See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/332589876

SDN aided Mobility Management for Connected Vehicle Networks

Conference Paper · April 2019

CITATIONS READS
0 17

3 authors:

Kuljaree Tantayakul Riadh Dhaou

Prince of Songkla University Ecole Nationale Supérieure d’Electrotechnique, d’Electronique, d’Informatique, d’H…
7 PUBLICATIONS   13 CITATIONS    76 PUBLICATIONS   347 CITATIONS   

SEE PROFILE SEE PROFILE

Béatrice Paillassa
Ecole Nationale Supérieure d’Electrotechnique, d’Electronique, d’Informatique, d’H…
45 PUBLICATIONS   208 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

UAV, M2M, Hybrid Satellite-Terrstrial Networks, SDN, View project

All content following this page was uploaded by Riadh Dhaou on 23 April 2019.

The user has requested enhancement of the downloaded file.

 SDN aided Mobility Management for Connected
Vehicle Networks
Kuljaree Tantayakul∗ , Riadh Dhaou† , Beatrice Paillassa†
∗ Collegeof Computing, Prince of Songkla University, Phuket, Thailand
† IRIT-ENSEEIHT, University of Toulouse, France
kuljaree.t@phuket.psu.ac.th, {Riadh.Dhaou, Beatrice.Paillassa}@enseeiht.fr
Abstract—Connected Vehicle is provided the Internet access
with built-in modems providing onboard 4G LTE and allows
to share Wi-Fi Internet access with other mobile devices both
internal and external vehicle. The large volume of data is
possibly transmitted over the Internet through the connected
services or infotainment service by passengers. From the origin
to destination on the road, the connected vehicle may frequently
change its point attachment with high speed, leading to the
large number of packet loss impact on quality of experience
(QoE). To address this problem, in this paper, we propose a new
mobility management approach with the benefits of Software-
Defined Networking (SDN) for prediction and pre-uploading the
data to next Road Side Unit (RSU) in vehicular context.
Index Terms—Connected Vehicle, Software-defined Network.
I. I NTRODUCTION
Connected vehicle (CV) is one area of particular interest
to investors from the automotive industry perspective that
companies are pouring budget into the Internet of Things (IoT)
connected vehicles. The first, the connected vehicles were
designed for a safety service, and only voice is enabled to
contact the emergency call center when an accident occurs.
After the capacities of connected vehicles have been increased,
for the localization with Global Positioning System (GPS) and
the ability to transfer voice and data at the same time by sup-
porting 4G LTE. The connected vehicles are driven based on
the roadmap from integrated navigation which supports online
services to respond to a driver for intelligent routing, traffic
alerts, fuel stations, existing attractions, etc. Moreover, many
infotainment services are provided for the passengers since
Wi-Fi hotspot can be enabled in connected vehicles. While
the passengers are enjoying the entertaining program along
with their trip, it is possible that the connected vehicles may
change to the other point of attachment, leading to a possible
disconnection and disturbed services that makes the passengers
unhappy. In this paper, we address this problem by proposing a
new possible approach for providing mobility management in
vehicular networks with the leverage of Software-Defined Net-
working (SDN) and Network Function Virtualization (NFV).
SDN is centralized network management. SDN contributes to
the creation of network automation that enables a policy-based
decision. As NFV focuses on optimizing the network services
for flexibility, scalability and cost-effective, leading to fast and
straightforward to create new services.
The increasing of connected vehicles and the increasing
demand of video streaming, refer to the large data will be
exchanged over the Internet while the connected vehicles are
running on the road. Even though the cloud service was
designed to provide the hardware and system software in the
data center over the Internet, it becomes full-grown from the
massive scale of vehicles network. To satisfy the quality of
service (QoS) requirements of vehicle services with seamless
mobility, in this paper proposes SDN with a caching policy
in the local Wi-Fi network. In the other side, the operator
requirement is to decreases the operating cost. Therefore
in this paper considers the suitable distance between RSUs
coverage to decrease the number of RSU devices without
impacting the Quality of Experience (QoE).
The rest of this paper is organized as follows: the related
works are introduced in Section II, SDN Vehicle Mobility
Management is our proposed approach which is described in
the next section. Then, the results of performance analysis on
experimental topology are shown in Section IV and Section
V. Finally, the conclusion and future works will be presented
in Section VI.
II. R ELATED W ORKS
A. Existing Handover Approaches in Vehicular Network
Currently, the standard of mobility management is essential
for providing fast and seamless mobility for vehicular net-
works, but it is not defined yet. So, several ways to handle
mobility are designed based on Internet mobility management
protocol. There are two ways to handle vehicle mobility by
applying the IP mobility protocol: host mobility and network
mobility (NEMO).
An enhanced PFMIPv6 (ePFMIPv6) [1][2] and Vehicular
Fast handover for Mobility IPv6 (VFMIPv6) [3] were pro-
posed to apply the existing mobility protocols: Fast handover
for Mobility IPv6 (FMIPv6) and Fast Handover for PMIPv6
(PFMIPv6) for mobility management in a vehicular network.
Both schemes were proposed for improving the quality of
services (QoS) regarding the handover latency, packet loss,
and the signaling cost.
NEMO-enabled PMIPv6 (N-PMIPv6) [4] have extended
the PMIPv6 protocol to provide network mobility, but it
experiences large tunneling overhead since it uses multiple
encapsulations. Then, network mobility support in PMIPv6
network (N-NEMO) [5] was proposed to enhance the effi-
ciency of N-PMIPv6 and to reduce tunneling overhead by
splitting the tunnel to the mobile router (MR). But N-NEMO

Fig. 1. The SDNVMM architecture for connected vehicles.
suffers from high overhead. So PNEMO [6] was proposed to
minimize this problem by creating only one tunnel between
Location Management Agent (LMA ) and MAGs.
Even though an ePFMIPv6, N-PMIPv6, N-NEMO, and
PNEMO do not require the mobility functionality at the mobile
terminal, they still suffer from the above problem with the tun-
neling establishment. These schemes also need to implement
the mobility protocol on the network components, leading to
inflexible management for the heterogeneous network.
B. The Alternative Mobility Management Approaches
Most of the schemes in the previous section were designed
based on PMIPv6 which is the network-based mobility that
does not require software implementation at the user side. But
they are worst regarding tunneling overhead. Therefore, SDN-
Mobility [7] was proposed to use SDN for providing mobility
management like as PMIPv6 without any legacy mobility
protocol implementation and also gains a higher performance
than PMIPv6 regarding the throughput, handover latency, and
avoids tunneling establishment overhead. But, it still suffers
from the data packet loss during MN handover. So, [8]
addressed this loss problem by proposing two caching policies
(ON-OFF and Adaptive caching policy) that minimized the
loss and also improved the quality of transfer in both user and
operator aspects.
However, in both [7] and [8] focused on the pedestrian
mobility model that is quite different with vehicle mobility
model regarding velocity and rapid handover.
III. P ROPOSED A PPROACH
In this section, we propose a new mobility management
approach in a vehicular network which is called SDN Vehicle
Mobility Management (SDNVMM). SDNVMM was designed
based on SDN-Mobility [7] for updating the routing path
during vehicle handover. Besides, SDNVMM was intended
to take the benefits of SDN for handover prediction and pre-
uploading the data to next Road Side Unit (RSU), leading
improve QoS regarding loss.
A. SDNVMM Architecture
Fig. 1 illustrates the SDNVMM architecture for connected
vehicles. It can be separated into three main layers: applica-
tion, Internet backbone, and access infrastructure.
1) Application Layer: Application layer provides connected
services such as intelligent navigation, traffic management, the
infotainment service, etc.
2) Internet Backbone Layer: The interconnection of many
large networks is called Internet backbone. Many network
service providers or NSPs are connected in Internet backbone
to provide the Internet for their customers. Several servers are
located on the Internet backbone such as web servers, mail
servers, media servers, etc.
3) Access Infrastructure Layer: An access infrastructure
layer consists of three main components: Road Side Units
(RSUs), SDN controller and Caching Policy.
• Road Side Units (RSUs)
Road Side Units are installed on the roadside to provide
vehicles connectivity. The RSUs can be equipped with the
wireless transceiver operating on either Dedicated Short-Range
Communication (DSRC) or Wi-Fi [9].
The general access infrastructure for connected vehicles
consists of eNBs and RSUs. The eNBs provide LTE commu-
nication that covers all city areas. The link communication
between connected vehicles and eNBs is more stable than
link communication with RSUs. Then, the LTE connectivity
is recommended for the time-sensitive services such as safety
services, intelligent navigation, and voice data. Video data or
streaming data is recommended to be transferred through Wi-
Fi link because Wi-Fi provides high data rate communication
[10].
In this paper, we focus on the infotainment application such
as video on demand and live streaming video. So RSUs based
on Wi-Fi technology and support the OpenFlow protocol. They
also support caching function and coordinate with OpenFlow
command messages with the SDN controller. In the following,
we distinguish the Previous Road Side Unit (P-RSU) from the
New Road Side Unit (N-RSU).
• SDN Controller
The SDN Controller is an essential component which sup-
ports the OpenFlow protocol. It is located on the same network
as RSUs and acts as the brain of the vehicular system that
can automatically and dynamically manage and control the
network components in the vehicular network. Because the
SDN controller can monitor the information of connected
vehicles such as origin, destination, the road path, vehicle
position, it also knows the network information such as the
load of each RSU, the bandwidth usage of each RSU, the
location of RSU, etc.
In this paper, the main function of an SDN controller can
be divided into two parts according to the purpose of the task:
routing path management and cache management.
The SDN controller automatically updates the routing path
when the connected vehicle drives in SDN vehicular network

Fig. 2. The SDNVMM Operation.
and when the connected vehicle changes to another RSU
network.
For cache management, the SDN controller uses the Open-
Flow protocol to enable the cache mode, configure and manage
the cached data on the RSUs. These processes show the ability
of SDN to easily and flexibly enable the network function
services.
After the cache rules are configured in the RSUs, RSUs
can decide to cache the CV data by themselves under caching
policy. This is an edge computing which can minimize the
number of signaling and share the load with SDN controller.
• Caching Policy
Caching Policy contains the defined rules for caching the
requested content of each connected vehicle. It can be installed
either on the SDN controller or the other servers. SDN
controller can get the rule of each connected vehicle from
caching policy through Application Programming Interface
(API).
We apply an ON-OFF caching policy [8] which only
transfers the data while the connected vehicle lives in the
good signaling area. Even though the experimental results
of ON-OFF caching policy with walk mobility model are
worst regarding total transmission time, the ON-OFF caching
policy gains a better jitter delay which suitable for in-vehicle
communication and infotainment services that’s why an ON-
OFF caching policy is chosen in this work.
B. SDNVMM Operation
The SDNVMM operation can be separated in two proce-
dures: registration procedure and pre-uploading data procedure
as illustrates in Fig. 2.
• Registration Procedure
When the engine of the connected vehicle is started, the
connected vehicle requires an IPv6 address by sending the
router solicitation (RS) message to the closest RSU (P-RSU).
After that, the P-RSU informs the CV attachment event to
the SDN controller to authenticate the CV permission with
the network/cache policy. In this case, assume that CV has
been defined in the network policy. Then, the CV routing path
will be added in all RSUs, and the caching requirement of the
CV will be configured at the P-RSU by the SDN controller.
Once the connected vehicle receives the router advertisement
(RA) message, an IPv6 address is assigned by IPv6 auto-
configuration. The connected vehicle can access the Internet to
update its information or to access the infotainment services.
While the connected vehicle drives on the road far away
from P-RSU, the bandwidth capacity of CV is changed,
leading to enabling the cache trigger. The P-RSU starts to
cache the CV data under the caching policy and informs the
caching event to the SDN controller. In this paper, we apply
an ON-OFF caching policy that will cache and stop to forward
the CV data after the cached trigger is enabled.
• Pre-uploading Data Procedure
The data pre-uploading procedure occurs after the caching
trigger is enabled. Meanwhile, the connected vehicle keeps
driving on the roadmap with the GPS navigation. The CV po-
sition can be known by GPS location, and the SDN controller
localizes the each RSU. Therefore the SDN controller can
predict the next RSU which the connected vehicle will attach.
Then, the routing path of the CV and the cached requirement
can be updated at the N-RSU before the connected vehicle
handover. Also, the cached data at the P-RSU will be offloaded
to N-RSU before the connected vehicle handover.
Then, the connected vehicle attaches to the new network;
the CV data is continually transferred directly from N-RSU.
IV. E XPERIMENTAL N ETWORK T OPOLOGY
We set up an experimental network topology to compare
the performance of SDNVMM and SDN-Mobility, that can
be separated into two aspects: user and operator. From origin
to destination, it is possible that a vehicle will be driven
through the road path without Internet connectivity. Maybe
some RSUs are broken or fixed. Then, for user aspect, we
propose three experimental network topology depending on
the network coverage: overlapping network coverage (Net1),
closed network coverage (Net2), and no-overlapping network
coverage (Net3). For the operator aspect, to decrease the
operating cost. Thus, we set up the experimental network
topology like as Net3 by varying the distance between the
Wi-Fi network coverages from 10 to 120 meters.
We use NS3 [11] to generate topology. Each connected
vehicle will move depending on NS3 mobility trace which
is generated by Simulation of Urban Mobility (SUMO) [12]
with a road map from OpenStreetMap.
V. P ERFORMANCE A NALYSIS
In this paper, we focus on in-vehicle communication with
the infotainment services such as live streaming video based on
Wi-Fi technology, while considering the two aspects: user and
operator. For the user aspect, they always require to receive
a good quality of video and to playback the video without

Fig. 3. The average percentage of packet loss.
interruption. So the quality of service regarding the loss and
transmission delay will be considered for user aspect.
For operator aspect, they prefer to serve their users with
the high user experience (UX) and also prefer to decrease the
operating expense (Opex). To achieve this, we vary the dis-
tance between the network coverages (cells) to find the suitable
range that can extend the network coverage by decreasing the
number of RSUs.
User Aspect Results
Each Net is tested with the same scenario that every 5
seconds, the connected vehicle departs from the same origin
to same destination (ENSEEIHT to Francois-Verdier) and
requests the same video with a video duration about 200
seconds and a constant data rate at 1Mbps. The experimental
results show the number of loss.
Fig. 3 illustrates the average percentage of packet loss
through the varying number of vehicles in the vehicular
network from 1 to 10 vehicles. For SDN-Mobility approach,
the number of packet loss is significantly increased when
the number of connected vehicles is large. For ten connected
vehicles with SDN-Mobility approach, Net1, Net2, and Net3
suffer from about 38%, 49% and 65% loss. But the SDNVMM
approach minimizes the loss in Net3 and prevents the loss in
Net1 and Net2.
Reducing the number of loss gives better video quality.
From the experimental results, we can conclude that SD-
NVMM approach provides the benefit with regarding loss,
leading to the better user experience.
Operator Aspect Results
Fig. 4 is illustrating the average percentage of packet loss for
different distances between the Wi-Fi network coverages. The
SDNVMM approach prevents the packet loss at all distances.
The number of loss of SDN-Mobility will be increased when
the distance increases.
VI. C ONCLUSION AND F UTURE W ORK
This paper has proposed a SDNVMM approach to provide
mobility management in a vehicular network with the ability
of SDN for handling the vehicle mobility and enabling the
cache function for reducing the packet loss. The performance
View publication stats
Fig. 4. The percentage of packet loss through the varying of the distance
between the Wi-Fi network coverages.
analysis has been measured regarding the percentage of loss
and the total transmission time. From the obtained results,
we can conclude that SDNVMM can improve the quality of
services regarding packet loss, but it is worse concerning total
transmission time. But this weakness will not affect the user
experience because the connected vehicles can be equipped
with large storage. So the data can be held in the storage before
is forwarded to the passengers in the vehicle for smooth video
playback. In this paper, we have also considered the operator
aspect; the results can be used to adjust the distance between
the RSUs. The operator can choose a suitable distance to locate
the RSUs and can reduce the number of RSUs to decrease the
operating cost without affecting the user experience.
R EFERENCES
[1] M. Kim, S. Lee, and N. Golmie, Enhanced fast handover for proxy mobile
IPv6 in vehicular networks, Journal Wireless Network, vol. 18, no. 4, pp.
401-411, May 2012.
[2] M. Balfaqih, M. Isamail, R. Nordin, and Z. A. Balfaqih, Proxy Mobile
IPv6 Management in Vehicular Networks: State of the Art, Taxonomy
and Directions for Future Research, Wireless Pers Commun 2015, vol.
84, no. 2, pp. 1509-1534, September 2015.
[3] L. Banda, M. Mzyece, and G. Noel, Fast Handover Management in IP-
based Vehicular Networks, IEEE ICT 2013, February 2013.
[4] I. Soto, C. J. Bernardos, M. Calderon, and A. Banchs, NEMO-Enabled
Localized Mobility Support for Internet Access in Automotive Scenarios,
IEEE Communications Magazine 2009, vol. 47, no. 5, May 2009.
[5] Z. Yan, H. Zhou and I. You, N-NEMO: A Comprehensive Network
Mobility Solution in Proxy Mobility IPv6 Network, Journal of Wireless
Mobile Network (JOWUA), vol. 1, no.2/3, pp.52-70, September 2010.
[6] F. Teraoka and Tetsuya Arita, PNEMO: A network-based localized
mobility management protocol for mobile networks, IEEE ICUFN 2011,
June 2011.
[7] K. Tantayakul, R. Dhaou, and B. Paillassa, Impact of SDN on Mobility
Management, IEEE IANA 2016, pp. 260-265, March 2016.
[8] K. Tantayakul, R. Dhaou, and B. Paillassa, Mobility Management with
Caching Policy over SDN Architecture, IEEE NFV-SDN 2017, November
2017.
[9] T. H. Luan, X. Ling and X. Shen, Mac in motion: impact of mobility on
the mac of drive-thru internet, IEEE Trans. Mobile Comput., vol. 11, no.
2, pp. 305-319, February 2012.
[10] C. Chen, T. H. Luan, X. Guan, N. Lu, and Y. Liu, CONNECTED VE-
HICULAR TRANSPORTATION: Data Analytics and Traffic-Dependent
Networking, IEEE VEHICULAR TECHNOLOGY MAGAZINE, pp. 42-
54, September 2017.
[11] Gustavo J.A.M Carneiro. NS-3:Network Simulator 3. Available:
https://www.nsnam.org/tutorials/NS-3-LABMEETING-1.pdf
[12] SUMO, Simulation of Urban,” Available:
http://www.sumo.dlr.de/userdoc/Downloads.html.