z. Zhong, “Queuing Models With Applications to Mode Selection
in Device-to-Device Communications Underlaying Cellular Networks,” IEEE Trans. Wirel. Commun., vol. 13, no. 12,
pp. 6697–6715, 2014.

[54] C. H. Yu, K. Doppler, C. B. Ribeiro, and O. Tirkkonen, “Resource sharing optimization for device-to-device
communication underlaying cellular networks,” IEEE Trans. Wirel. Commun., vol. 10, no. 8, pp. 2752–2763, 2011.

[55] N. Reider and G. Fodor, “A distributed power control and mode selection algorithm for D2D communications,”
EURASIP J. Wirel. Commun. Netw., vol. 2012, no. 1, pp. 1–25, 2012.

[56] M. Jung, K. Hwang, S. Choi, and I. I. S. Y. M. Odel, “Joint Mode Selection and Power Allocation Scheme for Power-
Efficient Device-to-Device (D2D) Communication,” Proc. IEEE VTC-Spring, pp. 1–5, 2012.

[57] D. Della Penda, L. Fu, and M. Johansson, “Energy efficient D2D communications in dynamic TDD systems,” arXiv
Prepr. arXiv1506.00412, pp. 1–31, 2015.

[58] H. S. Chae, J. Gu, B.-G. Choi, and M. Y. Chung, “Radio resource allocation scheme for device-to-device communication
in cellular networks using fractional frequency reuse,” 17th Asia Pacific Conf. Commun., vol. 1, no. October, pp. 58–62,
2011.

[59] S. Mumtaz, K. Mohammed, S. Huq, A. Radwan, J. Rodriguez, and R. L. Aguiar, “Energy Efficient Interference-Aware
Resource Allocation in LTE-D2D Communication,” 2014 IEEE Int. Conf. Commun., pp. 282–287, 2014.

[60] T. Bansal, K. Sundaresan, S. Rangarajan, and P. Sinha, “R2D2: Embracing device-to-device communication in next
generation cellular networks,” Proc. - IEEE INFOCOM, pp. 1563–1571, 2014.

[61] T. Chen, G. Charbit, and S. Hakola, “Time hopping for device-to-device communication in LTE cellular system,” IEEE
Wirel. Commun. Netw. Conf. WCNC, 2010.

[62] N. Chen, H. Tian, and Z. Wang, “Resource Allocation for Intra-Cluster D2D Communications Based on Kuhn-Munkres
Algorithm,” 2014 IEEE 80th Veh. Technol. Conf. (VTC Fall), pp. 1–5, 2014.

[63] J. Wang, E. Lo, and M. L. Yiu, “Identifying the most connected vertices in hidden bipartite graphs using group testing,”
IEEE Trans. Knowl. Data Eng., vol. 25, no. 10, pp. 2245–2256, 2013.

[64] and S. C. Yong Li, Chaoming Song, Depeng Jin, “a Dynamic Graph Optimization Framework for Multihop Device-To-
Device Communication Underlaying Cellular Networks,” no. October, pp. 52–61, 2014.

[65] P. Jänis, V. Koivunen, Ć. Ribeiro, J. Korhonen, K. Doppler, and K. Hugl, “Interference-aware resource allocation for
device-to-device radio underlaying cellular networks,” IEEE Veh. Technol. Conf., pp. 1–5, 2009.

[66] L. B. Le, “Fair resource allocation for device-to-device communications in wireless cellular networks,” 2012 IEEE Glob.
Commun. Conf., pp. 5451–5456, 2012.

[67] I. Kim, I.-S. Park, and Y. H. Lee, “Use of Linear Programming for Dynamic Subcarrier and Bit Allocation in Multiuser
{OFDM},” IEEE Trans. Veh. Technol., vol. 55, no. 4, pp. 1195–1207, 2006.

[68] A. H. Sakr, H. Tabassum, E. Hossain, and D. I. Kim, “Cognitive Spectrum Access in Device-to-Device-Enabled Cellular
Networks,” Commun. Mag. IEEE, vol. 53, no. 7, pp. 126–133, 2015.

[69] W. Zhibo, T. Hui, C. Nannan, and H. Yao, “Device-to-Device Resource Allocation for QoS Support Using a Graphic
Theory,” Consum. Commun. Netw. Conf., pp. 1003–1008, 2014.
 27

[70] R. Y. Chang, Z. Tao, J. Zhang, and C. C. J. Kuo, “Multicell OFDMA downlink resource allocation using a graphic
framework,” IEEE Trans. Veh. Technol., vol. 58, no. 7, pp. 3494–3507, 2009.

[71] C. Xu, L. Song, Z. Han, Q. Zhao, X. Wang, and B. Jiao, “Interference-aware resource allocation for device-to-device
communications as an underlay using sequential second price auction,” IEEE Int. Conf. Commun., pp. 445–449, 2012.

[72] C. Xu, L. Song, Z. Han, D. Li, and B. Jiao, “Resource allocation using a reverse iterative combinatorial auction for
device-to-device underlay cellular networks,” GLOBECOM - IEEE Glob. Telecommun. Conf., pp. 4542–4547, 2012.

[73] L. Wang, L. Liu, X. Cao, X. Tian, and Y. Cheng, “Sociality-aware resource allocation for device-to-device
communications in cellular networks,” IET Commun., vol. 9, pp. 342–349, 2015.

[74] X. Chen, X. Gong, L. Yang, and J. Zhang, “A social group utility maximization framework with applications in database
assisted spectrum access,” Proc. - IEEE INFOCOM, pp. 1959–1967, 2014.

[75] X. Gong, X. Chen, and J. Zhang, “Social Group Utility Maximization Game with Applications in Mobile Social
Networks.”

[76] Y. Li, D. Jin, J. Yuan, and Z. Han, “Coalitional Games for Resource Allocation in the device-to-device uplink
underlaying cellular networks,” Wirel. Commun. IEEE Trans., vol. 13, no. 7, pp. 3965–3977, 2014.

[77] R. Zhang, L. Song, Z. Han, X. Cheng, and B. Jiao, “Distributed Resource Allocation for Device-to-Device
Communications Underlaying Cellular Networks,” Commun. (ICC), 2013 IEEE Int. Conf., pp. 1889–1893, 2013.

[78] G. Yu, S. Member, L. Xu, D. Feng, and R. Yin, “Joint Mode Selection and Resource Allocation for Device-to-Device
Communications,” IEEE Trans. Commun., vol. 62, no. 11, pp. 3814–3824, 2014.

[79] S. Wen, X. Zhu, Z. Lin, X. Zhang, and D. Yang, “Optimization of interference coordination schemes in Device-to-
Device(D2D) communication,” 2012 7th Int. ICST Conf. Commun. Netw. China, CHINACOM 2012 - Proc., pp. 542–
547, 2012.

[80] C. Chien, Y. Chen, and H. Hsieh, “Exploiting Spatial Reuse Gain through Joint Mode Selection and Resource Allocation
for Underlay Device-to-Device Communications,” Int. Symp. Wirel. Pers. Multimed. Commun. (WPMC), 2012, pp. 80–
84, 2012.

[81] X. Wu, S. Tavildar, S. Shakkottai, T. Richardson, J. Li, R. Laroia, and A. Jovicic, “FlashLinQ: A synchronous
distributed scheduler for peer-to-peer ad hoc networks,” IEEE/ACM Trans. Netw., vol. 21, pp. 1215–1228, 2013.

[82] G. Fodor and N. Reider, “A distributed power control scheme for cellular network assisted D2D communications,”
GLOBECOM - IEEE Glob. Telecommun. Conf., pp. 1–6, 2011.

[83] D. Tse and P. Viswanath, Fundamentals of wireless communication. Cambridge university press, 2005.

[84] W. Zhong, Y. Xu, and H. Tianfield, “Game-theoretic opportunistic spectrum sharing strategy selection for cognitive
MIMO multiple access channels,” IEEE Trans. Signal Process., vol. 59, no. 6, pp. 2745–2759, 2011.

[85] P. Jänis, V. Koivunen, C. B. Ribeiro, K. Doppler, and K. Hugl, “Interference-avoiding MIMO schemes for device-to-
device radio underlaying cellular networks,” IEEE Int. Symp. Pers. Indoor Mob. Radio Commun. PIMRC, pp. 2385–
2389, 2009.

[86] H. Min, J. Lee, S. Park, and D. Hong, “Capacity enhancement using an interference limited area for device-to-device
uplink underlaying cellular networks,” IEEE Trans. Wirel. Commun., vol. 10, no. 12, pp. 3995–4000, 2011.

[87] H. Tang, C. Zhu, and Z. Ding, “Cooperative MIMO Precoding for D2D Underlay in Cellular Networks,” in IEEE
International Conference on Communications, 2013, pp. 5517–5521.
 28

[88] W. Zhong, Y. Fang, S. Jin, K. Wong, and S. Member, “Joint Resource Allocation for Device-to-Device Communications
Underlaying Uplink MIMO Cellular Networks,” IEEE J. Sel. Areas Commun., vol. 33, no. 1, pp. 41–54, 2015.

[89] L. K. S. Jayasinghe, P. Jayasinghe, N. Rajatheva, and M. Latva-Aho, “MIMO physical layer network coding based
underlay device-to-device communication,” IEEE Int. Symp. Pers. Indoor Mob. Radio Commun. PIMRC, pp. 89–94,
2013.

[90] J. C. F. Li, M. Lei, and F. Gao, “Device-to-device (D2D) communication in MU-MIMO cellular networks,”
GLOBECOM - IEEE Glob. Telecommun. Conf., pp. 3583–3587, 2012.

[91] K. Jayasinghe, S. Member, P. Jayasinghe, and S. Member, “Linear Precoder-Decoder Design of MIMO Device-to-
Device Communication Underlaying Cellular Communication,” IEEE Trans. Commun., vol. 62, no. 12, pp. 4304–4319,
2014.

[92] J. C. Liberti and T. S. Rappaport, Smart antennas for wireless communications: IS-95 and third generation CDMA
applications. Prentice Hall PTR, 1999.

[93] Q. H. Spencer and A. L. Swindlehurst, “A hybrid approach to spatial multiplexing in multiuser MIMO downlinks,”
EURASIP J. Wirel. Commun. Netw., vol. 2004, no. 2, pp. 236–247, 2004.

[94] Q. H. Spencer, a. L. Swindlehurst, and M. Haardt, “Zero-forcing methods for downlink spatial multiplexing in multiuser
MIMO channels,” IEEE Trans. Signal Process., vol. 52, no. 2, pp. 461–471, 2004.

[95] J. G. Andrews, “Interference Cancellation for Cellular Systems : a Contemporary Overview,” IEEE Wirel. Commun., no.
April, pp. 19–29, 2005.

[96] H. Min, W. Seo, J. Lee, S. Park, and D. Hong, “Reliability improvement using receive mode selection in the device-to-
device uplink period underlaying cellular networks,” IEEE Trans. Wirel. Commun., vol. 10, no. 2, pp. 413–418, 2011.

[97] V. S. Annapureddy and V. V. Veeravalli, “Gaussian interference networks: Sum capacity in the low-interference regime
and new outer bounds on the capacity region,” IEEE Trans. Inf. Theory, vol. 55, no. 7, pp. 3032–3050, 2009.

[98] S. T. Chung and J. M. Cioffi, “The capacity region of frequency-selective Gaussian interference channels under strong
interference,” IEEE Trans. Commun., vol. 55, no. 9, pp. 1812–1821, 2007.

[99] S. Xu, H. Wang, T. Chen, Q. Huang, and T. Peng, “Effective interference cancellation scheme for device-to-device
communication underlaying cellular networks,” IEEE Veh. Technol. Conf., pp. 3–7, 2010.

[100] W. Xu, L. Liang, H. Zhang, S. Jin, J. C. F. Li, and M. Lei, “Performance enhanced transmission in device-to-device
communications: Beamforming or interference cancellation?,” in GLOBECOM - IEEE Global Telecommunications
Conference, 2012, pp. 4296–4301.

[101] W. Fu, R. Yao, F. Gao, J. C. F. Li, and M. Lei, “Robust null-space based interference avoiding scheme for D2D
communication underlaying cellular networks,” IEEE Wirel. Commun. Netw. Conf. WCNC, pp. 4158–4162, 2013.

[102] C. Ma, G. Sun, X. Tian, K. Ying, H. Yu, and X. Wang, “Cooperative relaying schemes for device-to-device
communication underlaying cellular networks,” GLOBECOM - IEEE Glob. Telecommun. Conf., pp. 3890–3895, 2013.

[103] C. H. Yu and O. Tirkkonen, “Device-to-device underlay cellular network based on rate splitting,” IEEE Wirel. Commun.
Netw. Conf. WCNC, pp. 262–266, 2012.

[104] K. Kobayashi, “A new achievable rate region for the interference channel,” IEEE Trans. Inf. Theory, vol. 27, no. 1, pp.
49–60, 1981.

[105] S. Mumtaz, K. Mohammed, S. Huq, and J. Rodriguez, “Coordinated Paradigm for D2D Communications,” Comput.
Commun. Work. (INFOCOM WKSHPS), 2014 IEEE Conf., pp. 718–723, 2014.
 29

[106] V. R. Cadambe and S. A. Jafar, “Interference alignment and degrees of freedom of the K-user interference channel,”
IEEE Trans. Inf. Theory, vol. 54, no. 8, pp. 3425–3441, 2008.

[107] H. E. Elkotby, K. M. F. Elsayed, and M. H. Ismail, “Exploiting interference alignment for sum rate enhancement in
D2D-enabled cellular networks,” in IEEE Wireless Communications and Networking Conference, WCNC, 2012, pp.
1624–1629.

[108] H. E. Elkotby, K. M. F. Elsayed, and M. H. Ismail, “Shrinking the reuse distance: Spectrally-efficient radio resource
management in D2D-enabled cellular networks with Interference Alignment,” IFIP Wirel. Days, pp. 1–6, 2012.

[109] L. Yang, W. Zhang, and S. Jin, “Interference Alignment in Device-to-Device LAN Underlaying Cellular Networks,”
IEEE Trans. Wirel. Commun., vol. 1276, pp. 1–1, 2015.

[110] T. S. Rappaport, R. Mayzus, Y. Azar, K. Wang, G. N. Wong, J. K. Schulz, M. Samimi, and F. Gutierrez, “Millimeter
Wave Mobile Communications for 5G Cellular: It Will Work!,” IEEE Access, vol. 1, pp. 335–349, 2013.

[111] A. Ghosh, T. Thomas, M. Cudak, and R. Ratasuk, “Millimeter wave enhanced local area systems: A high data rate
approach for future wireless networks,” IEEE J. Sel. Areas Commun., vol. 32, no. 6, pp. 1152–1163, 2014.

[112] Y. Niu, C. Gao, Y. Li, L. Su, D. Jin, and A. V. Vasilakos, “Exploiting Device-to-Device Communications in Joint
Scheduling of Access and Backhaul for mmWave Small Cells,” IEEE J. Sel. Areas Commun., vol. 33, pp. 2052–2069,
2015.

[113] E. Hossain, M. Hasan, and S. Member, “5G Cellular : Key Enabling Technologies and Research Challenges,” Instrum.
Meas. Mag. IEEE, vol. 18, no. 3, pp. 11–21, 2015.