SPECIAL SECTION ON RECENT ADVANCES IN SOFTWARE DEFINED NETWORKING

FOR 5G NETWORKS

Received August 16, 2015, accepted October 8, 2015, date of publication November 10, 2015,
date of current version November 18, 2015.
Digital Object Identifier 10.1109/ACCESS.2015.2498609

INVITED PAPER

Spatial Domain Management and Massive

MIMO Coordination in 5G SDN
SONGLIN SUN1 , (Member, IEEE), BO RONG2 , (Member, IEEE),
ROSE QINGYANG HU3 , (Senior Member, IEEE), AND YI QIAN4 , (Senior Member, IEEE)
1 Schoolof Information and Communication Engineering, Beijing University of Posts and Telecommunications, Beijing 100876, China
2 Communications Research Centre Canada, Ottawa, ONK2H 8S2, Canada
3 Department of Electrical and Computer Engineering, Utah State University, Logan, UT 84322-4120, USA
4 Department of Electrical and Computer Engineering, University of Nebraska–Lincoln, Lincoln, NE 68182-0572, USA

Corresponding author: R. Q. Hu (rose.hu@usu.edu)

This work was supported by the National Natural Science Foundation of China under Project 61471066.

ABSTRACT In 5G mobile communication systems, massive multiple-input multiple-output (MIMO) and

heterogeneous networks (HetNets) play crucial roles to achieve expected coverage and capacity across
venues. This paper correspondingly addresses software-defined network (SDN) as the central controller of
radio resource management in massive MIMO HetNets. In particular, we identify the huge spatial domain
information management and complicated MIMO coordination as the grand challenges in 5G systems. Our
work accordingly distinguishes itself by considering more network MIMO aspects, including flexibility and
complexity of spatial coordination. In our proposed scheme, SDN controller first collects the user channel-
state information in an effective way, and then calculates the null-space of victim users and applies linear
precoding to that null-space. Simulation results show that our design is highly beneficial and easy to be
deployed, due to its high quality of service performance but low computation complexity.

INDEX TERMS 5G SDN, massive MIMO, HetNets, precoding, channel estimation.

I. INTRODUCTION To address these challenges, operators are coping with

In the past few years, the fourth-generation (4G) mobile
communication systems have been rapidly deployed and
successfully operated in many countries. Nevertheless, the
dramatic adoption by end-users of smart-phones and tablets
is fueling an increasing demand for mobile data access.
To provide enhanced connectivity for more diversified
devices with higher mobility, both academia and industry
are focusing on the fifth-generation (5G) mobile system, and
related technologies have become popular research topics.
A major challenge in the era of 5G is to efficiently sup-
port the increasing demand of network capacity while the
spectrum resource remains scarce [1]. For instance, it is
expected that the future networks should be able to handle
the sophisticated content operations of a tenfold increase
in traffic with guaranteed quality of service (QoS) [2], [3].
In addition, energy efficiency has also been recognized as
an urgent requirement, since service providers are more
and more concerned about the operating expense (OPEX),
as well as the impact to global climate change and air
pollution [4], [5].
the trend by: 1) deploying denser cellular infrastructures;
2) increasing the available bandwidth; and 3) exploring mas-
sive MIMO solutions [6], [7]. In particular, one of the most
promising technologies is the HetNet architecture, in which
a cellular system consists of a large number of densified low
power nodes (LPNs) [8], [9]. In HetNets, LPNs provide high
data rate to nearby mobile stations and improve the system
capacity with frequency reuse. Moreover, LPNs can transmit
signals with lower power, resulting in a significant reduction
in energy consumption [10], [11].
Despite the promising features of massive MIMO based
HetNets, with the drastic increase of ultra-dense deploy-
ment of small cells in 5G networks, the total energy con-
sumption may still likely exceed the acceptable level [12].
Moreover, the traffic fluctuation also contributes consider-
able increase to energy consumption and thus causes extra
OPEX to service providers [13]. In short, wireless net-
work operators are now facing rapid growth of users and
traffics, which should be effectively curbed by introducing
intelligence [14], [15].

2169-3536
2015 IEEE. Translations and content mining are permitted for academic research only.
2238 Personal use is also permitted, but republication/redistribution requires IEEE permission. VOLUME 3, 2015
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
S. Sun et al.: Spatial Domain Management and Massive MIMO Coordination in 5G SDN
This ever increasing network complexity is driving opera-
tors toward a virtualization of network functionality that calls
for a paradigm shift from a hardware-based approach to a
software-based approach [16], [17]. In this work, we develop
an intelligent management framework based on the concept
of SDN, which is featured by the decoupling of control plane
from data plane [18], [19]. The intelligent SDN framework
aims to provide a viable way to solve the existing challenges
in a unified manner [20], [21].
In essence, SDN recognizes the network as an operating
system and abstracts the applications from the hardware [22],
which enables the management-related functions to be imple-
mented in a centralized manner [23]. Accordingly, network
intelligence can be realized logically in a centralized SDN
controller that manages the entire network globally. More
specifically, an SDN controller is programmed to learn the
physical topology of the network and status of each individual
network element through certain discovery mechanisms or
appropriate databases, and thus can orchestrate the whole
network to function in a cost-efficient and energy-efficient
manner [24].
This work studies how SDN benefits the RRM of next
generation mobile networks. Our proposed network architec-
ture is in line with 5G concepts and also compatible to cur-
rent 3GPP LTE Advanced (LTE-A) standard. In 4G systems,
RRM addresses frequency and time domain management,
such as frequency reuse [25], inter cell interference coordi-
nation (ICIC) [26], enhanced inter cell interference coordina-
tion (eICIC), etc. As to 5G, however, the system has to reach
the staggering densities of devices and network infrastructure
nodes, as well as supporting an enormous number of anten-
nas in massive MIMO installations [27], [28]. Fortunately,
mmWave technologies at very high carrier frequencies make
the size of antenna smaller and open the door to massive
MIMO [29], [30]. Nevertheless, massive MIMO leads to
the problem of managing huge spatial domain resource and
processing vast information in 5G HetNets. SDN shows its
inherent advantage of integrating and processing information
in dealing with such situations.
In 5G HetNets, there are macro base stations (Macro-BSs)
and small cells all employing multiple antennas and leading
to huge spacial domain information [31]. To retrieve, store
and utilize the spatial information becomes a critical task
for 5G network [32]. In our design, SDN is responsible
for collecting and managing huge spacial domain informa-
tion and processing null-space calculation regarding massive
MIMO coordination. SDN offers a platform to integrate spa-
tial information and perform massive MIMO coordination.
With this effort, the null-space information is integrated to
enable whole network coordination and thus achieves a better
performance than non-SDN based networks [33], [34]. This
novel architecture implements coordinated RRM and eases
the operation pressure of BS and LPNs significantly.
The rest of this paper is organized as follows. Section II
introduces the cutting edge technologies of 5G pertinent
to spatial domain management. In Section III, we present
VOLUME 3, 2015
status-quo of SDN and OpenFlow specifications. In
Section IV, we propose a novel SDN architecture for 5G
networks and an intelligent massive MIMO scheme operated
in 5G SDN. Section V demonstrates the performance evalu-
ation. In the end, Section VI concludes the paper.
II. 5G CUTTING EDGE TECHNOLOGIES
REGARDING SPATIAL DOMAIN
A. HETEROGENEOUS NETWORKS
HetNet has been recognized as the most viable solution
to the impending mobile data traffic crunch in the context
of LTE-A. It consists of eNodeBs of different transmission
powers to constitute different cell sizes or layers, such as
macro, micro, pico and femto. HetNets can significantly
increase the spectral efficiency by manipulating these layers
in an appropriate manner [35], [36], and can expand network
coverage and improve network scalability. It can also enhance
the availability, reliability and survivability of the network,
which are the key requirements for disaster rescue scenarios.
HetNets take full advantage of the complementary character-
istics of different network tiers, and thus become an inevitable
trend for future development of information networks [37].
However, different tiers of networking access technologies
and various service requirements cause a new problem we
have to handle, i.e., spatial domain management [31], [38].
In 5G HetNets, we rely on SDN controller to collect and
manage spacial information from a variety of high and low
power nodes.
The deployment of low power nodes, namely small cell
network, as shown in Fig. 1, has been considered as an
important technology of next generation cellular networks to
improve spectrum efficiency, power efficiency, and coverage
effectively, as well as reduce the capital expenses (CapEx)
and OPEX [47], [48].
FIGURE 1. A small cell network.
With the explosive demand for mobile broadband services
and the emergence of new high capacity mobile devices,
mobile networks have to continuously evolve to meet capac-
ity and coverage demands with the latest technologies.
At the same time, there is an apparent trend of declining
2239
 S. Sun et al.: Spatial Domain Management and Massive MIMO Coordination in 5G SDN
profitability of mobile data despite the recent exponential
growth of mobile data usage. Furthermore, dense small cell
networks have attracted great attentions, where a mass of
small cell base stations (SBSs) are deployed to improve
QoS further [49].
B. MASSIVE MIMO
Massive MIMO employs the antenna arrays of a few hundred
elements to simultaneously serve many tens of mobile users
with the same time-frequency resource [39], [40]. Massive
MIMO scales up the conventional MIMO to reap all the
benefits at a higher level, which includes: 1) increasing the
capacity 10 times or more and simultaneously improving
the radiated energy efficiency in the order of 100 times;
2) inexpensive, low-power components; 3) a significant
reduction of latency on the air interface; 4) robustness against
both unintended man-made interference and intentional
jamming; 5) reducing the constraints on accuracy and lin-
earity of each individual amplifier and radio frequency (RF)
chain. Overall, massive MIMO is an enabler for future broad-
band (fixed and mobile) networks, which will be spectrum-
efficient, energy-efficient, and robust.
Massive MIMO antenna arrays generate vast amounts of
spacial domain information in real time [41], which in turn
significantly raises the computational complexity and brings
about the problem of MIMO coordination. The network can
achieve good performance, only if SDN processes spacial
information well.
C. MILLIMETER-WAVE
Millimeter-wave technology is a promising technology for
future 5G cellular systems [42]. In the core network,
fiber-based millimeter-wave systems are expected to bridge
high-capacity wireless access networks in the future [43].
In particular, millimeter-wave signals on optical carriers
are up and down linked between central stations and base
antenna stations. In radio access network, millimeter-wave
has an operating frequency between microwave and light
and owns the advantages of both. Future 5G cellular systems
can achieve super wide bandwidth from millimeter-wave, as
its frequency ranges from 26.5 to 300 GHz. Furthermore,
millimeter-wave has a much narrower antenna beam size
compared to the microwave, so that it can more precisely aim
the target [44].
The main challenge of millimeter-wave to 5G is the acqui-
sition of transmission characteristic in the air and the require-
ment of high device precision. Spatial domain management
makes the control of millimeter-wave devices more efficient
without the need to install hardware for every new service,
and thus offers a chance of low equipment cost and opera-
tional cost [45], [46].
III. REVIEW ON SDN AND OpenFlow SPECIFICATIONS
Massive MIMO HetNets lead to spatial diversity and thus
raise a critical task to retrieve, store and process the spatial
information [50], [51]. Therefore, SDN is critical for spatial
2240
domain management in massive MIMO HetNets [52], [53].
In this section, we provide an overview on the main existing
SDN standards and OpenFlow specifications.
A. MAIN EXISTING SDN STANDARDS
AND OpenFlow PROTOCOLS
The concept of SDN was first introduced in the 1990s and
became popular in the 21th century. The architecture of SDN
was formally defined by a non-profit industry consortium
called open networking forum (ONF) [54]. ONF is also
responsible for the maintenance of the OpenFlow standard
and technical specifications for the OpenFlow switch as well
as the conformance test of SDN enabled devices. In the archi-
tecture of SDN, network devices are composed of three lay-
ers, i.e., application, control and infrastructure. The functions
of two lower layers are called OpenFlow controller and Open-
Flow switch, corresponding to the control and data planes of
traditional IP/MPLS network switches and routers. Through
the OpenFlow standard, the controller instructs the OpenFlow
switch to define the standard functional messages such as
packet-received, send-packet-out, modify-forwarding-table,
and get-stats.
Apart from ONF, internet engineering task force (IETF),
international telecommunication union telecommunication
standardization sector (ITU-T), European telecommunica-
tions standards institute (ETSI) and China communications
standards association (CCSA) also started the standardiza-
tion work for SDN. IETF issued an RFC concerning the
requirements, application issues about SDN from operators’
perspective [55], while ITU-T has no formal recom-
mendation published since 2012 when the project was
launched. Table 1 summarizes and lists all formally published
SDN standards.
TABLE 1. Current main SDN standards.
Currently, SDN has found its best practice in campus net-
works and data centers and has been paid increasing attention
by more and more network organizations around the world.
For example, Fig.2 illustrates a generalized architecture of
SDN controlled hybrid cellular networks. SDN is regarded
as a promising method to solve current and emerging prob-
lems, though necessary extensions are required for future
5G HetNets.
VOLUME 3, 2015
S. Sun et al.: Spatial Domain Management and Massive MIMO Coordination in 5G SDN
FIGURE 2. A generalized architecture of SDN controlled hybrid cellular network.
It has been several years for the development of OpenFlow
specification to become an academic initiative [56], [57].
In early 2011, ONF formalized the OpenFlow specification.
A lot of the largest network operators in the world, for
example, Microsoft, Google, Yahoo, Facebook, Verizon,
Deutsche Telekom and NTT, are all directors of ONF.
It is important to note that these companies are all end users
of networking technology, rather than network equipment
manufacturers. This phenomenon indicates a strong mar-
ket interest in OpenFlow technology. Over 70 equipment
manufacturers (e.g. Cisco, IBM, Brocade, Juniper Networks)
have joined the ONF in such condition. In June 2012, ONF
released the latest version of its specification, OpenFlow 1.3.
Therefore we must pay close attention to implementing only
the approved industry standard version of these protocols to
maximize interoperability in a multivendor network and fully
realize the benefits intended by the ONF, because SDN and
OpenFlow are still evolving.
The OpenFlow standard integrates the network control
plane into software running on an attached server or net-
work controller, which makes the flow of network traffic
be controlled dynamically, without the need to rewire the
data center network. SDN concludes both industry standard
network virtualization overlays and the emerging OpenFlow
industry standard protocol. Most modern Ethernet switches
and routers contain flow tables, which run at line rate and
are used to implement functions such as QoS, security fire-
walls, and statistical analysis of data streams. OpenFlow
takes advantage of that to achieve its capabilities. OpenFlow
standardizes a common set of functions that operate on these
VOLUME 3, 2015
flows and will be extended in the future as the standard
evolves.
An OpenFlow switch consists of three parts: 1) a remote
controller on a server, 2) flow tables in the switch, and
3) a secure communication channel between them. In the
flow table, the OpenFlow protocol allows external servers
to define entries. In practice, a flow could be all the packets
from a particular MAC or IP address, a TCP connection, or all
packets with the same virtual local area network (VLAN) tag.
Each flow table entry is associated with a particular flow to
accomplish a specific action, for example, encapsulating and
forwarding the flow to a controller for processing, forwarding
the flow to a given switch port (at line rate), or dropping a flow
packet.
As illustrated in Fig.3, in the OpenFlow architecture, the
OpenFlow switch acts as a forwarding device containing
one or more flow tables. It contains an abstraction layer
that securely communicates with a controller via OpenFlow
protocol. Flow tables consist of flow entries, each of which
determines how to process and forward packets belonging to
a flow.
Typically, flow entries consist of: 1) counters, used to
collect statistics for the particular flow, such as number of
received packets, number of bytes and duration of the flow;
2) a set of instructions, or actions, to be applied upon a match;
they dictate how to handle matching packets; and 3) match
fields, or matching rules, used to match incoming packets;
match fields may contain information found in the packet
header, ingress port, and metadata. When a packet arrives at
an OpenFlow switch, packet header fields are extracted and
2241
 S. Sun et al.: Spatial Domain Management and Massive MIMO Coordination in 5G SDN

FIGURE 3. Communication between the controller and the forwarding devices happens via
OpenFlow protocol.

matched against the matching fields portion of the flow table

entries. If there is a matching entry, the switch applies the
appropriate set of instructions or actions associated with the
matched flow entry. On the other hand, the action taken by
the switch will depend on the instructions defined by the
table-miss flow entry if there isn’t a match during the flow
table look-up procedure. In order to handle table misses,
every flow table must contain a table-miss entry, which
specifies a set of actions to be performed when there is no
match for an incoming packet. On such condition, this entry
may continue the matching process on the next flow table
or forward the packet to the controller over the OpenFlow
channel. It may just drop the packet as well.
It is important to note that OpenFlow supports multiple
tables and pipeline processing from version 1.1. Another
possibility is to forward non-matching packets using regular
IP forwarding schemes in the case of hybrid switches which
have both OpenFlow and non-OpenFlow ports.
Overall, OpenFlow improves scalability, enables multite-
nancy and resource pooling in cloud computing environments
and will likely coexist with other layer 2/3 protocols and FIGURE 4. Conventional SDN architecture.
network overlays for some time. OpenFlow protocol enables
communication between controllers and switches by defining
controller that allows attached servers to control features
a set of messages which can be exchanged between these
such as packet flows, topology changes, and network
entries over a secure channel. By using the OpenFlow pro-
management [10], [58].
tocol, a remote controller can add, update and delete flow
An OpenFlow architecture, as shown in Fig.5, consists of
entries from the switch’s flow tables, which can happen
three main components: an OpenFlow-compliant switch, a
reactively or proactively.
secure channel and a controller. Switches forward packets
by means of flow tables which are lists of flow entries.
B. CONVENTIONAL SDN AND OpenFlow ARCHITECTURE Each entry consists of counters, instructions and match
As illustrated in Fig.4, SDN advocates virtualizing the fields. Firstly, the incoming packets are compared with the
data center network with a software overlay and network matching fields of each entry and then the packet is

2242 VOLUME 3, 2015

S. Sun et al.: Spatial Domain Management and Massive MIMO Coordination in 5G SDN
FIGURE 5. OpenFlow architecture.
processed according to the action contained by that entry
if there is a match. It is also possible to encapsulate and
send the packet to the controller. Counters are responsible for
keeping statistics about packets [59], [60]. The controller is
a software program following the OpenFlow protocol, which
is responsible for manipulating the switch’s flow table. The
OpenFlow-compliant switch is responsible for forwarding
packets according to the rules defined in the flow table. The
secure channel is the interface connecting the controller to
all the switches, through which the controller interacts with
the switches, including receiving and sending packets to the
switches as well as managing these switches.
C. POTENTIALS OF OpenFlow FOR 5G CORE NETWORK
OpenFlow has been one of the most commonly deployed
SDN technology [61] which standardizes the way that a
controller communicates with network devices in an SDN
architecture. OpenFlow was proposed to enable researchers
to test new ideas in a production environment and provide
a specification to migrate the control logic from a switch
into the controller. It also shows its significance in support-
ing communication between the controller and the switches
by defining an OpenFlow protocol. In an SDN architec-
ture, OpenFlow standardizes the communication between
the switches and the software-based controller [62], [63].
It doesn’t require the vendors to expose the code of their
devices when controlling a switch. The original intention
of OpenFlow was to provide a platform that would allow
researchers to run experiments in production networks.
However, industry has developed SDN and OpenFlow as a
strategy to reduce costs and hardware complexity, as well as
increasing the functionality of the network.
There are many specific capabilities in OpenFlow net-
works. For example, in OpenFlow networks, it is possible
to control multiple switches from a single controller and to
analyze traffic statistics using software. It is also feasible
to update forwarding information dynamically and manage
different types of traffic can be abstractly as flows. Research
communities have exploited these capabilities to experiment
VOLUME 3, 2015
with innovative ideas and propose new applications. In Open-
Flow area, most research focus on network management,
security, availability, network and data center virtualization,
wireless applications and ease of configuration. They have
been implemented in different environments, such as real
hardware networks and virtual simulations [64], [65].
Many specific capabilities of OpenFlow-based architec-
tures are worthy of exploitation to experiment with new
ideas and test novel applications. For example, centralized
control, flow abstraction, dynamic updating of forwarding
rules and software-based traffic analysis. OpenFlow based
applications aim to simplify network management and ease
the configuration of a network, as well as adding security
features. It also assists in virtualizing networks and data
centers and deploying mobile systems. These applications run
on top of networking operating systems such as Node. Flow,
Floodlight, Beacon, Maestro, Trema and Nox. By deploying
larger scale OpenFlow infrastructures, the research commu-
nities can run experiments and test their applications in a
more realistic scenario. Moreover, many investigations have
measured the performance of OpenFlow networks through
experimentation and modelling.
With respect to 5G core network, we next elaborate the
details of dynamic QoS support as an example of the advan-
tage of SDN paradigm.
In OpenFlow networks, it offers a new paradigm to make
routing more flexible by allowing different routing rules asso-
ciated with data flows so that the modification of partitions
of the networks layout and traffic flows can be as instant
as in an SDN. Acting as the brain of the network, the con-
troller component of OpenFlow is where routing changes
are determined. In this way, different data flows associated
with different algorithms in the controller may yield different
routing choices. The controller provides access to flow tables
as well as the rules which instruct the network forwarders
how to direct traffic flows. As illustrated in Fig. 6, the
QoS-supporting controller offers various functions and inter-
faces. In the classical Internet model, some of them have been
part of a router.
The main interfaces of a controller are Controller-
Forwarder interface and Controller-Application interface.
At Controller-Forwarder interface, the controller attaches to a
network of forwarders with a secure OpenFlow protocol. This
interface is responsible for discovering network topology,
sending flow tables associated with data flows, and receiving
traffic status information and notifications. At Controller-
Application interface, the controller provides application ser-
vice providers an secure and open (non-real time) interface to
make reservations of new data partitions (individual or groups
of data flows). It can also define new routing rules associated
with partitions if necessary. There are several key functions
under the management of the controller:
1) ROUTE MANAGEMENT
Analyzing availability and packet performance of routes to
aid the route calculation. This function requires collecting
2243
 S. Sun et al.: Spatial Domain Management and Massive MIMO Coordination in 5G SDN
FIGURE 6. The OpenFlow controller and interfaces.
performance data from the forwarders on a periodic or on a
notification basis.
2) ROUTE CALCULATION
Finding and calculating routes for different data flows.
Several routing algorithms can run in parallel to meet the
performance requirements and objectives of different flows.
Network topology and route management information are
input to this function along with service reservations.
3) TOPOLOGY MANAGEMENT
Determining and maintaining network connectivity topology
by analyzing data collected or received from forwarders.
4) SERVICE LEVEL AGREEMENT (SLA) MANAGEMENT
Providing users SLA templates for different types of
allowable data flow patterns.
5) TRAFFIC POLICING
Determining if data flows agree with their SLAs, and apply
policy rules when they do not.
6) SERVICE MANAGEMENT
Managing and storing different data flow request.
The service management function of controller works like
this: first, it receives a service reservation module (e.g. video
service) requiring QoS option from the network, then it
decides if other reservations being made can deliver the
requested service SLA. Once accepting the reservation, it
manages the network resources for the service starting at
the begin time of the data flow. When the route calculation
2244
function computes the exact route and uploads the flow table
to appropriate forwarders, the video server signals the con-
troller and data flow may start. The route management func-
tion will reactivate the route calculation function to determine
a new route for the data flow if there is congestion in the
network. It is worth noting that, in order to make sure the end
points conform to their SLAs stated in their QoS contract, the
traffic policing function must also be implemented.
IV. SDN BASED MASSIVE MIMO
COORDINATION SCHEME
In future 5G system, more effective technologies have to be
employed for spectrum utilization, traffic control, resource
allocation, density management, security, etc., so as to sup-
port the interconnection of more diversified user equip-
ment (UE) and devices. While existing SDN standards mainly
address the management of wired network such as 5G core,
5G mobile network, however, also involves a large number
of wireless technologies as another essential part, i.e., radio
access network (RAN), specially including massive MIMO
and HetNets. Correspondingly, in this paper we endeavor to
extend the function of SDN to cooperate with these edg-
ing technologies in 5G network, including Hetnets, massive
MIMO, density small cells, etc. We also emphasize the func-
tion of SDN on coping with the radio resource management
in massive MIMO HetNets.
A. A NOVEL SDN ARCHITECTURE
FOR FUTURE 5G NETWORKS
The exploitation of SDN technology shall empower the
network with intellectualization, so as to largely reduce
the complexity of 5G networks, lower the cost on network
deployment/maintenance, and facilitate the future network
evolution. The current standardizations, however, have not
taken the 5G RAN into consideration. Next, we will present
the necessary extensions to current standards and propose our
novel SDN architecture. In particular, we identify the huge
spatial domain information management and complicated
MIMO coordination as the targeted problems.
Fig. 7 illustrates our proposed architecture, where an SDN
controller is employed to manage spatial information and
conduct massive MIMO coordination for 5G networks. In this
architecture, the network consists of BS and densely deployed
small cells to support spectral efficiency and coverage. Both
BS and LPNs can be extended to have a huge number of
antennas, i.e. massive MIMO. And SDN controller is respon-
sible for the RRM of the whole Hetnet to overcome the
massive data processing constraints. Firstly, BS collects the
user information and report it to SDN. SDN controller will
perform the information processing such as CSI analysis
and null-space calculation. Then SDN controller sends the
processed results and instructions back to BS and LPNs so as
to achieve better coordination regarding the global knowledge
of network spatial information. In our proposed architecture,
SDN takes over the heavy operation from BS and LPNs to
release the efficiency of 5G networks.
VOLUME 3, 2015
S. Sun et al.: Spatial Domain Management and Massive MIMO Coordination in 5G SDN
FIGURE 7. 5G SDN architecture for spatial domain management.
FIGURE 8. Block diagram of our proposed scheme.
B. CSI ACQUIREMENT
We present in this paper a novel SDN architecture for 5G net-
works and propose our scheme to implement massive MIMO
coordination in 5G SDN. Fig. 8 illustrates the the proposed
processing diagram by highlighting two critical steps, i.e., the
acquisition of CSI for LPNs and the generation of precoding
FIGURE 9. Proposed CSI acquirement methods.
VOLUME 3, 2015
matrix based on the null-space of the victim users. We will
elaborate the details in the following.
Practically LPNs can only collect the local CSI though the
backward channel of their own users and can not access to the
CSI of the external victim users. However, owing to the hub-
spoke structure of the network, the CSI of all MIMO channels
can be collected and disseminated through the backhaul link.
Fig. 9 gives the CSI acquisition method.
BS collects user information and then reports the channel
matrix on victim users of each LPN to the SDN controller.
That is, to send SDN every [HH j,Kj +1 , ..., Hj,Kj +Lj ] corre-
H H
sponding to the j-th LPN, which contains Lj × M elements.
The SDN will perform the block-diagonal (BD) algorithm
(0)
and gain null-space vector Ṽj,i for LPN j. Then the SDN
(0)
computes on the null-space vector Ṽj,i to obtain precod-
ing matrix Tj,i (i = 1, 2...Kj ) by the precoding algorithm
proposed in the next subsection. Finally, Tj,i is sent to each
2245
 S. Sun et al.: Spatial Domain Management and Massive MIMO Coordination in 5G SDN

LPN through downlink transmission. Thus, all the matrix 1) ZF

decomposition computation and precoding methods will be We can rewrite the ZF precoding matrix as
performed by the SDN controller and it is suitable for 1  −1
the situation that LPNs are simple and devices are in low Tj,ZF = Pj H Pj Pj H , (7)
βj
performance. r
H −1 tr(Bj BHj )
H
, and βj =


C. PRECODING BASED ON NULL-SPACE
where Bj = Pj Pj Pj Ptr,j .
Obviously, the interference from the victim users can-
not affect the intended users, when the precoding vectors 2) MMSE
By the criterion of minimum mean square error, the precoding


Tj,k k = 1, 2...Kj span a subspace of the null-space of the
channel vectors from victim users. Based on this understand- vector Tj,MMSE can be expressed as
ing, the precoding matrix can be constructed by processing 1 H

Kj
−1
the massive-MIMO channel matrix. Tj,MMSE = Pj Pj Pj + H
IK , (8)
βj Ptr,j j
For the j-th LPN, the complementary space concerning the r  
victim users can be obtained by  −1 tr Bj BHj
h iH
H H K
where Bj = Pj Pj Pj + Ptr,j IK , and βj = Ptr,j .
H̃j,v = Hj,Kj +1 H · · · Hj,Kj +Lj H , (1)
3) MRT
where H̃j,v is combined by the channel matrix of all the When Ptr,j → 0, the precoding matrix can be obtained by
victim users interfered by the j-th LPN. To avoid the interfer-
ence from the victim users, Tj,i should satisfy the following 1 H
Tj,MRT = Pj , (9)
condition βj
r
H̃j,v Tj,i = 0Lj ×1 . (2) tr(Bj BHj )
where Bj = Pj and βj =
H
Ptr,j .

By SVD, H̃j,v can be further written as

D. QoS ANALYSIS
H̃j,v = Ũj,v 3̃j,v ṼH
j,v
In this section, we deduce the signal-to-interference-and-
  noise-ratio (SINR) performance for null-space based on ZF,
X H MMSE and MRT precoding.
ṽj,v,1 ṽj,v,2 · · · ṽj,v,Mj .

= Ũj,v  0 (3)
j,v Lj ×M 1) ZF
(0) Hj,k Tj,k 2
 
Define Ṽj,i = [ṽj,i,Li+1 ...ṽj,i,Mj ]. Since the column vectors SINRj,k =
(0) Kj 2
belonging to Ṽj,i locate in the null-space of all victim users,

σj,k 2 +
P
Hj,k Tj,k 0 
 
(0) 0 0
we have H̃j,v Ṽj,i = 0. k =1,k 6=k

For the k-th user of the j-th LPN, define the projection
   −1  2
Ptr,j Hj,k PH H

j P P
j j j,k 

matrix Jj,k based on the null-space of victim users by
=  .
−1 2

Kj   
(0) H
  P
(0) σj,k
  H H H
2 tr
Ṽj,v , Bj Bj + Ptr,j Hj,k Pj Pj Pj
 
Jj,k = Ṽj,v (4) 0 0


k =1 j,k 
0
k 6=k
and the new channel matrix Pj,k after the projection by
(10)
Pj,k = Hj,k Jj,k . (5)
2) MMSE
Define the channel matrix after the projection of all Kj Hj,k Tj,k 2
 
users served by the j-th LPN by SINRj,k =
Kj  2
σj,k 2 +
P
Hj,k Tj,k 0 
   
Pj,1 0 0
 Pj,2  k =1,k 6=k

 · · · , Pj ∈ C , Pj,k ∈ C 1×M .
Kj ×M
Pj =   (6)    −1 2
Ptr,j Hj,k PH H + Kj I

j P j Pj P Kj j,k 

tr,j
Pj,Kj = .
  Kj    −1 2
Kj
σj,k 2 tr Bj BH H H
P 
2
j + Ptr,j Hj,k Pj Pj Pj + P IKj
 0 
Definethe power allocation Ptr,j by Ptr,j = E[ xj ] =
0 tr,j j,k 
k =1
tr THj Tj . With Pj , we can obtain the precoding matrices
0
k 6=k
corresponding to the three fundamental linear algorithms. (11)

2246 VOLUME 3, 2015

S. Sun et al.: Spatial Domain Management and Massive MIMO Coordination in 5G SDN

3) MRT TABLE 2. Simulation parameters of Fig. 10 - Fig. 14.

Hj,k Tj,k 2
 
SINRj,k =
Kj  2
σj,k 2 +
P
Hj,k Tj,k 0 
 
0 0
k =1,k 6=k
2
 (0)

Ptr,j Hj,k HH

j,k 
=
Kj H
!
Kj  2 .
2 tr P Ṽ(0) Ṽ(0) HH H(0) + P P H(0) HH

σj,k
 
tr,j  j,k 0

j,i j,i j,i j,i
i=1 0
k =1
0
k 6 =k
To meet the QoS requirement of each user, we must have
B log 1 + SINRj,k ≥ Rj,k ,


k = 1, 2, . . . , K ; j = 1, 2, . . . , J ,
where Rj,k is the data rate demanded by user k and B is the
total bandwidth. Thus the minimum Pj,k can be solved with
the following linear equation system:
B log 1 + SINRj,k = Rj,k , k = 1, 2, . . . , K . (14)


The total power demand can be obtained by equations (10),
(11) and (12).
E. ADVANTAGES OF OUR PROPOSED ALGORITHM
Compared with the original linear precoding algorithms, the
proposed algorithm has the following advantages.
1) FEASIBLE CSI ACQUIREMENT
In traditional HetNets, the served UE reports its CSI to the
access point it associates with, but not to the other nodes inter-
fering it. Thus, an LPN doesn’t know the victim users’ CSI.
By performing our proposed algorithm, the SDN controller
can get enough information to achieve the precoding vector
to avoid the interference to the victim users. In addition,
the computational burden on LPN can be further relaxed by
letting SDN perform the null-space construction.
2) REDUCED COMPLEXITY
Unlike the existing linear precoding algorithms, the proposed
scheme can obtain a good complexity-performance trade-off.
In fact, by combining the BD algorithm and linear prediction
methods, the proposed scheme is of relatively low complexity
and good performance. Moreover, in [66], all the users receive
signals from all the LPNs, which is impossible in reality
and of high complexity in analyzing and computing. Here
we define the service area and interference area within the
LPNs’ coverage and select the LPNs that do cause measurable
interference to a specified victim user. The complexity is sig-
nificantly reduced, without considering the weak interference
from LPNs too far away.
3) IMPROVED QoS
The interference to the victim users can be totally avoided by
performing precoding based on null-space of the victim users,
VOLUME 3, 2015
j,k 
(12) TABLE 3. Channel parameters in the numerical evaluation.
(13)
for the generated precoding vectors and the channel matrix
of victim users are orthogonal to each other. At the same
time, the SINR calculation is amended to further conform
to the practice. The interference can be effectively decreased
to achieve improved QoS for users with satisfying power
consumption, applying appropriate precoding algorithm.
V. SIMULATION RESULTS
In this section, we present the numerical results to demon-
strate the performance of our proposed precoding scheme for
5G SDN. Here we use the network topology that combines
massive MIMO and small cells by implementing our algo-
rithm to achieve the target QoS with low complexity. We take
the property of null-space to alleviate the interference to the
victim users and deploy the liner precoding to mitigate the
interference among the associated users.
Fig. 10 shows the simulation scenario with intended users
marked in red and victim users marked in green. At the
same time, we evaluate the system performance over channel
realizations and different simulation parameters. The main
simulation parameters that characterize the macro cell and the
LPNs can be found in Table 2 and the main channel model
parameters are summarized in Table 3.
The channel model is similar to Case 1 for heterogeneous
deployments suggested by the 3GPP LTE standard, but the
small-scale fading is assumed to be Rayleigh according to
recent works on massive MIMO [66]. The correlation matrix
between the BS and each user is modeled according to the
physical channel model in [8, eq. (34)], where the main char-
acteristics are antenna correlation and reduced rank channels.
The propagation loss is different for BS and LPNs.
Next, Fig. 11 - Fig. 14 present the simulation results in
the the scenario as described in Fig. 10, where each user has
the rate demand of 20Mbps. The simulation results contain
both the conventional precoding scheme and the proposed
2247
 S. Sun et al.: Spatial Domain Management and Massive MIMO Coordination in 5G SDN
FIGURE 10. The overall topology. (a) The topology of the macro cell. (b) The topology of the LPN 2.
FIGURE 11. Impact of the different number of antennas.
null-space based scheme. In the proposed precoding scheme,
the BS performs the null-space construction and transmits the
corresponding part of the precoding vector to the LPNs as
described in Section IV. The BS and LPNs perform the linear
precoding separately, when using the conventional precoding
scheme.
A. POWER CONSUMPTION IMPACT BY
DIFFERENT NUMBER OF BS ANTENNAS
Fig. 11 illustrates the total power consumption impacted
by the number of massive antennas at BS and compares
the conventional ZF precoding with the proposed null-
space based ZF precoding. Massive MIMO showed a perfor-
mance improvements in energy efficiency and the increased
antennas can eliminate the interference to the victim users.
We can observe that the total power consumption can be
substantially reduced by increasing the number of antennas
at BS. Furthermore, the proposed precoding scheme has a
less power consumption than the conventional one at the same
user data rate. The reason is that the multi-antenna BS/LPNs
2248
FIGURE 12. Impact of the total users and the number of the LPNs.
can avoid the interference to the users associated to the other
LPNs but within the interference area.
B. POWER CONSUMPTION IMPACT BY
DIFFERENT NUMBER OF LPNs
Fig. 12 compares the total power consumption when the BS
is overlaid by different numbers of LPNs. The power con-
sumption decreases when the macro cell is overlaid by more
LPNs. The reason is, by performing our proposed scheme,
the LPNs can unload the burden from the BS and the extra
LPNs with multi-antennas can alleviate the interference of
the victim users. Therefore, the total power consumption
will decrease with the QoS requirements of each user
achieved.
C. UTILITY OF DIFFERENT ALGORITHMS IMPACTED
BY DIFFERENT QoS REQUIREMENTS
Fig. 13 studies the different algorithms versus the different
QoS requirements. Two proposed algorithms are involved:
1) Our proposed algorithm using the linear precoding of
VOLUME 3, 2015
S. Sun et al.: Spatial Domain Management and Massive MIMO Coordination in 5G SDN
FIGURE 13. Impact of the different QoS requirements.
FIGURE 14. Impact of the extra antennas.
the MRT. 2) Our proposed algorithm using the linear precod-
ing of the ZF. The performance of the two algorithms are also
compared to the conventional ZF precoding. Fig. 13 shows
the relationship of the QoS and the corresponding power
consumption. As a result of using the proposed precoding
scheme, the interference of the intended users from the other
BS/LPNs can be totally eliminated. This is corroborated by
Fig. 13, which illustrates that the proposed precoding scheme
consumes less power than the conventional one assuring the
same data rate.
D. SINR IMPACT BY DIFFERENT NUMBER
OF EXTRA ANTENNAS
Fig. 14 shows the SINR distribution of the conventional ZF
precoding and the proposed null-space based ZF precoding.
The conventional precoding gets worse SINR due to the
fact that their victim users experience the strong interference
from other nodes. In comparision, the proposed scheme has
an improved SINR, which results in a better system perfor-
mance. The proposed scheme uses the extra antennas to indi-
cate the victim users and project its users to the corresponding
VOLUME 3, 2015
null-space. Therefore, the strong interference to the victim
users will be eliminated and the SINR of the victim user
will increase. Fig. 14 shows the situation that the BS/LPNs
does not equip with sufficient antennas and only part of
the interference can be removed. The extra antennas can
indicate more victim users and the new system can eliminate
more interference. When the number of antennas is sufficient,
it totally eliminates the interference from other BS/LPNs.
This indicates that more extra antennas help significantly
to achieve higher SINR. In general, the proposed massive
MIMO scheme can significantly alleviate the interference and
achieve a desirable system performance for next generation
HetNets. The projection to the null-space protects the victim
users and the combined linear precoding algorithms deal with
the interference among the associated users. Compared with
the conventional design, the new system has a satisfying
power consumption and low complexity by taking advantage
of the CSI acquirement.
VI. CONCLUSION
Massive MIMO HetNet is a promising technology to improve
the capacity of 5G mobile networks, where a mobile sta-
tion can be served by multiple power nodes with different
scales of coverage range. A major challenge in 5G HetNets
is to manage and optimize spatial domain resource, com-
pared to 4G system where frequency domain has caught the
major attention. To address this issue, this work explores the
probability of utilizing SDN technology to form a central-
ized architecture of control plane. Furthermore, we propose
a lightweight CSI acquisition based null-space precoding
scheme for the implementation. Numerical simulation results
justify that our proposed system can significantly improve the
SINR of intended users in LPN-covered small cells, while
oppressing the impact on neighboring victim users.
REFERENCES
[1] R. Q. Hu and Y. Qian, ‘‘An energy efficient and spectrum efficient wireless
heterogeneous network framework for 5G systems,’’ IEEE Commun. Mag.,
vol. 52, no. 5, pp. 94–101, May 2014.
[2] M. Palkovic, P. Raghavan, M. Li, A. Dejonghe, L. Van der Perre,
and F. Catthoor, ‘‘Future software-defined radio platforms and map-
ping flows,’’ IEEE Signal Process. Mag., vol. 27, no. 2, pp. 22–33,
Mar. 2010.
[3] Y. Xu, R. Q. Hu, L. Wei, and G. Wu, ‘‘QoE-aware mobile association and
resource allocation over wireless heterogeneous networks,’’ in Proc. IEEE
Global Commun. Conf. (GLOBECOM), Dec. 2014, pp. 4695–4701.
[4] I. Chih-Lin, C. Rowell, S. Han, Z. Xu, G. Li, and Z. Pan, ‘‘Toward
green and soft: A 5G perspective,’’ IEEE Commun. Mag., vol. 52, no. 2,
pp. 66–73, Feb. 2014.
[5] R. Q. Hu, Y. Qian, S. Kota, and G. Giambene, ‘‘HetNets—A new paradigm
for increasing cellular capacity and coverage [Guest Editorial],’’ IEEE
Wireless Commun., vol. 18, no. 3, pp. 8–9, Jun. 2011.
[6] R. Q. Hu and Y. Qian, Eds., Heterogeneous Cellular Networks. New York,
NY, USA: Wiley, 2013.
[7] R. L. G. Cavalcante, S. Stanczak, M. Schubert, A. Eisenblaetter, and
U. Tuerke, ‘‘Toward energy-efficient 5G wireless communications tech-
nologies: Tools for decoupling the scaling of networks from the growth of
operating power,’’ IEEE Signal Process. Mag., vol. 31, no. 6, pp. 24–34,
Nov. 2014.
[8] H. Masutani et al., ‘‘Requirements and design of flexible NFV network
infrastructure node leveraging SDN/OpenFlow,’’ in Proc. Int. Conf. Opt.
Netw. Design Modeling, May 2014, pp. 258–263.
2249
 S. Sun et al.: Spatial Domain Management and Massive MIMO Coordination in 5G SDN
[9] B. A. A. Nunes, M. Mendonca, X.-N. Nguyen, K. Obraczka, and T. Turletti,
‘‘A survey of software-defined networking: Past, present, and future of
programmable networks,’’ IEEE Commun. Surveys Tuts., vol. 16, no. 3,
pp. 1617–1634, Third Quarter 2014.
[10] X. Duan and X. Wang, ‘‘Authentication handover and privacy protection
in 5G HetNets using software-defined networking,’’ IEEE Commun. Mag.,
vol. 53, no. 4, pp. 28–35, Apr. 2015.
[11] S. Sun, M. Kadoch, L. Gong, and B. Rong, ‘‘Integrating network function
virtualization with SDR and SDN for 4G/5G networks,’’ IEEE Netw.,
vol. 29, no. 3, pp. 54–59, May/Jun. 2015.
[12] S. Sun, B. Rong, and Y. Qian, ‘‘Artificial frequency selective channel
for covert cyclic delay diversity orthogonal frequency division multiplex-
ing transmission,’’ Secur. Commun. Netw., vol. 8, no. 9, pp. 1707–1716,
Jun. 2015.
[13] Q. Li, R. Q. Hu, Y. Qian, and G. Wu, ‘‘Cooperative communications
for wireless networks: Techniques and applications in LTE-advanced sys-
tems,’’ IEEE Wireless Commun., vol. 19, no. 2, pp. 22–29, Apr. 2012.
[14] D. Choudhury, ‘‘5G wireless and millimeter wave technology evolution:
An overview,’’ in Proc. IEEE MTT-S Int. Microw. Symp. (IMS), May 2015,
pp. 1–4.
[15] S. Han, I. Chih-Lin, Z. Xu, and C. Rowell, ‘‘Large-scale antenna systems
with hybrid analog and digital beamforming for millimeter wave 5G,’’
IEEE Commun. Mag., vol. 53, no. 1, pp. 186–194, Jan. 2015.
[16] V. Jungnickel et al., ‘‘The role of small cells, coordinated multipoint, and
massive MIMO in 5G,’’ IEEE Commun. Mag., vol. 52, no. 5, pp. 44–51,
May 2014.
[17] B. Panzner et al., ‘‘Deployment and implementation strategies for massive
MIMO in 5G,’’ in Proc. Globecom Workshops (GC Wkshps), Dec. 2014,
pp. 346–351.
[18] T. R. Omar, A. E. Kamal, and J. M. Chang, ‘‘Downlink spectrum allo-
cation in 5G HetNets,’’ in Proc. Int. Wireless Commun. Mobile Comput.
Conf. (IWCMC), Aug. 2014, pp. 12–17.
[19] D. Liu, L. Wang, Y. Chen, T. Zhang, K. Chai, and M. Elkashlan, ‘‘Dis-
tributed energy efficient fair user association in massive MIMO enabled
HetNets,’’ IEEE Commun. Lett., vol. 19, no. 10, pp. 1770–1773, Oct. 2015.
[20] C.-F. Lai, R.-H. Hwang, H.-C. Chao, M. Hassan, and A. Alamri,
‘‘A buffer-aware HTTP live streaming approach for SDN-enabled 5G
wireless networks,’’ IEEE Netw., vol. 29, no. 1, pp. 49–55, Jan./Feb. 2015.
[21] P. Ameigeiras, J. J. Ramos-Munoz, L. Schumacher, J. Prados-Garzon,
J. Navarro-Ortiz, and J. M. Lopez-Soler, ‘‘Link-level access cloud archi-
tecture design based on SDN for 5G networks,’’ IEEE Netw., vol. 29, no. 2,
pp. 24–31, Mar./Apr. 2015.
[22] S. Talwar, D. Choudhury, K. Dimou, E. Aryafar, B. Bangerter, and
K. Stewart, ‘‘Enabling technologies and architectures for 5G wireless,’’ in
Proc. IEEE MTT-S Int. Microw. Symp. (IMS), Jun. 2014, pp. 1–4.
[23] Y. Mehmood, W. Afzal, F. Ahmad, U. Younas, I. Rashid, and I. Mehmood,
‘‘Large scaled multi-user MIMO system so called massive MIMO systems
for future wireless communication networks,’’ in Proc. IEEE 11th Int.
Conf. Autom. Comput. (ICAC), Sep. 2013, pp. 1–4.
[24] E. Larsson, O. Edfors, F. Tufvesson, and T. Marzetta, ‘‘Massive MIMO for
next generation wireless systems,’’ IEEE Commun. Mag., vol. 52, no. 2,
pp. 186–195, Feb. 2014.
[25] T. L. Narasimhan, P. Raviteja, and A. Chockalingam, ‘‘Large-scale mul-
tiuser SM-MIMO versus massive MIMO,’’ in Proc. Inf. Theory Appl.
Workshop (ITA), Feb. 2014, pp. 1–9.
[26] Z. Li, D. Grace, and P. Mitchell, ‘‘Cell division, migration and death
for energy efficient 5G ultra-small cell networks,’’ in Proc. Globecom
Workshops (GC Wkshps), Dec. 2014, pp. 942–947.
[27] R. Guerzoni, I. Vaishnavi, A. Frimpong, and R. Trivisonno, ‘‘Virtual link
mapping for delay critical services in SDN-enabled 5G networks,’’ in Proc.
1st IEEE Conf. Netw. Softwarization (NetSoft), Apr. 2015, pp. 1–9.
[28] H.-H. Cho, C.-F. Lai, T. K. Shih, and H.-C. Chao, ‘‘Integration of SDR and
SDN for 5G,’’ IEEE Access, vol. 2, pp. 1196–1204, Sep. 2014.
[29] W. Liu, S. Han, C. Yang, and C. Sun, ‘‘Massive MIMO or small cell
network: Who is more energy efficient?’’ in Proc. IEEE Wireless Commun.
Netw. Conf. Workshops (WCNCW), Apr. 2013, pp. 24–29.
[30] W. Liu, S. Han, and C. Yang, ‘‘Energy efficiency comparison of massive
MIMO and small cell network,’’ in Proc. IEEE Global Conf. Signal Inf.
Process. (GlobalSIP), Dec. 2014, pp. 617–621.
[31] J. J. V. Olmos and I. T. Monroy, ‘‘Millimeter-wave wireless links
for 5G mobile networks,’’ in Proc. 17th Int. Conf. Transparent Opt.
Netw. (ICTON), Jul. 2015, p. 1.
2250
[32] H. Mehrpouyan, M. Matthaiou, R. Wang, G. K. Karagiannidis, and Y. Hua,
‘‘Hybrid millimeter-wave systems: A novel paradigm for HetNets,’’ IEEE
Commun. Mag., vol. 53, no. 1, pp. 216–221, Jan. 2015.
[33] R. Guerzoni, R. Trivisonno, and D. Soldani, ‘‘SDN-based architecture
and procedures for 5G networks,’’ in Proc. 1st Int. Conf. 5G Ubiquitous
Connectivity (5GU), Nov. 2014, pp. 209–214.
[34] N. Cvijetic, ‘‘Optical network evolution for 5G mobile applications and
SDN-based control,’’ in Proc. 16th Int. Telecommun. Netw. Strategy
Planning Symp. (Networks), Sep. 2014, pp. 1–5.
[35] X. Chen, R. Q. Hu, and Y. Qian, ‘‘Distributed resource and power
allocation for device-to-device communications underlaying cellular net-
work,’’ in Proc. IEEE Global Commun. Conf. (GLOBECOM), Dec. 2014,
pp. 4947–4952.
[36] Z. Zhang, R. Q. Hu, Y. Qian, A. Papathanassiou, and G. Wu, ‘‘D2D
communication underlay uplink cellular network with fractional frequency
reuse,’’ in Proc. 11th Int. Conf. Design Rel. Commun. Netw. (DRCN),
Mar. 2015, pp. 247–250.
[37] T. Djerafi, O. Kramer, N. Ghassemi, A. B. Guntupalli,
B. Youzkatli-El-Khatib, and K. Wu, ‘‘Innovative multilayered millimetre-
wave antennas for multi-dimensional scanning and very small footprint
applications,’’ in Proc. 6th Eur. Conf. Antennas Propag. (EUCAP),
Mar. 2012, pp. 2583–2587.
[38] S. Scott-Hayward and E. Garcia-Palacios, ‘‘Multimedia resource alloca-
tion in mmwave 5G networks,’’ IEEE Commun. Mag., vol. 53, no. 1,
pp. 240–247, Jan. 2015.
[39] M. Haugh and R. Milford, ‘‘Interoperability event technical issues report,’’
Open Netw. Found., Tech. Rep. ONF TR-501, Jun. 2013.
[40] Software-Defined Networking: A Perspective From Within a Service
Provider Environment, Internet Eng. Task Force, Fremont, CA, USA,
Mar. 2014.
[41] L. L. Wei, R. Q. Hu, T. He, and Y. Qian, ‘‘Device-to-device (D2D) com-
munications underlaying MU-MIMO cellular networks,’’ in Proc. IEEE
GlOBECOM, Dec. 2013, pp. 4902–4907.
[42] A. Dixit, F. Hao, S. Mukherjee, T. V. Lakshman, and R. Kompella,
‘‘Towards an elastic distributed SDN controller,’’ ACM SIGCOMM
Comput. Commun. Rev., vol. 43, no. 4, pp. 7–12, Oct. 2013.
[43] P. Bosshart et al., ‘‘Forwarding metamorphosis: Fast programmable match-
action processing in hardware for SDN,’’ ACM SIGCOMM Comput.
Commun. Rev., vol. 43, no. 4, pp. 99–110, Oct. 2013.
[44] M.-K. Shin, K.-H. Nam, and H.-J. Kim, ‘‘Software-defined
networking (SDN): A reference architecture and open APIs,’’ in Proc. Int.
Conf. ICT Converg. (ICTC), Oct. 2012, pp. 360–361.
[45] S. Schmid and J. Suomela, ‘‘Exploiting locality in distributed SDN con-
trol,’’ in Proc. 2nd ACM SIGCOMM Workshop Hot Topics Softw. Defined
Netw., Aug. 2013, pp. 121–126.
[46] S. Scott-Hayward, G. O’Callaghan, and S. Sezer, ‘‘SDN security:
A survey,’’ in Proc. IEEE SDN Future Netw. Services, Nov. 2013, pp. 1–7.
[47] S. Salsano, N. Blefari-Melazzi, A. Detti, G. Morabito, and L. Veltri,
‘‘Information centric networking over SDN and OpenFlow: Architectural
aspects and experiments on the OFELIA testbed,’’ Comput. Netw., vol. 57,
no. 16, pp. 3207–3221, 2013.
[48] A. Shalimov, D. Zuikov, D. Zimarina, V. Pashkov, and R. Smeliansky,
‘‘Advanced study of SDN/OpenFlow controllers,’’ in Proc. 9th Central
Eastern Eur. Softw. Eng. Conf. Russia, Oct. 2013, Art. ID 1.
[49] N. Cvijetic, A. Tanaka, P. N. Ji, K. Sethuraman, S. Murakami, and
T. Wang, ‘‘SDN and OpenFlow for dynamic flex-grid optical access and
aggregation networks,’’ J. Lightw. Technol., vol. 32, no. 4, pp. 864–870,
Feb. 15, 2014.
[50] H. Ali-Ahmad et al., ‘‘An SDN-based network architecture for extremely
dense wireless networks,’’ in Proc. IEEE SDN Future Netw. Services,
Nov. 2013, pp. 1–7.
[51] K. Bakshi, ‘‘Considerations for software defined networking (SDN):
Approaches and use cases,’’ in Proc. IEEE Aerosp. Conf., Mar. 2013,
pp. 1–9.
[52] D. Hock, M. Hartmann, S. Gebert, M. Jarschel, T. Zinner, and
P. Tran-Gia, ‘‘Pareto-optimal resilient controller placement in SDN-based
core networks,’’ in Proc. 25th Int. Teletraffic Congr. (ITC), Sep. 2014,
pp. 1–9.
[53] P. Porras, S. Shin, V. Yegneswaran, M. Fong, M. Tyson, and G. Gu,
‘‘A security enforcement kernel for OpenFlow networks,’’ in Proc. 1st
Workshop Hot Topics Softw. Defined Netw., Aug. 2012, pp. 121–126.
[54] C. Rotsos, N. Sarrar, S. Uhlig, R. Sherwood, and A. W. Moore, ‘‘OFLOPS:
An open framework for OpenFlow switch evaluation,’’ in Proc. 13th Int.
Conf. Passive Active Meas., Jan. 2012, pp. 85–95.
VOLUME 3, 2015
S. Sun et al.: Spatial Domain Management and Massive MIMO Coordination in 5G SDN
[55] J. H. Jafarian, E. Al-Shaer, and Q. Duan, ‘‘OpenFlow random host muta-
tion: Transparent moving target defense using software defined network-
ing,’’ in Proc. 1st Workshop Hot Topics Softw. Defined Netw., Aug. 2012,
pp. 127–132.
[56] X. Xu, H. Zhang, X. Dai, Y. Hou, X. Tao, and P. Zhang, ‘‘SDN based
next generation mobile network with service slicing and trials,’’ China
Commun., vol. 11, no. 2, pp. 65–77, Feb. 2014.
[57] J. Sanchez, I. G. Ben Yahia, N. Crespi, T. Rasheed, and D. Siracusa,
‘‘Softwarized 5G networks resiliency with self-healing,’’ in Proc. 1st Int.
Conf. 5G Ubiquitous Connectivity (5GU), Nov. 2014, pp. 229–233.
[58] X. Li and H. Zhang, ‘‘Creating logical zones for hierarchical traffic engi-
neering optimization in SDN-empowered 5G,’’ in Proc. Int. Conf. Comput.,
Netw. Commun. (ICNC), Feb. 2015, pp. 1071–1075.
[59] S. Kuklinski, Y. Li, and K. T. Dinh, ‘‘Handover management in SDN-
based mobile networks,’’ in Proc. Globecom Workshops (GC Wkshps),
Dec. 2014, pp. 194–200.
[60] V. Jungnickel et al., ‘‘Software-defined open architecture for front- and
backhaul in 5G mobile networks,’’ in Proc. 16th Int. Conf. Transparent
Opt. Netw. (ICTON), Jul. 2014, pp. 1–4.
[61] L. Liu, R. Muñoz, R. Casellas, T. Tsuritani, R. Martínez, and I. Morita,
‘‘OpenSlice: An OpenFlow-based control plane for spectrum sliced elastic
optical path networks,’’ Opt. Exp., vol. 21, no. 4, pp. 4194–4204, 2013.
[62] A. Lara, A. Kolasani, and B. Ramamurthy, ‘‘Network innovation using
OpenFlow: A survey,’’ IEEE Commun. Surveys Tuts., vol. 16, no. 1,
pp. 493–512, First Quarter 2014.
[63] L. Liu et al., ‘‘First field trial of an OpenFlow-based unified control plane
for multi-layer multi-granularity optical networks,’’ in Proc. Opt. Fiber
Commun. Conf. Expo., Nat. Fiber Opt. Eng. Conf., Mar. 2012, pp. 1–3,
paper PDP5D-2.
[64] T. Luo, H.-P. Tan, and T. Q. S. Quek, ‘‘Sensor OpenFlow: Enabling
software-defined wireless sensor networks,’’ IEEE Commun. Lett., vol. 16,
no. 11, pp. 1896–1899, Nov. 2012.
[65] M. Channegowda et al., ‘‘Experimental demonstration of an OpenFlow
based software-defined optical network employing packet, fixed and flex-
ible DWDM grid technologies on an international multi-domain testbed,’’
Opt. Exp., vol. 21, no. 5, pp. 5487–5498, 2013.
[66] Further Advancements for E-UTRA Physical Layer Aspects (Release 9),
document 3GPP TS 36.814, Mar. 2010.
SONGLIN SUN received the B.S. and M.S.
degrees from the Shandong University of Tech-
nology, in 1997 and 2000, respectively, and the
Ph.D. degree from the Beijing University of Posts
and Telecommunications, in 2003. He is currently
an Associate Professor with the Beijing University
of Posts and Telecommunications. He has long
been engaged in the research and development and
teaching work on wireless communications, multi-
media communication, the Internet of Things, and
embedded systems.
BO RONG (M’07) received the B.S. degree from
Shandong University, Jinan, China, in 1993, the
M.S. degree from the Beijing University of Aero-
nautics and Astronautics, Beijing, China, in 1997,
and the Ph.D. degree from the Beijing University
of Posts and Telecommunications, in 2001.
He is currently a Research Scientist with
the Communications Research Centre Canada,
Ottawa, ON. He is also an Adjunct Professor with
the Ecole de technologie superieure, Universite du
Quebec, Canada. He has authored or co-authored over 100 technical papers in
major journals and conferences, and two book chapters in wireless network-
ing and communications. Many of these publications have theoretical and
practical significance to the research community and industry. His research
interests include modeling, simulation, and performance analysis of next-
generation wireless networks. He is a member of the IEEE Communications
Society and the IEEE Broadcasting Society. He serves as an Associate Editor
of the IEEE COMMUNICATIONS LETTERS and a Guest Editor of special issues of
the IEEE Communications Magazine, the IEEE Wireless Communications
Magazine, and the IEEE INTERNET OF THINGS JOURNAL.
VOLUME 3, 2015
ROSE QINGYANG HU (S’95–M’98–SM’06)
received the B.S. degree from the University of
Science and Technology of China, the M.S. degree
from the Polytechnic Institute of New York Uni-
versity, and the Ph.D. degree from the University
of Kansas. She has over ten years of research and
development experience with Nortel, Blackberry,
and Intel as a Technical Manager, Senior Wire-
less System Architect, and Senior Research Sci-
entist, actively participating in industrial 3G/4G
technology development, standardization, system level simulation, and per-
formance evaluation. She is currently an Associate Professor with the
Electrical and Computer Engineering Department, Utah State University.
She has authored extensively and holds numerous patents in her research
areas. Her current research interests include next-generation wireless com-
munications, wireless system design and optimization, green radios, mul-
timedia Quality of Service/Quality of Experience, communication and
information security, wireless system modeling, and performance anal-
ysis. She is a member of the Phi Kappa Phi Honor Society and the
Epsilon Pi Epsilon Honor Society. She was a recipient of best paper
awards from the IEEE Globecom 2012 and IEEE ICC 2015. She is an
IEEE Communications Society Distinguished Lecturer Class 2015–2016.
She serves on the Editorial Boards of the IEEE Wireless Communications
Magazine, the IEEE COMMUNICATIONS SURVEYS AND TUTORIALS, the IEEE
INTERNET OF THINGS JOURNAL, the Security and Communication Networks
Journal, Wireless Communications and Mobile Computing, and KSII Trans-
actions on Internet and Information Systems. She has been a seven-time
Guest Editor of the IEEE Communications Magazine, the IEEE Wireless
Communications Magazine, and the IEEE Network Magazine. She served
as the TPC Co-Chair of ICNC 2014, the TPC Vice Chair of the IEEE Green-
com 2013, and the Symposium Co-Chair of the IEEE ICC 2012/2014/2015,
the IEEE WCNC 2013, ICNC 2013, and the IEEE SmartgridComm 2012.
YI QIAN (M’95–SM’07) received the Ph.D.
degree in electrical engineering from Clemson
University, Clemson, SC, USA.
He was with the telecommunications industry,
academia, and the government. Some of his pre-
vious professional positions include serving as a
Senior Member of the Scientific Staff and Techni-
cal Advisor with Nortel Networks, a Senior Sys-
tems Engineer and Technical Advisor with several
start-up companies, an Assistant Professor with
the University of Puerto Rico at Mayaguez, and a Senior Researcher with the
National Institute of Standards and Technology. He is currently an Associate
Professor with the Department of Electrical and Computer Engineering,
University of Nebraska–Lincoln. His research interests include information
assurance and network security, network design, network modeling, sim-
ulation and performance analysis for next generation wireless networks,
wireless ad-hoc and sensor networks, vehicular networks, smart grid com-
munication networks, broadband satellite networks, optical networks, high-
speed networks, and the Internet. He is a member of ACM. He serves on the
Editorial Board of several international journals and magazines, including
serving as the Associate Editor-in-Chief of the IEEE Wireless Communi-
cations Magazine. He is the Chair of the IEEE Communications Society
Technical Committee for Communications and Information Security. He is
a Distinguished Lecturer of the IEEE Vehicular Technology Society.
2251