SPECIAL SECTION ON ARTIFICIAL INTELLIGENCE ENABLED NETWORKING

Received July 2, 2015, accepted July 17, 2015, date of publication August 11, 2015, date of current version August 28, 2015.
Digital Object Identifier 10.1109/ACCESS.2015.2467174

Artificial Intelligence-Based Techniques for

Emerging Heterogeneous Network: State
of the Arts, Opportunities,
and Challenges
XIAOFEI WANG, (Member, IEEE), XIUHUA LI, (Student Member, IEEE),
AND VICTOR C. M. LEUNG, (Fellow, IEEE)
Department of Electrical and Computer Engineering, The University of British Columbia, Vancouver, BC V6T 1Z4, Canada
Corresponding author: X. Wang ([email protected])
This work was supported in part by the Canadian Natural Sciences and Engineering Research Council under Grant RGPIN-2014-06119
and Grant RGPAS-462031-2014, in part by the Chinese Scholarship Council through the Four-Year Post-Graduate Scholarship, and
in part by the National Natural Science Foundation of China under Project 61271182.

ABSTRACT Recently, mobile networking systems have been designed with more complexity of
infrastructure and higher diversity of associated devices and resources, as well as more dynamical formations
of networks, due to the fast development of current Internet and mobile communication industry. In such
emerging mobile heterogeneous networks (HetNets), there are a large number of technical challenges
focusing on the efficient organization, management, maintenance, and optimization, over the complicated
system resources. In particular, HetNets have attracted great interest from academia and industry in deploying
more effective solutions based on artificial intelligence (AI) techniques, e.g., machine learning, bio-inspired
algorithms, fuzzy neural network, and so on, because AI techniques can naturally handle the problems of
large-scale complex systems, such as HetNets towards more intelligent and automatic-evolving ones. In this
paper, we discuss the state-of-the-art AI-based techniques for evolving the smarter HetNets infrastructure
and systems, focusing on the research issues of self-configuration, self-healing, and self-optimization,
respectively. A detailed taxonomy of the related AI-based techniques of HetNets is also shown by discussing
the pros and cons for various AI-based techniques for different problems in HetNets. Opening research issues
and pending challenges are concluded as well, which can provide guidelines for future research work.

INDEX TERMS Artificial intelligence, genetic algorithms, ant colony optimization, self-organization
networks, heterogeneous networks.

I. INTRODUCTION Worldwide Interoperability for Microwave Access (WiMAX),

Due to the fast development of current Internet and Mobile and enhanced NodeBs (eNBs) in Long-Term Evolu-
Communication industry, the mobile traffic load has tion (LTE), which is with the largest transmission power
experienced an explosive growth during the last few years and thus coverage (up to several kilometres outdoor),
as well. Accordingly, the systems of mobile network oper- picocell which has relatively less coverage mainly targeting
ators (MNOs) have been designed and improved with more indoor coverage enrollment, femtocell which has relatively
complexity of infrastructure, higher diversity of associated small-size low-cost low-power user-deployed access points in
devices and resources, and more dynamical formations of offices and homes, relay which is an MNO-deployed access
networks [1], and they are evolving towards the promising point that only routes signals between the macro cell and end
future mobile networking paradigm, the heterogeneous users in poor coverage areas and dead spots. Therefore, the
networks (HetNets). complicated network infrastructure is facing to troubles for
In HetNets, there are various types of cells with different effectively organizing, managing and optimizing the
transmission power, coverage, and working mechanisms, network resources, while the skyrocketing mobile traffic
as shown in Fig. 1, e.g., macrocell, base station (BS) in demand is generating more pressure. Although there have

2169-3536
2015 IEEE. Translations and content mining are permitted for academic research only.
VOLUME 3, 2015 Personal use is also permitted, but republication/redistribution requires IEEE permission. 1379
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 X. Wang et al.: AI-Based Techniques for Emerging HetNet

been many proposed techniques for directly improving we will not penetrate into technical details of each algorithms
HetNets performance, how to automatically deal with the and schemes, but will try to introduce and summarize up-to-
HetNets’ complexity via other evolutionary techniques and date research issues in related areas with sufficient breadth
algorithms becomes a hot research topic. and proper depth.
‘‘Brains exist because the distribution of resources The remainder of this article is organized as following:
necessary for survival and the hazards that threaten survival We first introduce why intelligence is demanded for evolving
vary in space and time [2]. The similar concept of bringing the emerging HetNets in Sec. II. Then we describe issues of
intelligence has been greatly expected in research areas on HetNets that can benefit from AI-based techniques in Sec. III.
HetNets, which are increasingly dynamic, heterogeneous, The taxonomy of the AI-based techniques for supporting
large-scale, and complex. For a long time, researchers have HetNets is summarized in Sec. IV. New challenges and
been sparing no effort in deploying novel solutions by adopt- opportunities are discussed in Sec. V, and Sec. VI concludes
ing Artificial Intelligence (AI) techniques for the automatic the paper.
management and optimization of HetNets.
AI techniques include multi-disciplinary techniques from II. THE EMERGING HetNets
machine learning, bio-inspired algorithms, fuzzy neural The emerging mobile networks are becoming increasingly
network and so on, and they have been extensively studied dynamic, heterogeneous, large-scale, and systematically
and applied to optimize computer systems and networks complex, and hence the infrastructure consists of various
in diverse scenarios and complicated environments. It has wireless access technologies with various capabilities,
been proved that AI techniques can achieve outstanding constraints, and operating functions towards the emerging
performance [3]–[5], as most of them are inspired from nature HetNet.
findings or motivated by the ways of thinking of human More specifically, in the multi-tier infrastructure of the
beings. They have relatively lower complexity enabled by newest release of LTE-Advanced [7], macro-cells, pico-cells,
recursive feedback-based learning and local interactions, and femto-cells, as well as relay stations and Remote Radio Heads
hence faster speed of finding sub-optimal solutions compared (RRHs), are envisaged. Macro-cell base stations use dedi-
to conventional techniques [3], [4]. Therefore, the integration cated backhaul, for providing guaranteed bandwidth. Pico-
of AI techniques into the design of smart infrastructure cells and femto-cells are with significant popularity recently,
becomes one promising trend for effectively solving and are small-size low-cost low-power user-deployed access
problems in HetNets. e.g., cell planning, fault-tolerance, points in buildings and houses, offloading data traffic locally.
resource management and optimization etc. Relays are just MNO-deployed access points routing only
From the aspect of reducing the operational and capital signals from the macro cell to end users and vice versa in
expenditure (O/CAPEX) of MNOs, AI-based techniques can poor coverage areas and dead spots. RRHs are compact-size,
substantially diminish human involvement in operational high-power, and low-weight units, mounted outside the con-
tasks, and optimize network capacity, coverage, and Quality ventional macro BS and connected to it generally through a
of Service (QoS) in HetNets, according to the advanced fiber optic cable for creating a distributed BS. The associated
features discussed in Self-Organizing Networks (SONs) centre macro BS is in charge of control and baseband signal
along following directions [3], [4], [6]: 1) self-configuration, processing, and some radio circuitry is moved into RRHs.
where newly deployed high variety of cells are automati- Details of each types times of cells and stations are discussed
cally configured and updated before entering operation tasks, and shown in Fig. 1 as well.
2) self-healing, where cells and networks can automatically Deploying various types of cells aims at offloading the
detect and recover from failure and even execute traffic load via different requirements on coverage, locations,
compensation mechanisms whenever failures occur, and environment characteristics and user dynamics, for boosting
3) self-optimization, where cellular systems can measure the the spectrum re-usage effectively. Therefore, HetNets can
network conditions and optimize the settings to improve the drastically reduce the O/CAPEX of MNOs. However, the
performance in terms of coverage, bandwidth, interference complexity of HetNets brings substantial design challenges
avoidance, and QoS, while the corresponding issues of and problems, where the integration of AI-based and the
system scalability and energy conservation are also framework of ‘‘Self-Organization Networks’’ (SON) with the
considered. HetNets can be one effective way towards smart future mobile
In this article, we survey the state-of-the-art AI-based networks.
techniques for evolving the smarter HetNets infrastructure
and systems. Discussions on how researchers have been III. ‘‘SELF-EVOLUTION’’ IN HetNets
trying to adopt Machine Learning, Genetic Algorithms, During the last few years, the technology of SONs has
Ant Colony Optimization Artificial Neural Networks, Fuzzy experienced an explosive growth in its study. SONs are to
System, Markov Models and Bayesian Models for improving effectuate substantial reductions of operation cost by dimin-
the smartness, efficiency, performance and QoS of HetNets ishing human involvement. The essential idea of SONs is to
will be carried out, followed by a comprehensive taxonomy integrate network planning, configuration, and optimization
of AI-based techniques used for evolving HetNets. Note that into a single and mostly automated process requiring minimal

1380 VOLUME 3, 2015

X. Wang et al.: AI-Based Techniques for Emerging HetNet

FIGURE 1. Infrastructure of heterogeneous networks.

FIGURE 2. Illustration of AI-based techniques for self-organization on HetNets.

manual intervention; particularly AI-based techniques may introduce related AI-techniques for SONs in HetNets with
offer efficient solutions for SONs in HetNets. Main regards to the aforementioned three features. And features of
features in SONs in HetNets include self-configuration, SONs for HetNets are illustrated in Fig. 2.
self-optimization, and self-healing, which designs HetNets Note there are already several studies focusing on
with the ability of ‘‘self-evolution’’ [5], [8]. The abilities of proposals of the three main features of SONs comprehen-
‘‘self-planning’’ and ‘‘self-organisation’’ are sometime sively. The study in [8] has proposed SONs for Radio Access
discussed separately from self-configuration, but in this Networks (RANs) in HetNets with centralized annealing
article we consider them as one essential automatic pre- approach for coverage and capacity optimization, a hybrid
operational starting phase. In following subsections, we will neighbourhood based approach for cell outage detection, and

VOLUME 3, 2015 1381

 X. Wang et al.: AI-Based Techniques for Emerging HetNet

a centralized greedy approach for cell outage compensation, E. coli bacteria [14]. Since the convergence time of
The study in [3] also surveys different aspects of swarm BioRARSA is insensitive to the initial sampling step-size
intelligence inspired mechanisms as well as their designing of the algorithm, it exhibits a robustness against all initial
principles, and examines various optimization algorithms that parameters and the dynamic nature of distributed HetNets.
have been applied to artificial SON systems for HetNets. And
we will detail most up-to-date related literature for improving
B. SELF-HEALING
the self-‘‘X’’ ability of HetNets in following sections.
Robustness is one of the most critical issues for commer-
A. SELF-CONFIGURATION
cial HetNets, MNOs need to stably support a large number
of various devices and diverse users. SON functionality in
In the planning phase of HetNets, the locations of base
HetNets provides the essential self-healing ability based on
stations (or other types of cell stations), links between
active tracking, passive monitoring, proper measurement for
base stations, and associated various network devices as
faults, and appropriate consequent remedial actions, such
well as the corresponding parameters (output power, antenna
as restarting, falling back to backup system, switching to
angle, etc.) should be determined before all equipments are
compensation equipment and so on. SONs also offer some
practically installed. However, because of the coexistence of
means for automatically upgrading the software of network
multiple types of cells in the HetNets and high dynamics of
equipments to fix legacy bugs and problems, as well as adjust-
users and services, an ever increasing number of parameters
ing parameters adaptively to avoid problematic situations.
need to be managed and optimized. A better planning can
Self-healing saves huge amount of human intervention for
shape the cell coverage optimally and prevent severe prop-
networking system maintenance, and recovers the network
agation losses at the cell edge. Therefore, it is important to
failures as much autonomously as possible.
derive optimal parameter settings automatically so that the
Therefore, AI techniques can be adopted conveniently for
amount of human-beings’ labor is minimized in HetNets [9].
the self-healing feature to ensure services and resources avail-
AI-based techniques (e.g., Genetic Algorithm) can be
ability. As one representative work of cell outage compen-
easily utilized to evolve cell planning and coverage opti-
sation, the proposal in [15] proposes an autonomic particle
mization with pilot power adjustment [10]. And the most
swarm compensation algorithm to generate a centralized cell
initialisation phase is the automatic physical cell identifier
outage compensation management scheme in LTE HetNets,
assignment and radio resource configuration in
in order to mitigate the performance degradation induced by
HetNets [9]. Ant then in the configuration phase after the
the cell outage.
planning and placement phase, newly deployed cell base
Autonomous re-configuration is considered as one key
stations (and related hardware) should be able to get automati-
feature of self-healing of SON in HetNets. By observing
cally configured, throughly tested, and autonomously authen-
changes of users and network conditions, HetNets requires
ticated by downloading and running firmware/software
necessary re-configurations in a real-time manner without
before entering the operational state, which is called ‘‘plug-
termination of mobile services. This has been studied in
and-play’’ mode. There have been a certain number of
[16], which has proposed an adaptive policy-based dynamic
studies focusing on the self-configuration in HetNets, mostly
reconfiguration framework, by creating and updating poli-
on the methodology for deriving appropriate parameters for
cies dynamically in response to changing reconfiguration
specific HetNets scenarios. For example, the study in [11]
requirements. Through the use of Reinforcement Learning
addresses the issues of smart low-power node deployment
methodology, already-deployed policies are updated dynami-
in 5G HetNets, and proposes to associate appropriate sec-
cally within the two-layer policy based framework. However,
torization with radio resource allocation during the adaptive
adaptability of the network control mechanisms and on-the-
SON by integrating cognitive radio with inter-cell
fly derivation of new policy rules is not taken into considera-
interference coordination. Also relay placement requires
tion.
sophisticated modeling and configurations as researched
in [12] for determining the parameters of interfere-limited
relay channel management to maximize capacity without C. SELF-OPTIMIZATION
committing to protracted system simulation studies. Much emphasis in this article is put to the self-optimization
Some other studies focus on the distributed beamforming techniques in HetNets. Due to the running of HetNets, the per-
configuration in HetNets to achieve some breakthrough for formance should keep being improved from a large number of
optimal coverage and signal quality. Compared with aspects, such as load balancing, power adaptation, neighbour-
conventional beamforming techniques that require priori hood maintenance, mobility management and so on, taking
knowledge of channel conditions at transmitters, into account radio characteristics, traffic dynamics, and user
the bio-inspired robust adaptive random search algorithm demands in the service area. However, practically applying
(BioRARSA) [13] is proposed to enable a convergence time real-time optimization is difficult, because it typically entails
that scales linearly with the number of distributed transmit- heavy work load for comprehensive measurement, statistics
ters, as inspired by a heuristic random search mechanism that learning, optimization problem solving, and decision making
mimics the foraging strategy and life cycles of over the parameters.

1382 VOLUME 3, 2015

X. Wang et al.: AI-Based Techniques for Emerging HetNet

a) Automatic coverage optimization and load balancing optimize the relaying topology based on the awareness of
can be achieved by adjusting the antenna settings and thus the intercell interference and the spatial traffic distribution
shaping the radio coverage, and by adjusting the handover dynamics, where the topologies are encoded as a set of
parameters to logically change the cell size. Automatic chromosomes, and special crossover and mutation operations
coverage optimization also aims to maximize the system are proposed to search for the optimum. The aforementioned
capacity and ensure an appropriate overlapping area study in [12] presents closed-form capacity expressions for
between neighbor cells based on cooperatively adjusting interfere-limited relays as well as the deployment parameters
antenna tilt and pilot power among the related cells. For that may maximize the capacity practically.
instance, the study in [17] proposes self-optimization of f) Virtical Handoff (VHO) in HetNets [22] plays an
antenna tilt and power by using a fuzzy neural network important role in fulfilling seamless mobile service when
optimization method based on cooperative reinforcement users cross different cells with different link layer technolo-
learning in HetNets. gies for RANs. Current VHO algorithms mainly focus on
b) Mobility optimization is to eliminate unnecessary when to trigger and what connection QoS to improve, but
handover and to provide appropriate handover timing, neglect the synthetical consideration of all currently avail-
by automatically adjusting the thresholds related to cell able candidate networks, where AI-based techniques can help
re-selection and handover. And its target is also to load bal- to get optimal decisions on parameters by overall evalua-
ancing optimization, which can automatically force special tion of the complicated conditions. For instance, the study
users covered by the edge of a congested cell to reselect or in [23] addresses the VHO problem for user QoS enhance-
handover to some neighbor cells which are less congested ment and system performance improvement in HetNets, and
by adjusting the parameter thresholds. Mobility optimization proposes an adaptive parameter adjustment algorithm based
highly relies on the support of efficient neighbourhood main- on neural network model. Similar studies can be discovered
tenance, the users can automatically refresh and reconfigures in [24] and [25].
the neighbourhood list for holding the minimum set of cells Note that practical design of SON features into HetNets
necessary for roaming based on measurements. systems and networks requires necessary deployment of
c) Link quality estimation must be always performed monitoring and maintenance systems into core networks
with a high reliability to facilitate a secure transmission with mostly. The study in [17] designs a dedicated SON server
robustness in HetNets. Rather than conventional static installed in HetNets with communication functions for
link-quality aware routing metrics that adopt simple moving- distributed SON modules at each network devices and cells,
average estimators, bio-inspired estimator based on the neural integrated with database for storing collected information.
network paradigm can be utilized to improve the effectiveness The server periodically executes optimization algorithms on
of link-quality estimation [18]. the basis of the analysis results, and SON features can be
d) Device-to-Device communications supply HetNets realized without having a large impact on existing systems.
with higher scalability and large capacity, by utilizing the
organized device-to-device links owing to the effective inter- IV. TAXONOMY OF AI-RELATED SON TECHNIQUES
ference management. The desired resource allocation can IN HetNets
benefit much from the AI-based techniques. For instance, In this section, we survey and discuss the AI-related
the study in [19] proposes Ant Colony Optimization base SON techniques in HetNets by classifying them based on the
intelligent management, by which D2D pairs can effectively, types of AI techniques, while discussing their correlation and
autonomously, and jointly select the suitable resource blocks benefits with SONs in HetNets in detail. Note that all
and adjust power to guarantee signal quality of all users. discussed studies are clearly listed in Table 1.
e) Relay-based multi-hop transmission can effectively
extend the service coverage and strengthen the sustainabil- A. MACHINE LEARNING
ity of HetNets by leveraging intermediate relay nodes, so Machine learning (ML) is evolved from the study of
that the transmission can cross over multiple hops, while pattern recognition and computational learning theory in
the link interference and multi-hop path formation become AI areas [26]. It effectively learns the way of human brains,
quite challenging. One study in [20] presents a traffic-light- explores the construction and study of algorithms, and makes
related approach to autonomous self-optimization of tradeoff data-driven predictions or decisions [6]. Among many ML
performance indicators (i.e., coverage and capacity) in LTE techniques, the reinforcement learning (RL) is one particular
HetNets, where by introducing a low-complexity interference learning algorithm inspired by behaviorist psychology,
approximation model, the related optimization problem is concerned with how software agents ought to take actions in
formulated as a mixed-integer linear program and is an environment so as to maximize some notion of cumulative
embedded into a self-organized network operation and rewards [27]. From the illustration in Fig. 3, RL aims at
optimization framework. The hierarchical cooperative finding a policy that maximizes the observed rewards (i.e.,
relay-based HetNets are discussed in [21] for providing payoffs) over a certain time period, and it is often used for
cost-effective coverage extension based on the convergence channel access and transmit power allocation in network-
of heterogeneous radio networks. GA can be also used to ing systems. One representative work is the study in [28]

VOLUME 3, 2015 1383

 X. Wang et al.: AI-Based Techniques for Emerging HetNet

TABLE 1. Taxnomony of AI-based techniques for smart infrastructure of HetNets.

which presents game dynamics and performance analysis of improve spectral efficiency and energy efficiency of antenna
RL schemes for 4G HetNets where users try to learn their tilting in HetNets in [35], where BSs do not need knowledge
own optimal payoff and their optimal strategy simultaneously of location information. Also the study in [36] uses a
with limited information, and shows the advantages when regret-learning based algorithm to improve the interference
dealing with imperfect, noisy and delayed measurement and coordination in closed-loop spatial multiplexing Multi-Input-
randomly changing environment. Multi-Output HetNets, where cells jointly estimate own
utility and learn strategies locally in a decentralized manner.
Note that optimal solution cannot be always obtained by
distributed methods, but near-optimal one can be achieved
rapidly.
Fuzzy-reinforcement learning is one further step of
RL techniques by integrating fuzzy state representations and
fuzzy goals for uncertain environments in HetNets. There-
fore, its target mostly falls into the optimization on dynamical
resource management with SON-enabled fractional power
control in HetNets [37]. The Fuzzy Q-learning can also be
decentralized for HetNets, as studied in [38], which focuses
on cross-layer interference mitigation in an OFDMA based
FIGURE 3. Illustration of learning-based optimization in HetNets. HetNet deployment [39], [40]. By the comprehensive
comparison among decentralized Q-learning, fuzzy
Q-learning is a form of RL, which is a model-free Q-learning, improved Q-learning and expertness-based
RL technique to find an optimal Quality Value (Q-Value) of Q-learning methods, the study in [41] delves into the
action-selection policy for any given (finite) Markov decision challenges of interference management solutions.
process [29]. And thus it can be utilized for most of the self-
optimization objectives in HetNets, e.g., Q-learning based
dynamical resource allocation with positioning information B. GENETIC ALGORITHMS
during carrier aggregation in semi-and uncoordinated deploy- Among many popular evolutionary algorithms discovered
ment of HetNets in [30] and [31]. Also RL can facilitate and designed for optimizing complex systems, the Genetic
the coordination-based context-aware mobility manage- Algorithm (GA) must be the representative one [42], [43].
ment [32], as well as the autonomous and adaptive The GA adopts the gene evolution procedure, including
reconfiguration management in HetNets efficiently, as two main tasks, crossover, which facilitates the obtainment
researched in [16]. of optimal solution, and mutation, which prevents falling to
Due to the substantial amount of devices with various types regional optimal solutions. The offspring inherits the gene of
in HetNets, interference management strategies in HetNets superior chromosome in crossover operation, and eliminates
can benefit from distributed Q-learning as discussed in [33], poor chromosomes through competition. Fig. 4 illustrates the
where each cell reinforces the learning collaboratively with procedure using GA for HetNets.
others, in order to achieve time-domain-adaptive enhanced Proved by years of studies, GA is efficient for solving
inter-cell interference coordination. Similar interference opti- problems where the potential solution scope is too high to be
mization study with distributed self-organizing functions is searched exhaustively in reasonable time, and it can appropri-
also studied in [34]. Another similar approach is applied to ately converge to optimal (or suboptimal) results, in shorter

1384 VOLUME 3, 2015

X. Wang et al.: AI-Based Techniques for Emerging HetNet

sequential GA to optimize the relaying topology in multi-hop

HetNets with awareness of the intercell interference and the
spatial traffic distribution dynamics, where the topologies are
encoded as a set of chromosomes, and special crossover and
mutation operations are proposed to search for the optimum
topology.

C. SWARM INTELLIGENCE AND ANT COLONY

Ant colony optimization (ACO) is one popular algorithm of
swarm intelligence [49]. In the ant colony, a huge amount
of ants are discovering optimal routes to food based on pre-
viously accumulated but dismissing pheromones. Because of
the similarity between networking systems and ant colony (an
ant corresponding to a data packet, and a path corresponding
to the networking route), ACO is widely discussed in studies
on HetNets as an effective methodology for approaching
the optimal performance [3] with regard to interference
FIGURE 4. Illustration of GA optimization flow for HetNets. management, routing, and coverage [5], [50], as shown in
Fig. 5.
time in most cases compared with other bio-inspired meth- For mitigating inter-cell and cross-tier interference
ods [44]–[46]. Particularly, GA can solve multi-objective coordination, the study in [51] explores ant colony optimiza-
optimization problems easily. Therefore, GAs are popularly tion for dynamic sub-channel allocation by utilizing ACO
used to facilitate HetNets, mostly focusing on the issues of for channel allocation. The study in [52] employs quan-
cell planning and node placement optimization when a great tum particle swarm optimization (QPSO) to propose a sub-
number of parameters need to be evaluated. optimal energy-efficient user association scheme in HetNets
For example, the study in [47] proposes a multi-objective where cells can be switched off (cell sleeping) along with
GA to address a multi-objective communication nodes (e.g., the dynamic traffic load. ACO has automatic self-healing
antennas and relays) placement problem in HetNets, aiming capability, and thus Li et al. [15] proposed an autonomic par-
at maximizing the communication coverage as well as the ticle swarm compensation algorithm (APSCA) for cell outage
total capacity bandwidth, and minimizing the placement cost. compensation management scheme, in order to mitigate the
By adaptively adjusting the pilot power, GA can be also performance degradation induced by the cell outage.
used to evolve cell coverage optimization as studied in [10]. Network routing in HetNets can obtain huge gain of swarm
And the detailed research work in [48] presents a novel intelligence due to their inherent path optimization strategies

FIGURE 5. Illustration of ACO for HetNets.

VOLUME 3, 2015 1385

 X. Wang et al.: AI-Based Techniques for Emerging HetNet

FIGURE 6. Illustration of artificial neural networks for HetNets.

of entities (i.e., ants) [53]. Most of the routing issues in because of its rapid convergence. ANNs can be also utilized
HetNets take place in multi-hop transmissions where the for addressing the vertical handoff problem for user QoS
ACO algorithm can address a multi-objective routing opti- enhancement and HetNets system performance improvement
mization problem that uses network performance measures as research in [23], which proposes an adaptive parameter
such as delay, hop distance, load, cost and reliability [54]. adjustment algorithm based on neural network model by
The study in [55] uses ant colony algorithm to solve the which user parameters can be determined optimally.
problem of coverage optimization for dense deployment of
small cell in HetNets by finding the optimal pilot transmit E. FUZZY SYSTEM
power of each small cell through the minimization of the A fuzzy system is a control system based on fuzzy logic,
cost function. Also for HetNets with huge amount of wireless which mathematically analyzes analog input values in terms
sensors [56] the efficient-energy coverage problem requires of logical variables that take on continuous values
and motivates the design of a novel ACO algorithm by using between 0 and 1 [59]. Fuzzy logic is able to create rules by
three types of pheromone (i.e., local pheromone to help an inferring knowledge from imprecise, uncertain, or unreliable
ant organize its coverage set with fewer sensors, and two information [60]. And hence it can rapidly help to make
global pheromone for optimizing the number of required decisions for what action to take, based on a set of ‘‘rules’’
active sensors per Point of Interest (PoI) and forming a sensor over great amount information of physical and systemati-
set that has as many sensors as an ant has selected the number cal resources with varying conditions and characteristics in
of active sensors) to find the solution efficiently. HetNets. How to utilize Fuzzy System for improving HetNets
is illustrated in Fig. 7. Fuzzy system techniques are mostly
D. ARTIFICIAL NEURAL NETWORKS extended for handoff decision algorithms with certain adapt-
Artificial neural networks (ANNs) are one type of statistical ability to specific network environments, e.g., the adaptive
learning models inspired from the biological neural networks neuro-fuzzy based VHO decision algorithm in [24].
of human brains [57]. ANNs are generally presented as In HetNets with vehicular users, multi-parameters handoff
systems consisting of interconnected ‘‘neurons’’ within decisions should be made adaptively to user speed [25], where
numerically weighted values which can be tuned for mak- the type-2 fuzzy logic contributes significantly by naturally
ing ANNs adaptive to inputs and capable of learning [58]. handling more uncertainty in vehicular HetNets. Vertical
Therefore, the neural network paradigm assures the ability to handoff algorithm can balance the flow load by the proposal
learn from the unsupervised environments, and thus exhibits in [61] based on determining the optimal flow-dividing ratio
an effectiveness in applying to HetNets for estimating or of the traffic delivered in HetNets.
approximating functions that depend on many unknown input Fuzzy logic provides easy understanding over com-
conditions [3]. Note that the functionality and working flow plicated HetNets and has been discussed for improving
are shown in Fig. 6. performance (e.g., load balancing, coverage optimization,
The study in [17] proposes self-optimization of antenna and energy efficiency) in commercial LTE [62] and
tilt and power by using a fuzzy neural network optimization WiMAX networks [63], respectively. More specifically,
method based on RL, which is one of the most important tasks the intercell-interference coordination in HetNets can be
in the context of SON and can meet the need of practical addressed by employing fuzzy logic controllers as studied
applications of self-optimization in a dynamic environment in [64].

1386 VOLUME 3, 2015

X. Wang et al.: AI-Based Techniques for Emerging HetNet

FIGURE 7. Illustration of fuzzy logic handoff in HetNets.

F. MARKOV MODELS AND BAYESIAN-BASED GAMES as well. Hence, HetNets may induce significant technical
Although Markov Models [65] and Bayesian Methods [66] problems and raise substantial challenges.
are not exactly AI techniques, they still provide statistical
solutions for HetNets with certain ability of automatic opti- A. OPTIMIZATION ON VIRTUAL RESOURCE IN THE
mization. Along with the stochastically transitions among DATA CENTER FOR CLOUD-BASED HetNets
various states in Markov Models and the Bayesian-based One essential direction of 5G evolution is the resource
learning, the localization and mobility issues in HetNets can virtualization based on cloud computing techniques for elas-
be improved based on decision-tree based prediction. tic and dynamic resource allocation according to the ever
Hidden Markov Models (HMM) are often used for changing service requirements [73], [74]. Not only RRHs
modeling the signal strength history to achieve accurate will cooperate with the virtual resource of cloud-based RAN,
localization as studied in [67]. Also the study in [68] proposes but also in HetNets, much of the computation load of the
Preferred Route Indoor Mobility Model for indoor environ- computation-intensive tasks for radio access and processing
ments in HetNets, based on HMM as well to perform mobility will be offloaded to the centralized cloud in MNOs’ data
modelling and learning. BS-cooperative power management centers.
with self-organized sleep mode in virtual cell-based For instance, the proposal in [75] has introduced a
HetNets [69] uses Semi-Markov stochastic process to derive concept of cloud cooperated heterogeneous cellular
detailed analytic formulas of average cumulative delay and network (C-HetNet) where femtocell and picocell over-
interference time as well as the energy consumption. laid on a macrocell are connected to cloud radio access
The study in [70] formulates the network selection problem network (C-RAN) to operate cooperatively. Also, the study
in HetNets with incomplete information as a Bayesian game in [76] develops a theoretic computation framework of the
and uses Bayesian Nash equilibrium as the solution by deployment cost of the HetNets (modelled using various
applying the Bayesian best response dynamics and aggregat- spatial point processes) to explore whether heterogeneous
ing best response dynamics. The study in [71] investigates the C-RAN is cost-effective. These frameworks offer potential
application of sparse Bayesian methods for anomaly detec- research models for how to utilize AI algorithms to improve-
tion in HetNets, and constructs a detection system, which ment virtualized HetNets.
can provide human interpretability. More complicated studies The heterogeneity in terms of cell types and RRHs will
based on game theories can be referred to [72]. allow the MNOs to further enhance the performance with
respect to co-channel interference management, cell
V. FUTURE CHALLENGES coverage, traffic engineering, routing optimization as well as
The paradigm of mobile networks has been shifted from mobility and handoff, by sufficiently utilizing the collabo-
traditional centralized infrastructure to distributed and hier- rative resources on virtual machines according to the user
archical HetNets with more autonomous, cooperated, and demands and link conditions. Due to the convenient obtaining
intelligent rollouts. In the near future, cells (base stations) of status of all cells (virtualized devices in the same data
are equipped with certain automated mechanisms, and hence center), many learning-based and bio-inspired algorithms can
can reposition themselves optimally corresponding to traffic be effectively utilized for multi-objective optimization over
dynamics, user demands, and physical environment interfer- virtual resource, but yet have not been exploited.
ence conditions.
Although till now AI-based self-evolutionary techniques B. UTILIZING THE INTELLIGENCE OF SOFTWARE-
for HetNets have been researched a lot, it is not yet clear how DEFINED NETWORKS IN 5G HetNets
and when will HetNets work towards realistic deployment. The Software-Defined Network (SDN) technology [77] is
AI-based techniques for HetNets can be seen as an excellent one important enabler for the evolution of current networks
opportunity for enhancements, and also introduce challenges towards the intelligent 5G HetNets [78]. It offers real-time

VOLUME 3, 2015 1387

 X. Wang et al.: AI-Based Techniques for Emerging HetNet

packet reporting and re-activity functions of traffic

flow characteristics, and thus has been bringing basic
intelligence to the networks. In addition, SDN’s fine-grained
packet classifier and decoupling of control plane from data
plane of network hardware can offer central controllability
and powerful programmability to enable an intelligent carrier
cloud environment by allocating networking resources for
better consolidation of heterogeneous hardware resource with
high interoperability, andresilience.
Therefore, based on SDN’s inherent packet classifier for
traffic recognition and real-time programmability, there is a
large potential of designing appropriate methods and algo-
rithms for network failure recovery (self-healing) to ensure
services and resources availability. For example, the study
in [79] aims to simplify authentication handover by global
management of HetNets through sharing of user-dependent
security context information among SDN-assisted related
access points.
Bio-inspired algorithms like ACO and GA, which are often
used for route optimization, can accelerate the flow opti-
mization along with the programability of SDN, e.g., the
study in [80] presents an OpenFlow-based load balancing
system with the GA but it does not consider the heterogeneity FIGURE 8. Potential relationship between bee hive and BS caching.
and complexity of the emerging HetNets, and hence we can
expect a huge performance improvement in the near future. and social structure of HetNets, and how to formulate
collaborative BS caching among various cells in distributed
C. DISTRIBUTED EDGE INTELLIGENCE IN HetNets manner, by borrowing creative ideas from nature cooperative
Decentralizing current mobile network is not as easy as evolution of living organism systems (e.g., cooperation of
virtualizing computation load into the cloud data center, but bees in their hive as illustrated in Fig. 8) become critical but
moving services toward the network edge will be put to a challenging. Research on BS caching just started but it unveils
priority [81]. The significance of this trend is twofold: 1) it a new stage of edge computing and service, where more
reduces latency for users, and service providers can enjoy needs to be discovered. Also, how to integrate cognitive radio
more revenue opportunities in HetNets, e.g., local content, techniques [84], [85] with SON (e.g., game-based, learning-
localized radio resource management, and value-added data based, bio-inspired ones) becomes a promising research topic
processing at network edge; 2) it can significantly reduce for cognitive edge intelligence of HetNets in the near
network congestion in the core network, which is increasingly future [86], [87].
a risk due to the proliferation of media-intensive applications.
Furthermore, some critical challenges in the emerging D. AI TECHNIQUES FOR MANAGING MACHINE-TO-
HetNets, i.e., the coordination and management of various MACHINE (M2M) COMMUNICATIONS AND
types of devices, must be resolved with decentralized meth- INTERNET OF THINGS (IoT) IN HetNets
ods, by which we expect more service functions and manage- HetNets not only facilitate the mobile communication ser-
ment decisions can be made locally at the involved devices, vices of mobile users, but also integrate with the technology
rather than always gathering and processing in the center of M2M Communications [88] and IoT [89], which realize
modules. the information exchange among devices with intelligence,
One emerging technical issue of edge intelligence is e.g., wireless sensors in body area networks, embedded smart
the Base Station Caching [81], [82], by which most of the devices in buildings, outdoor environments, and even tradi-
content delivery can be locally achieved by caching at the tionally difficult-to-reach locations. However, for supporting
base stations of small cells. The preliminary study in [83] such an extremely large number of embedded devices,
proposes a transfer learning-based caching scheme for small conventional networking approaches seem unsuitable for sce-
cell base stations in HetNets by exploiting the rich contextual narios of high heterogeneity, scalability and complexity [90].
information (i.e., users’ content viewing history, social ties, It is expected to induce natural AI-based techniques to control
etc.), which can be further extended by exploring means of and optimize the collaboration of sensors and to provide
state-of-art ML and bio-inspired techniques in the future. stable operations and service management functionalities in
Thus according to the dynamical user demands and high fully distributed manner.
diversity of contents, how to carry out more effective Bio-ecosystems naturally run over huge amount of entities,
learning-based caching strategies for exploiting the spatial and thus bio-inspired techniques can deal with HetNets with

1388 VOLUME 3, 2015

X. Wang et al.: AI-Based Techniques for Emerging HetNet

M2M and IoT inherently. For instance, a new bio-inspired [11] S. Sun, M. Kadoch, and T. Ran, ‘‘Adaptive SON and cognitive smart
approach, BIONETS [91], is proposed to enhance the system LPN for 5G heterogeneous networks,’’ Mobile Netw. Appl., Jan. 2015, doi:
10.1007/s11036-014-0563-2
performance in terms of scalability, robustness and efficiency [12] W. Guo and T. O’Farrell, ‘‘Relay deployment in cellular networks:
based on bio-inspired management. So the challenge is to Planning and optimization,’’ IEEE J. Sel. Areas Commun., vol. 31, no. 8,
treat the whole HetNets as a living ecosystem, where the pp. 1597–1606, Aug. 2013.
[13] C.-S. Tseng, C.-C. Chen, and C. Lin, ‘‘A bio-inspired robust adaptive
functions, modules and services play the role of organisms, random search algorithm for distributed beamforming,’’ in Proc. IEEE
evolving and combining themselves to successfully adapt ICC, Jun. 2011, pp. 1–6.
to the dynamic environment (e.g., user demands, network [14] K. M. Passino, ‘‘Biomimicry of bacterial foraging for distributed optimiza-
tion and control,’’ IEEE Control Syst. Mag., vol. 22, no. 3, pp. 52–67,
conditions etc.) [56]. Future research and development on Jun. 2002.
HetNets technologies must address new requirements from [15] W. Li, P. Yu, Z. Jiang, and Z. Li, ‘‘Centralized management mechanism for
M2M and IoT services, by adopting effective management cell outage compensation in LTE networks,’’ Int. J. Distrib. Sensor Netw.,
vol. 2012, pp. 1–8, Jul. 2012.
and optimization techniques inspired from bio-ecosystems, [16] Z. Zhao, J. Chen, and N. Crespi, ‘‘A policy-based framework for autonomic
e.g., GA and ACO of swarm intelligence [92]. reconfiguration management in heterogeneous networks,’’ in Proc. 7th Int.
Conf. Mobile Ubiquitous Multimedia, 2008, pp. 71–78.
[17] S. Fan, H. Tian, and C. Sengul, ‘‘Self-optimization of coverage and
VI. CONCLUSION capacity based on a fuzzy neural network with cooperative reinforce-
The emerging HetNets will be characterised by a higher ment learning,’’ EURASIP J. Wireless Commun. Netw., vol. 2014, no. 1,
pp. 57:1–57:14, Apr. 2014.
degree of capillarity, density and higher bit rates. By integrat-
[18] M. Caleffi and L. Paura, ‘‘Bio-inspired link quality estimation for wireless
ing AI-based SON features, HetNets may have high potential mesh networks,’’ in Proc. IEEE Int. Symp. WoWMoM, Jun. 2009, pp. 1–6.
to significantly boost the mobile service performance. Most [19] A.-H. Tsai, L.-C. Wang, J.-H. Huang, and T.-M. Lin, ‘‘Intelligent resource
of the contributions are from adaptive spectrum reutilisation, management for device-to-device (D2D) communications in heteroge-
neous networks,’’ in Proc. 15th Int. Symp. WPMC, Sep. 2012, pp. 75–79.
high automatic-ability on system development, and the [20] A. Engels, M. Reyer, X. Xu, R. Mathar, J. Zhang, and H. Zhuang,
evolutive optimization of systems, as well as the effectiveness ‘‘Autonomous self-optimization of coverage and capacity in LTE cellular
of the system fault-tolerance. networks,’’ IEEE Trans. Veh. Technol., vol. 62, no. 5, pp. 1989–2004,
Jun. 2013.
Through this survey and analysis, AI-based techniques are [21] M. Peng, Y. Liu, D. Wei, W. Wang, and H.-H. Chen, ‘‘Hierarchical coop-
discussed and proved to be able to acclimatize and be compe- erative relay based heterogeneous networks,’’ IEEE Wireless Commun.,
tent for the smart improvement of HetNets by the self-‘‘X’’ vol. 18, no. 3, pp. 48–56, Jun. 2011.
[22] M. Liu, Z. Li, X. Guo, and E. Dutkiewicz, ‘‘Performance analysis and
features. Furthermore, throughout the taxonomy, we have optimization of handoff algorithms in heterogeneous wireless networks,’’
identified their major advantages as well as their technical IEEE Trans. Mobile Comput., vol. 7, no. 7, pp. 846–857, Jul. 2008.
challenges and research problems. Particularly, interesting [23] R. Chai, J. Cheng, X. Pu, and Q. Chen, ‘‘Neural network based vertical
handoff performance enhancement in heterogeneous wireless networks,’’
opening research issues include AI-based auto-optimization in Proc. WiCOM, Sep. 2011, pp. 1–4.
on virtualized resource integrated with SDN, M2M and IoT. [24] A. Çalhan and C. Ceken, ‘‘An adaptive neuro-fuzzy based vertical handoff
decision algorithm for wireless heterogeneous networks,’’ in Proc. IEEE
21st Int. Symp. PIMRC, Sep. 2010, pp. 2271–2276.
REFERENCES [25] B. Ma and X. Liao, ‘‘Speed-adaptive vertical handoff algorithm based on
[1] W. Guo, S. Wang, X. Chu, J. Zhang, J. Chen, and H. Song, ‘‘Auto- fuzzy logic in vehicular heterogeneous networks,’’ in Proc. 9th Int. Conf.
mated small-cell deployment for heterogeneous cellular networks,’’ IEEE FSKD, May 2012, pp. 371–375.
Commun. Mag., vol. 51, no. 5, pp. 46–53, May 2013. [26] S. B. Kotsiantis, ‘‘Supervised machine learning: A review of classification
[2] J. Allman, Evolving Brains (Scientific American Library). New York, NY, techniques,’’ Informatica, vol. 31, no. 3, pp. 249–268, 2007.
USA: Freeman, 2000. [27] L. P. Kaelbling, M. L. Littman, and A. W. Moore, ‘‘Reinforcement learn-
ing: A survey,’’ J. Artif. Intell. Res., vol. 4, pp. 237–285, 1996.
[3] Z. Zhang, K. Long, J. Wang, and F. Dressler, ‘‘On swarm intelligence
[28] M. A. Khan, H. Tembine, and A. V. Vasilakos, ‘‘Game dynamics and cost
inspired self-organized networking: Its bionic mechanisms, designing prin-
of learning in heterogeneous 4G networks,’’ IEEE J. Sel. Areas Commun.,
ciples and optimization approaches,’’ IEEE Commun. Surv. Tuts., vol. 16,
vol. 30, no. 1, pp. 198–213, Jan. 2012.
no. 1, pp. 513–537, First Quarter 2014.
[29] C. Watkins and P. Dayan, ‘‘Q-learning,’’ Mach. Learn., vol. 8, nos. 3–4,
[4] O. G. Aliu, A. Imran, M. A. Imran, and B. Evans, ‘‘A survey of self
pp. 279–292, 1992.
organisation in future cellular networks,’’ IEEE Commun. Surv. Tuts.,
[30] P. T. Semov, A. Mihovska, R. Prasad, and V. Poulkov, ‘‘Use of positioning
vol. 15, no. 1, pp. 336–361, First Quarter 2013.
information for performance enhancement of uncoordinated heteroge-
[5] Z. Zhang, W. Huangfu, K. Long, X. Zhang, X. Liu, and B. Zhong, neous network deployment,’’ in Proc. 3rd Int. Conf. VITAE, Jun. 2013,
‘‘On the designing principles and optimization approaches of bio-inspired pp. 1–6.
self-organized network: A survey,’’ Sci. China Inf. Sci., vol. 56, no. 7, [31] P. T. Semov, V. Poulkov, A. Mihovska, and R. Prasad, ‘‘Increasing through-
pp. 1–28, Jul. 2013. put and fairness for users in heterogeneous semi coordinated deploy-
[6] A. Imran, A. Zoha, and A. Abu-Dayya, ‘‘Challenges in 5G: How to ments,’’ in Proc. IEEE WCNC, Apr. 2014, pp. 40–45.
empower SON with big data for enabling 5G,’’ IEEE Netw., vol. 28, no. 6, [32] M. Simsek, M. Bennis, and I. Güvenc, ‘‘Context-aware mobility man-
pp. 27–33, Nov./Dec. 2014. agement in HetNets: A reinforcement learning approach,’’ in Proc. Netw.
[7] HetNet/Small Cells. [Online]. Available: http://www.3gpp.org/hetnet, Internet Archit., May 2015. [Online]. Available: arxiv.org/pdf/1505.01625
accessed Aug. 2014. [33] Q. Li, H. Xia, Z. Zeng, and T. Zhang, ‘‘Dynamic enhanced inter-cell
[8] X. Su, J. Zeng, and C. Xiao, ‘‘Key technologies for SON in next generation interference coordination using reinforcement learning approach in het-
radio access networks,’’ in Proc. 23rd ICCCN, Aug. 2014, pp. 1–8. erogeneous network,’’ in Proc. IEEE ICCT, Nov. 2013, pp. 239–243.
[9] M. Peng, D. Liang, Y. Wei, J. Li, and H.-H. Chen, ‘‘Self-configuration [34] M. Bennis, S. M. Perlaza, P. Blasco, Z. Han, and H. V. Poor, ‘‘Self-
and self-optimization in LTE-advanced heterogeneous networks,’’ IEEE organization in small cell networks: A reinforcement learning approach,’’
Commun. Mag., vol. 51, no. 5, pp. 36–45, May 2013. IEEE Trans. Wireless Commun., vol. 12, no. 7, pp. 3202–3212, Jul. 2015.
[10] L. T. W. Ho, I. Ashraf, and H. Claussen, ‘‘Evolving femtocell coverage [35] W. Guo, S. Wang, Y. Wu, J. Rigelsford, X. Chu, and T. O’Farrell, ‘‘Spectral-
optimization algorithms using genetic programming,’’ in Proc. IEEE 20th and energy-efficient antenna tilting in a HetNet using reinforcement learn-
Int. Symp. PIMRC, Sep. 2009, pp. 2132–2136. ing,’’ in Proc. IEEE WCNC, Apr. 2013, pp. 767–772.

VOLUME 3, 2015 1389

 X. Wang et al.: AI-Based Techniques for Emerging HetNet

[36] M. Simsek, M. Bennis, and A. Czylwik, ‘‘Coordinated beam selection [58] A. V. Vasilakos and G. I. Papadimitriou, ‘‘A new approach to the design of
in LTE-advanced HetNets: A reinforcement learning approach,’’ in Proc. reinforcement schemes for learning automata: Stochastic estimator learn-
IEEE Globecom Workshops (GC Wkshps), Dec. 2012, pp. 603–607. ing algorithm,’’ Neurocomputing, vol. 7, no. 3, pp. 275–297, Apr. 1995.
[37] M. Dirani and Z. Altman, ‘‘Self-organizing networks in next generation [59] G. Feng, ‘‘A survey on analysis and design of model-based fuzzy control
radio access networks: Application to fractional power control,’’ Comput. systems,’’ IEEE Trans. Fuzzy Syst., vol. 14, no. 5, pp. 676–697, Oct. 2006.
Netw., vol. 55, no. 2, pp. 431–438, Feb. 2011. [60] A. V. Vasilakos, C. Ricudis, K. Anagnostakis, W. Pedryca, and
[38] M. Simsek and A. Czylwik, ‘‘Improved decentralized fuzzy Q-learning A. Pitsillides, ‘‘Evolutionary-fuzzy prediction for strategic QoS routing
for interference reduction in heterogeneous LTE-networks,’’ in Proc. Int. in broadband networks,’’ in Proc. IEEE Int. Conf. Fuzzy Syst., vol. 2.
OFDM Workshop, Aug. 2012, pp. 1–6. May 1998, pp. 1488–1493.
[39] D. Lopez-Perez, X. Chu, A. V. Vasilakos, and H. Claussen, ‘‘Power [61] N. Wang, W. Shi, S. Fan, and Y. Liu, ‘‘Flow diversion-based vertical
minimization based resource allocation for interference mitigation in handoff algorithm for heterogeneous wireless networks,’’ J. Comput. Inf.
OFDMA femtocell networks,’’ IEEE J. Sel. Areas Commun., vol. 32, no. 2, Syst., vol. 7, no. 13, pp. 4863–4870, 2011.
pp. 333–344, Feb. 2014. [62] P. Munoz, R. Barco, J. M. Ruiz-Aviles, I. de la Bandera, and A. Aguilar,
[40] D. Lopez-Perez, X. Chu, A. V. Vasilakos, and H. Claussen, ‘‘On dis- ‘‘Fuzzy rule-based reinforcement learning for load balancing techniques
tributed and coordinated resource allocation for interference mitigation in in enterprise LTE femtocells,’’ IEEE Trans. Veh. Technol., vol. 62, no. 5,
self-organizing LTE networks,’’ IEEE/ACM Trans. Netw., vol. 21, no. 4, pp. 1962–1973, Jun. 2013.
pp. 1145–1158, Aug. 2013. [63] S.-S. Seo, S.-S. Kim, N. Agoulmine, and J. W.-K. Hong, ‘‘On achieving
[41] M. Simsek, A. Czylwik, and M. Bennis, ‘‘On interference manage- self-organization in mobile WiMAX network,’’ in Proc. IEEE/IFIP Netw.
ment techniques in LTE heterogeneous networks,’’ in Proc. ICCCN, Oper. Manage. Symp. Workshops, Apr. 2010, pp. 43–50.
Jul./Aug. 2012, pp. 1–5. [64] A. Daeinabi, K. Sandrasegaran, and P. Ghosal, ‘‘An enhanced intercell
[42] M. Srinivas and L. M. Patnaik, ‘‘Genetic algorithms: A survey,’’ interference coordination scheme using fuzzy logic controller in
Computer, vol. 27, no. 6, pp. 17–26, Jun. 1994. LTE-advanced heterogeneous networks,’’ in Proc. Int. Symp. WPMC,
[43] I. E. Kassotakis, M. E. Markaki, and A. V. Vasilakos, ‘‘A hybrid genetic Sep. 2014, pp. 520–525.
approach for channel reuse in multiple access telecommunication net- [65] Y. Bengio, ‘‘Markovian models for sequential data,’’ Neural Comput. Surv.,
works,’’ IEEE J. Sel. Areas Commun., vol. 18, no. 2, pp. 234–243, vol. 2, pp. 129–162, 1999.
Feb. 2000. [66] P. Bajari, H. Hong, and D. Nekipelov, ‘‘Game theory and econometrics:
[44] Y. Song, L. Liu, H. Ma, and A. V. Vasilakos, ‘‘A biology-based algorithm A survey of some recent research,’’ in Proc. MIMEO, 2010, pp. 1–43.
to minimal exposure problem of wireless sensor networks,’’ IEEE Trans. [67] M. Ibrahim and M. Youssef, ‘‘A hidden Markov model for localization
Netw. Service Manag., vol. 11, no. 3, pp. 417–430, Sep. 2014. using low-end GSM cell phones,’’ in Proc. IEEE ICC, Jun. 2011, pp. 1–5.
[45] L. Liu, Y. Song, H. Zhang, H. Ma, and A. V. Vasilakos, ‘‘Physarum [68] L. Ning, Z. Wang, and Q. Guo, ‘‘Preferred route indoor mobility model
optimization: A biology-inspired algorithm for the Steiner tree problem in for heterogeneous networks,’’ IEEE Commun. Lett., vol. 18, no. 5,
networks,’’ IEEE Trans. Comput., vol. 64, no. 3, pp. 818–831, Mar. 2015. pp. 821–824, May 2014.
[69] W. Zheng, W. Li, Y. Xie, and X. Wen, ‘‘Hybrid BS-cooperative power
[46] A. Pitsillides, G. Stylianou, C. S. Pattichis, A. Sekercioglu, and
management scheme with self-organized sleep mode in virtual cell-based
A. V. Vasilakos, ‘‘Bandwidth allocation for virtual paths (BAVP): Investi-
femto networks,’’ Int. J. Distrib. Sensor Netw., vol. 2012, Sep. 2012,
gation of performance of classical constrained and genetic algorithm based
Art. ID 785418.
optimisation techniques,’’ in Proc. IEEE INFOCOM, vol. 3. Mar. 2000,
[70] K. Zhu, D. Niyato, and P. Wang, ‘‘Network selection in heterogeneous
pp. 1501–1510.
wireless networks: Evolution with incomplete information,’’ in Proc. IEEE
[47] O. Abdelkhalek, S. Krichen, A. Guitouni, and S. Mitrovic-Minic,
WCNC, Apr. 2010, pp. 1–6.
‘‘A genetic algorithm for a multi-objective nodes placement problem in
[71] J. Zhang and R. Kannan, ‘‘A sparse Bayesian framework for anomaly
heterogeneous network infrastructure for surveillance applications,’’ in
detection in heterogeneous networks,’’ in Quality, Reliability, Security and
Proc. IFIP WMNC, Oct. 2011, pp. 1–9.
Robustness in Heterogeneous Networks (Lecture Notes of the Institute
[48] B. Lorenzo and S. Glisic, ‘‘Optimal routing and traffic scheduling for
for Computer Sciences, Social Informatics and Telecommunications Engi-
multihop cellular networks using genetic algorithm,’’ IEEE Trans. Mobile
neering), vol. 74. Berlin, Germany: Springer-Verlag, 2012, pp. 75–87.
Comput., vol. 12, no. 11, pp. 2274–2288, Nov. 2013.
[72] C. Busch, R. Kannan, and A. V. Vasilakos, ‘‘Approximating
[49] D. Martens, B. Baesens, and T. Fawcett, ‘‘Editorial survey: Swarm intelli- congestion + dilation in networks via ‘quality of routing’ games,’’
gence for data mining,’’ Mach. Learn., vol. 82, no. 1, pp. 1–42, Jan. 2011. IEEE Trans. Comput., vol. 61, no. 9, pp. 1270–1283, Sep. 2012.
[50] A. V. Vasilakos, M. P. Saltouros, A. F. Atlassis, and W. Pedrycz, ‘‘Opti- [73] N. Zhang, N. Cheng, A. T. Gamage, K. Zhang, J. W. Mark, and X. Shen,
mizing QoS routing in hierarchical ATM networks using computational ‘‘Cloud assisted HetNets toward 5G wireless networks,’’ IEEE Commun.
intelligence techniques,’’ IEEE Trans. Syst., Man, Cybern. C, Appl. Rev., Mag., vol. 53, no. 6, pp. 59–65, Jun. 2015.
vol. 33, no. 3, pp. 297–312, Aug. 2003. [74] M. R. Rahimi, J. Ren, C. H. Liu, A. V. Vasilakos, and N. Venkatasub-
[51] R. Siddavaatam, ‘‘Joint graph coloring and ant colony ramanian, ‘‘Mobile cloud computing: A survey, state of art and future
optimization based sub-channel allocation algorithms for downlink directions,’’ Mobile Netw. Appl., vol. 19, no. 2, pp. 133–143, 2014.
LTE—A HetNets,’’ Master dissertation, Dept. Electr. Comput. Eng., [75] K. Sakaguchi, S. Sampei, H. Shimodaira, R. Rezagah, G. K. Tran, and
Ryerson Univ. Toronto, ON, Canada, Aug. 2012. K. Araki, ‘‘Cloud cooperated heterogeneous cellular networks,’’ in Proc.
[52] Y. Zhu, Z. Zeng, T. Zhang, L. An, and L. Xiao, ‘‘An energy efficient ISPACS, Nov. 2013, pp. 787–791.
user association scheme based on cell sleeping in LTE heterogeneous [76] V. Suryaprakash, P. Rost, and G. Fettweis, ‘‘Are heterogeneous cloud-
networks,’’ in Proc. WPMC, Sep. 2014, pp. 75–79. based radio access networks cost effective?’’ IEEE J. Sel. Areas Commun.,
[53] I. Kassabalidis, M. A. El-Sharkawi, R. J. Marks, P. Arabshahi, and vol. PP, no. 99, May 2015, doi: 10.1109/JSAC.2015.2435275.
A. A. Gray, ‘‘Swarm intelligence for routing in communication networks,’’ [77] B. A. A. Nunes, M. Mendonca, X.-N. Nguyen, K. Obraczka, and T. Turletti,
in Proc. IEEE GLOBECOM, vol. 6. Nov. 2001, pp. 3613–3617. ‘‘A survey of software-defined networking: Past, present, and future of
[54] H. Shokrani and S. Jabbehdari, ‘‘A survey of ant-based routing programmable networks,’’ IEEE Commun. Surveys Tuts., vol. 16, no. 3,
algorithms for mobile ad-hoc networks,’’ in Proc. ICSPS, May 2009, pp. 1617–1634, Third Quarter 2014.
pp. 323–329. [78] M. Mendonca, K. Obraczka, and T. Turletti, ‘‘The case for software-
[55] R. Han, C. Feng, H. Xia, and Y. Wu, ‘‘Coverage optimization for defined networking in heterogeneous networked environments,’’ in Proc.
dense deployment small cell based on ant colony algorithm,’’ in Proc. ACM Conf. CoNEXT Student Workshop, 2012, pp. 59–60.
IEEE VTC Fall, Sep. 2014, pp. 1–5. [79] X. Duan and X. Wang, ‘‘Authentication handover and privacy protection
[56] J.-W. Lee, B.-S. Choi, and J.-J. Lee, ‘‘Energy-efficient coverage of wire- in 5G hetnets using software-defined networking,’’ IEEE Commun. Mag.,
less sensor networks using ant colony optimization with three types of vol. 53, no. 4, pp. 28–35, Apr. 2015.
pheromones,’’ IEEE Trans. Ind. Informat., vol. 7, no. 3, pp. 419–427, [80] L.-D. Chou, Y.-T. Yang, Y.-M. Hong, J.-K. Hu, and B. Jean, ‘‘A genetic-
Aug. 2011. based load balancing algorithm in OpenFlow network,’’ in Advanced
[57] G. P. Zhang, ‘‘Neural networks for classification: A survey,’’ IEEE Technologies, Embedded and Multimedia for Human-Centric Computing
Trans. Syst., Man, Cybern. C, Appl. Rev., vol. 30, no. 4, pp. 451–462, (Lecture Notes in Electrical Engineering), vol. 260. Amsterdam,
Nov. 2000. The Netherlands: Springer-Verlag, 2014, pp. 411–417.

1390 VOLUME 3, 2015

X. Wang et al.: AI-Based Techniques for Emerging HetNet

[81] X. Wang, M. Chen, T. Taleb, A. Ksentini, and V. Leung, ‘‘Cache in the air: XIUHUA LI (S’12) received the B.S. and
Exploiting content caching and delivery techniques for 5G systems,’’ IEEE M.S. degrees from the Honors School and
Commun. Mag., vol. 52, no. 2, pp. 131–139, Feb. 2014. the School of Electronics and Information
[82] E. Bastug, M. Bennis, and M. Debbah, ‘‘Living on the edge: The role of Engineering, Harbin Institute of Technology,
proactive caching in 5G wireless networks,’’ IEEE Commun. Mag., vol. 52, in 2011 and 2013, respectively. He is currently
no. 8, pp. 82–89, Aug. 2014. pursuing the Ph.D. degree with the Department
[83] E. Baştuğ, M. Bennis, and M. Debbah, ‘‘A transfer learning approach for of Electrical and Computer Engineering, The Uni-
cache-enabled wireless networks,’’ in Proc. WiOpt D2D Workshop, 2015,
versity of British Columbia. His current research
pp. 1–6.
interests are resource allocation, optimization,
[84] M. Youssef, M. Ibrahim, M. Abdelatif, L. Chen, and A. V. Vasilakos,
‘‘Routing metrics of cognitive radio networks: A survey,’’ IEEE Commun. distributed antenna systems, cooperative backhaul
Surveys Tuts., vol. 16, no. 1, pp. 92–109, First Quarter 2014. caching, and traffic offloading in mobile content-centric networks.
[85] J. Esch, ‘‘A survey of security challenges in cognitive radio networks:
Solutions and future research directions,’’ Proc. IEEE, vol. 100, no. 12,
pp. 3172–3186, Dec. 2012.
[86] S.-S. Byun, I. Balashingham, A. V. Vasilakos, and H.-N. Lee, ‘‘Computa-
tion of an equilibrium in spectrum markets for cognitive radio networks,’’
IEEE Trans. Comput., vol. 63, no. 2, pp. 304–316, Feb. 2014.
[87] T. Jiang, H. Wang, and A. V. Vasilakos, ‘‘QoE-driven channel allocation
schemes for multimedia transmission of priority-based secondary users VICTOR C. M. LEUNG (S’75–M’89–SM’97–
over cognitive radio networks,’’ IEEE J. Sel. Areas Commun., vol. 30, no. 7, F’03) received the B.A.Sc. (Hons.) and
pp. 1215–1224, Aug. 2012. Ph.D. degrees in electrical engineering from
[88] J. Kim, J. Lee, J. Kim, and J. Yun, ‘‘M2M service platforms: Survey, issues, The University of British Columbia (UBC),
and enabling technologies,’’ IEEE Commun. Surveys Tuts., vol. 16, no. 1, in 1977 and 1981, respectively. He attended the
pp. 61–76, First Quarter 2014.
Graduate School, UBC, on a Natural Sciences
[89] L. Da Xu, W. He, and S. Li, ‘‘Internet of Things in industries: A survey,’’
and Engineering Research Council Post-Graduate
IEEE Trans. Ind. Informat., vol. 10, no. 4, pp. 2233–2243, Nov. 2014.
[90] Y. Yao, Q. Cao, and A. V. Vasilakos, ‘‘EDAL: An energy-efficient, delay- Scholarship. He received the APEBC Gold Medal
aware, and lifetime-balancing data collection protocol for heterogeneous as the Head of the Graduating Class in the Faculty
wireless sensor networks,’’ IEEE/ACM Trans. Netw., vol. 23, no. 3, of Applied Science.
pp. 810–823, Jun. 2015. He was a Senior Member of the Technical Staff and Satellite System
[91] I. Carreras, I. Chlamtac, F. De Pellegrini, and D. Miorandi, ‘‘BIONETS: Specialist with MPR Teltech Ltd., Canada, from 1981 to 1987. In 1988, he
Bio-inspired networking for pervasive communication environments,’’ was a Lecturer with the Department of Electronics, Chinese University of
IEEE Trans. Veh. Technol., vol. 56, no. 1, pp. 218–229, Jan. 2007. Hong Kong. He returned to UBC as a Faculty Member in 1989, where he is
[92] Y. Lin, J. Zhang, H. S.-H. Chung, W. H. Ip, Y. Li, and Y.-H. Shi, ‘‘An ant currently a Professor and the TELUS Mobility Research Chair in Advanced
colony optimization approach for maximizing the lifetime of heteroge- Telecommunications Engineering with the Department of Electrical and
neous wireless sensor networks,’’ IEEE Trans. Syst., Man, Cybern. C, Appl. Computer Engineering. He has co-authored over 800 technical papers in
Rev., vol. 42, no. 3, pp. 408–420, May 2012. international journals and conference proceedings, 31 book chapters, and
co-edited ten book titles. His research interests are in the areas of wireless
networks and mobile systems. Several of his papers had been selected for
best paper awards.
Dr. Leung is a fellow of the Royal Society of Canada, the Engineering
XIAOFEI WANG (S’06–M’13) received the Institute of Canada, and the Canadian Academy of Engineering. He was a
B.S. degree from the Department of Computer recipient of the IEEE Vancouver Section Centennial Award and the 2012
Science and Technology, Huazhong University UBC Killam Research Prize. He is a Registered Professional Engineer in
of Science and Technology, in 2005, and the the province of British Columbia, Canada. He was a Distinguished Lecturer
M.S. and Ph.D. degrees from the School of Com- of the IEEE Communications Society. He is a member of the Editorial
puter Science and Engineering, Seoul National Boards of the IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS—
University, in 2008 and 2013, respectively. He is SERIES ON GREEN COMMUNICATIONS AND NETWORKING, the IEEE WIRELESS
currently a Post-Doctoral Research Fellow with COMMUNICATIONS LETTERS, Computer Communications, and several other
the Department of Electrical and Computer Engi- journals. He has served on the Editorial Boards of the IEEE JOURNAL ON
neering, The University of British Columbia. His SELECTED AREAS IN COMMUNICATIONS—WIRELESS COMMUNICATIONS series, the
current research interests are social-aware multimedia service in cloud IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, the IEEE TRANSACTIONS
computing, cooperative backhaul caching, and traffic offloading in mobile ON VEHICULAR TECHNOLOGY, the IEEE TRANSACTIONS ON COMPUTERS, and the
content-centric networks. He was a recipient of the Korean Government Journal of Communications and Networks. He has guest-edited many journal
Scholarship for Excellent Foreign Students in the IT field from NIPA from special issues, and contributed to the organizing committees and technical
2008 to 2011. He received the Global Outstanding Chinese Ph.D. Student program committees of numerous conferences and workshops.
Award in 2012.

VOLUME 3, 2015 1391