Cost-efficient Heterogeneous Data Transmission in
Software Defined Vehicular Networks
Zongjian He, Daqiang Zhang
School of Software Engineering
Tongji University
Shanghai, China
E-mail: {hezongjian, dqzhang}@tongji.edu.cn
Junbin Liang
School of Computer and Electronic Information
Guangxi University
Nanning, Guangxi, China
E-mail: liangjb@gxu.edu.cn
Abstract—Vehicular networks have been regarded as the key
enabling technology of the future smart vehicles. Existing vehic-
ular communications mainly rely on a single network instance
or utilize data offloading over two networks (e.g., IEEE 802.11p
and cellular network). However, today’s vehicular network imple-
mentations are highly heterogeneous. Conventional homogeneous
communication and data offloading may not be able to satisfy the
requirement of the emerging vehicular networking applications.
In this research, we apply software defined network (SDN)
to the heterogeneous vehicular networks to bridge the gap.
With SDN, heterogeneous network resources can be managed
with unified abstraction. Moreover, we propose an SDN-based
wireless communication solution, which can schedule different
network resources to minimize communication cost. The problem
is formulated as an optimization problem, and two different
solutions have been designed to adapt divergent application
scenarios. We evaluate the proposed approaches using traffic
traces. The effectiveness and efficiency are validated by the
results.
Index Terms—Handoff, VANET, Software defined network,
SDN, optimization
I. INTRODUCTION
With the advancement of information communication and
technology, more and more things in people’s daily lives are
becoming smart. Smart phones, smart TVs and smart watches
are growing rapidly and have significantly changed people’s
life style. Besides, vehicles have been considered to be the next
impactful smart thing that can also potentially exert influence
on everyone [1]. To enable a smart vehicles, it is essential for
a vehicle to access Internet and communicate with ambient
neighbors through wireless communication. Thus, vehicular
networks are fundamentally important and have attracted much
attention from both industry and academy in recent years.
Vehicular network aims to connect vehicles to the cloud,
roadside units (RSU), and ambient vehicles using wireless
communications. Compared with smartphone based or smart
TV based networks, the smart vehicle based wireless networks
are more complicated and heterogeneous. Currently, many
vehicles have already be able to access Internet via cellular
networks like GPRS / EDGE / UMTS / LTE. Meanwhile,
IEEE has released standard 802.11p [2] for vehicular ad-
hoc network (VANET) that can establish vehicle-to-vehicle
(V2V) and vehicle-to-infrastructure (V2I) connections using
dedicated spectrum. In addition, other wireless technologies
like IEEE 802.11 ac/ad, ZigBee [3], Bluetooth [4], RFID
[5] and WiMAX [6], have also been applied as alternative
technologies for vehicular communication. What is more, it is
quite often for a vehicle to be equipped with multiple wireless
interfaces simultaneously (e.g., Cellular network and DSRC).
Communication over such a heterogeneous network poses
a number of challenges that need to be properly consid-
ered. First, the heterogeneity of these wireless technologies
makes the interconnection and interoperation an intractable
task, which leads to network fragmentation and inefficiency
of network resource utilization. Moreover, different wireless
technologies have different performance and communication
cost, and applications also have a variety of requirements for
network resources. Selecting one or more arbitrary network
interfaces for communication may not be able to satisfy
the requirement of application. Instead, the requirement and
network resources must be mapped carefully. In summary, two
problems are caused by the heterogeneity: 1) interoperability.
2) resource provisioning and utilization.
Researchers have already proposed many vehicular network-
ing applications, which range from safety related ones to
those aiming at improving driving efficiency. Most of these
applications only rely on a single type of wireless network.
For example, safety applications usually utilize IEEE 802.11p
for V2V communications, and infotainment applications often
access the Internet via cellular networks. In order to improve
the network performance and reduce the communication cost,
mobile data offloading [7] has been employed to vehicular
communication. In these scenarios, WLAN or other metro-
scale networks are generally used as a complementary network
to deliver data that originally leverage cellular networks [8]
[9]. Additionally, handoff among homogeneous network [10]
and heterogeneous network [11] [12] have also been proposed
to improve the network availability. These researches have
remarkably thrived the evolution of vehicular networks. How-
ever, imperfections remain for vehicular network to support
a wide range of real world applications. Foremost, statically
assigned data offloading and handoff are not flexible enough,
especially when the vehicles are moving fast and the external
environment changes rapidly. Moreover, current approaches
are also lack of unified abstractions. Consequently, existing
approaches may not be able to fully leverage the underlying