Sei sulla pagina 1di 11

Int. J. Electron. Commun.

(AE) 69 (2015) 15141524

Contents lists available at ScienceDirect

International Journal of Electronics and


Communications (AE)
journal homepage: www.elsevier.com/locate/aeue

REGULAR PAPER

A cross-layer mobility handover scheme for IPv6-based vehicular


networks
Wang Xiaonan , Le Deguang, Yao Yufeng
Changshu Institute of Technology, Jiangsu, Changshu 215500, China

a r t i c l e

i n f o

Article history:
Received 17 December 2013
Accepted 7 July 2015
Keywords:
Vehicular network
IPv6
Road domain
Road segment
Cluster

a b s t r a c t
This paper proposes a cross-layer mobility handover scheme for IPv6-based vehicular networks. In this
scheme, the architecture for vehicular networks is proposed and it is made up of three hierarchies including road domains, road segments and clusters. A vehicular network is made up of multiple road domains,
a road domain consists of multiple road segments, and a road segment includes multiple clusters. Based
on this architecture, the cluster generation algorithm based on the link duration time is proposed, and
the cross-layer mobility handover algorithm is presented. In the handover algorithm, the handover in
the network layer (L3) is launched before the one in the link layer (L2). Through the L3 handover process
the information on the L2 handover can be acquired in order to achieve the fast L2 handover. Moreover,
during the L3 handover process, a vehicle does not need to be congured with a care-of address, so the
L3 handover delay and packet loss are reduced. The performance of the proposed scheme is evaluated,
and the data results show that this scheme shortens the handover delay and lowers the packet loss rate.
2015 Elsevier GmbH. All rights reserved.

1. Introduction
With the technology development of vehicular networks and
the emergence of new applications, it is necessary to connect vehicular networks to the Internet in order to meet users demands for
new applications [1,2]. These applications require vehicular networks to support seamless wireless Internet services in vehicles
with high speed [3].
In wireless networks, the total handover is made up of the L2
handover and L3 handover. In the L2 handover, the channel scanning is time-consuming, and it is a main factor inuencing the
handover delay [4]. In the L3 handover, the care-of address conguration occupies a large proportion of the L3 handover delay. The
L3 handover standards such as mobile Internet protocol version 6
(MIPv6) [5] are typically applied in the wired networks. When these
protocols are applied in wireless networks, they cannot work efciently due to high packet loss and long delay [6]. Moreover, these
L3 handover protocols are totally separated from the L2 handover
ones, and they do not help improve the L2 handover performance.
In order to shorten the total handover delay and lower the
packet loss, this paper proposes a cross-layer mobility handover
scheme for vehicular networks. The main goal of this scheme is to

Corresponding author. Tel.: +86 15851550692.


E-mail address: ninawang9@163.com (X. Wang).
http://dx.doi.org/10.1016/j.aeue.2015.07.003
1434-8411/ 2015 Elsevier GmbH. All rights reserved.

combine the L3 handover with the L2 handover to improve the total


handover performance. This paper has the following contributions:

1) The architecture for vehicular networks is proposed, and it is


made up of three hierarchies including road domains, road segments and clusters. A vehicular network is made up of multiple
road domains, a road domain consists of multiple road segments,
and a road segment is composed of multiple clusters.
2) Based on this architecture, the cluster generation algorithm
based on the link duration time is proposed.
3) Based on this architecture, the cross-layer mobility handover
algorithm is presented. In this algorithm, the L3 handover is
launched before the L2 one. Through the L3 handover process, a
vehicle can acquire the channel information in order to achieve
the fast L2 handover without scanning all channels. Moreover,
in the L3 handover process, a vehicle does not need to be congured with a care-of address. As a result, the total handover delay
is reduced and the packet loss rate is lowered.

The remainder of this paper is organized as follows. In Section


2, the related work on the handover schemes is discussed. The
proposed handover scheme is presented in Section 3, and the performance is evaluated in Sections 4 and 5. This paper concludes
with a summary in Section 6.

X. Wang et al. / Int. J. Electron. Commun. (AE) 69 (2015) 15141524

2. Related work
In wireless networks, two kinds of handovers are included,
namely the L2 handover and L3 handover.
2.1. L2 handover
In the L2 handover, the channel scanning is time-consuming,
and it is a main factor inuencing the handover delay [4]. Therefore,
the L2 handover schemes focus on reducing the scanning delay.
Chiu et al. [7] propose a fast handover scheme where the information on the physical layer is shared with the link layer in order to
reduce the handover delay. This scheme operates based on mobile
multi-hop relay technique that allows inter-vehicle communications to access the Internet via a relay vehicle. In Ref. [8], the
channels usually used by access points are selected in order to avoid
the full scanning process and reduce the scanning delay.
In Refs. [9,10], a mobile node rst performs the full pre-scanning
process in order to get the information on all neighbor access points.
Based on the information, the mobile node selects the best access
point to perform the L2 handover. The data results show that the
L2 handover delay is reduced to some extent.
2.2. L3 handover
The L3 handover standards such as MIPv6 [5] are typically
applied in wired networks. When these protocols are applied in
wireless networks, they cannot work efciently due to high packet
loss and long delay [6].
Islam and Huh [11] propose the sensor PMIPv6 (SPMIPv6).
SPMIPv6 presents the network architecture and message formats
for the mobility handover process, and it also evaluates the mobility
handover cost. The results show that SPMIPv6 reduces the mobility handover cost signicantly. In Ref. [12], the mobility handover
process is achieved in the link layer, so the mobility handover delay
and cost are reduced.
Bag et al. [13] propose a scheme which reduces both the mobility
handover cost and tunnel establishment cost. This scheme depends
on dispatch types to determine source or destination of a packet. As
a result, intermediate nodes forwarding a packet have to identify
all dispatch types in order to determine the next hop, so the delay
is increased and the network scalability is also limited. Moreover,
a header structure is added between the adaptation layer and the
network layer, so the transmission delay is increased.
Denko and Wei [14] propose a mobility management scheme for
integrating MANETs into the Internet using multiple mobile gateway (MGs) and foreign agents (FAs). This scheme extends the ad hoc
on demand distance vector (AODV) and MIP to achieve the integration. The simulation results show that the use of both multiple MGs
and the hybrid gate discovery mechanism enhances the network
performance. Fan et al. [15] provide the localized mobility management scheme in mesh networks which uses the multi-path routing
to achieve the mobility handover. However, this scheme requires
some special signaling costs to deal with mobile terminals, so the
delay is prolonged to some extent.
Lee et al. [16] use an intermediate-mobile access gateway
(iMAG) to perform the mobility handover for vehicular networks.
iMAG must be geographically located between the home domain
and foreign domain, so this scheme cannot support the global
mobility management. In addition, iMAG must store the information on all road-side units, and maintaining the information
consumes a lot of network resources.
In Ref. [17], clusters are employed to improve the mobility handover performance. In this scheme, cluster heads are in charge of IP
mobility for other vehicles. Wang and Qian [18] propose a mobility handover scheme for IPv6-based vehicular networks, and this

1515

scheme improves the handover performance to some extent. However, this scheme does not use the information in the link layer to
shorten the handover delay.
Kim et al. [19] propose an enhanced PFMIPv6 (ePFMIPv6) for
vehicular networks. In ePFMIPv6, the serving MAG pre-establishes
a tunnel with multiple candidate MAGs. When the serving MAG
performs the mobility handover, it can forward the packets to the
next MAG. ePFMIPv6 shortens the mobility handover delay and
lowers the packet loss, but it increases the mobility handover cost.
In the above L3 handover schemes, a mobile node needs to be
congured with a care-of address. These schemes do not address
how to reduce the care-of address conguration delay although the
care-of address conguration delay occupies a large proportion of
the L3 handover delay.

2.3. Our solution


From the above discussion, it can be seen that the following
factors inuence the handover performance:
1) The L3 handover is totally separated from the L2 handover, and
it does not help improve the L2 handover performance.
2) The channel scanning is time-consuming and occupies a large
proportion of the L2 handover.
3) The care-of address conguration is time-consuming and occupies a large proportion of the L3 handover.
In order to shorten the handover delay and lower the packet
loss, this paper proposes a cross-layer mobility handover scheme
for IPv6-based vehicular networks. The main goal of this scheme is
to combine the L3 handover with the L2 handover to improve the
total handover performance. This scheme proposes the following
strategies to improve the handover performance:
1) The L3 handover provides the channel information in order to
help achieve the fast L2 handover without scanning all channels.
2) In the L2 handover, a vehicle does not need to scan all channels,
so the L2 handover delay is reduced.
3) In the L3 handover, a vehicle does not need to be congured with
a care-of address, so the L3 handover delay is reduced.

3. Cross-layer mobility handover


3.1. Architecture
A vehicular network is made up of access routers (ARs), base
stations and vehicles. An AR is connected to the IPv6 Internet, and
a base station is connected to an AR. The area covered by all base
stations connected to one AR is called a road domain (RD), and
the area covered by a base station is called a road segment (RS).
Vehicles are divided into three categories: a cluster head (CH) with
routing and forwarding function, a cluster member (CM) without
routing or forwarding function, and an isolated vehicle (IV). A CH
communicates directly with a base station, and a CM achieves the
communication with the Internet through its CH. An IV is a node
which does not join a cluster.
In this way, the architecture is made up of three hierarchies: an
RD which is identied by an AR, an RS which is identied by a base
station, and a cluster which is identied by a CH.
An RD is made up of multiple RS, and an RS consists of a number of clusters. A CH is usually acted by large automobiles, such as
buses. A vehicle is uniquely identied by its home address during
the mobility process, as shown in Fig. 1.

1516

X. Wang et al. / Int. J. Electron. Commun. (AE) 69 (2015) 15141524

assumed that the coordinate of the vehicle Vi /Vj is (xi , yi )/(xj , yj ),


the speed of Vi /Vj is vi /vj , the moving angle of Vi /Vj is  i / j (0  i ,
 j < 2), and the communication range of a vehicle is r. Then, the
link duration time Tij between Vi and Vj can be estimated according
to Eq. (1) [22].

Tij =

(a2 + c 2 )r 2 (ad bc) (ab + cd)


a2 + c 2

(1)

where
a = vi cos i vj cos j
b = xi xj
c = vi sin i vj sin j
d = yi yj

Fig. 1. Architecture.

3.4. Mobility handover for CH

Table 1
IPv6 address structure.
(128-i-j) bits

i bits

j bits

RD ID

RS ID

Vehicle ID

3.2. Address structure


Based on the proposed architecture, the hierarchical IPv6
address structure for vehicular networks is proposed, as shown in
Table 1.
In Table 1, an address consists of three parts. The rst part is
RD ID which is the global routing prex and uniquely identies an
RD. In an RD, the RD IDs of the base stations and the RD IDs of
the addresses acquired from this RD are the same, and the value is
equal to the one of the AR in the same RD. The second part is RS ID
which uniquely identies an RS. The RS IDs of the IPv6 addresses
acquired from one RS are the same, and the value is equal to the
one of the base station in the same RS. The third part is vehicle ID
which uniquely identies a vehicle. The address of an AR or a base
station is precongured, the RD ID and vehicle ID of an ARs address
are zero, and the vehicle ID of a base station is zero.
The values of i and j are determined by the size of a vehicular
network and the density of vehicles. Taking generality into account,
this scheme sets i to 16 and j to 32, as shown in Fig. 1. The IPv6
address conguration for vehicular networks is achieved through
our previous work [20]. After a CH acquires an address in an RS, the
AR in the same RD records the associate relationship between the
CH and the base station in the same RS.
3.3. Establishment of clusters
In this scheme, a CH is acted by a large automobile.
After a large bus starts, it marks itself as a CH and periodically broadcasts a dedicated short range communication (DSRC)
message-BasicSafetyMessage [21] whose payload includes the
node type, the speed, the mobile angle, the geographic coordinate,
the working channel and the address of the base station in the RS
where it is located.
If a vehicle is not a large automobile, then it marks itself as an IV
and scans all channels to receive the DSRC messages from neighbor CHs. Then, the IV selects the CH with the longest link duration
time, joins the cluster identied by the CH, marks itself as a CM and
begins to periodically broadcast BasicSafetyMessage whose payload includes the node type, the speed, the mobile angle, and the
geographic coordinate.
A vehicle can acquire its geographic coordinate through some
systems, for example, the global positioning system (GPS). It is

In this scheme, a base station stores the geographical coordinates and working channels of its neighbor base stations, and a CH
periodically broadcasts a DSRC message whose payload includes
the mobile angle, speed and geographic coordinate. The work in
Ref. [23] has shown that a node can determine the neighbor node
with the best communication performance via listening to a DSRC
message from its neighbors. This scheme adopts the method in Ref.
[23] to determine the next RS where a CH is entering. It is assumed
that the CH C1 is located in the RS S1 which is identied by the
base station B1. Then, B1 can acquire C1s mobile angle, speed and
geographic coordinate through receiving a DSRC message from C1.
If B1 detects that C1 is leaving its communication range, then it
calculates the link duration time between C1 and its neighbor base
stations according to formula (1) and selects as C1s next base station the base station B2 with the largest link duration time. That is,
the RS identied by B2 is C1s next RS.
3.4.1. CH inter-RS handover
If B1 and B2 belong to one RD where the AR is R1, then B1
launches the following operations:
1) B1 sends R1 a Handover message whose payload is the addresses
of C1 and B2.
2) After R1 receives the Handover message, it updates C1s associate base station with B2 and returns a Handover-Ack message
to B1.
3) After B1 receives the Handover-Ack message, it sends C1 a Handover message whose payload is B2s working channel.
4) After C1 receives the Handover message, it uses B2s working
channel to directly switch to B2 and begins to receive the data
messages from B2, as shown in Fig. 2(a) and (b).
In Fig. 2(a) and (b), at the time T1, C1 is leaving B1s communication range and entering the next RS identied by B2, R1 stores the
associate relationship between C1 and B1, and B1 launches the CH
inter-RS handover by sending a Handover message. At the time T2,
C1 switches to B2 and R1 stores the associate relationship between
C1 and B2. At this stage, the CH inter-RS handover process ends.
3.4.2. CH inter-RD handover
If B1 belongs to the RD where the AR is R1 and B2 belongs to the
RD where the AR is R2, then B1 launches the following operations:
1) B1 sends R2 a Handover message whose payload is the addresses
of C1 and B2.
2) After R2 receives the Handover message, it can determine
that B1 belongs to the different RD by checking B1s address.

X. Wang et al. / Int. J. Electron. Commun. (AE) 69 (2015) 15141524

1517

it stores these data messages. After C1 switches to B2, B2 forwards


these data messages to C1.

3.5. Mobility handover for CM


It is assumed that the CH C1 of the CM M1 is located in the RS
S1 which is identied by the base station B1, S1 belongs to the RD
where the AR is R1, and the IPv6 node N1 is located in the subnet
where the AR is R2. Then, the communication process between M1
and N1 is as follows:
1) M1 sends C1 a data message whose destination address is N1.
C1 forwards this data message to B1. B1 records the associate
relationship between M1 and C1, and then forwards the data
message to R1.
2) After R1 receives the data message, it records the associate relationship between M1 and C1, and then builds a tunnel reaching
R2. Through this tunnel, the data message reaches R2 which
forwards the data message to N1.
3) The data message returned by N1 rst reaches R2 which routes
the message to R1 through the tunnel. Based on the associate
relationship between M1 and C1 and the one between C1 and
B1, R1 forwards the data message to B1. Similarly, according
to the relationship between M1 and C1, B1 forwards the data
message to C1 which then forwards the data message to M1, as
shown in Fig. 4.

Fig. 2. CH inter-RS handover.

Therefore, R2 records the associate relationship between C1


and B2, and sends a Handover message to the AR HR of the
RD where C1 acquires the home address. The payload of the
Handover message is C1s address.
3) After HR receives the Handover message, it updates C1s
associate AR with R2 and returns a Handover-Ack message to R2.
4) After R2 receives the Handover-Ack message, it sends B1 a
Handover-Ack message. After B1 receives the Handover-Ack
message, it sends C1 a Handover message whose payload is B2s
working channel.
5) After C1 receives the Handover message, it uses B2s working
channel to directly switch to B2 and begins to receive the data
messages from B2, as shown in Fig. 3(a) and (b).
In Fig. 3(a) and (b), at the time T1, C1 is leaving B1s communication range and entering the next RS identied by B2, HR stores the
associate relationship between C1 and R1, R1 stores the associate
relationship between C1 and B1, and B1 launches the CH interRD handover by sending a Handover message. At the time T2, C1
switches to B2, HR stores the associate relationship between C1 and
R2, and R2 stores the associate relationship between C1 and B2. At
this stage, the CH inter-RD handover process ends.
In the CH inter-RS/inter-RD handover process, the L3 handover
is performed before the L2 one. If B2 receives the data messages
destined for C1 but it does not receive a DSRC message from C1, then

In Fig. 4, after the life time of the associate relationship between


M1 and C1 expires, it is deleted from B1 and R1.
In this scheme, a CH or CM periodically broadcasts a DSRC
message whose payload includes the mobile angle, speed and geographic coordinate. This scheme adopts the method in Ref. [23] to
determine the next cluster where a CM is entering. It is assumed
that the CH C1 of the CM M1 is located in the RS S1 which is identied by the base station B1, and S1 belongs to the RD where the
AR is R1. Through listening to a DSRC message from M1, C1 can
acquire M1s mobile angle, speed and geographic coordinate. If C1
detects that M1 is leaving its communication range, then it calculates the link duration time between M1 and its neighbor CHs
according to Eq. (1) and selects as M1s next CH the CH C2 with
the largest link duration time. That is, the cluster identied by C2
is M1s next cluster.
This scheme employs a DSRC message to determine the next
cluster head. Since a CM periodically broadcasts a DSRC message,
it does not need to support the additional operations to determine
the next cluster head. Moreover, it is the CH that determines the
next CH for its CM, so the CM does not need to report its neighbors
to its CH.
In this scheme, only if a CM is communicating with an IPv6 node
during its mobility process, the handover process for a CM is performed. The handover process for a CM is discussed according to
the following three situations.

3.5.1. CM intra-RS handover


If C1 and C2 belong to one RS and M1 is communicating with an
IPv6 node, then C1 launches the following operations:
1) C1 sends B1 a Handover message whose payload is the addresses
of M1 and C2.
2) After B1 receives the Handover message, it updates M1s CH with
C2 and then returns a Handover-Ack message to C1.
3) After C1 receives the Handover-Ack message, it sends M1 a Handover message whose payload is C2s working channel.

1518

X. Wang et al. / Int. J. Electron. Commun. (AE) 69 (2015) 15141524

Fig. 3. CH inter-RD handover.

Fig. 4. Communication process.

X. Wang et al. / Int. J. Electron. Commun. (AE) 69 (2015) 15141524

1519

Fig. 5. CM inter-RS handover.

4) After M1 receives the Handover message, it uses C2s working


channel to directly switch to C2 and starts receiving the data
messages from C2.
3.5.2. CM inter-RS handover
If C1 and C2 belong to different RS in the same RD, C2 is located
in the RS which is identied by the base station B2, and M1 is
communicating with an IPv6 node, then C1 launches the following
operations:

R1 and B2 store the associate relationship between M1 and C2. At


this stage, the CM inter-RS handover process ends.
3.5.3. CM inter-RD handover
If C1 and C2 belong to different RD, C2 is located in the RS which
is identied by the base station B2, C2 is located in the RD where the
AR is R2, and M1 is communicating with an IPv6 node in the subnet
where the AR is R3, then C1 launches the following operations:

1) C1 sends B2 a Handover message whose payload is the addresses


of M1 and C2.
2) After B2 receives the Handover message, it stores the associate
relationship between M1 and C2 and sends a Handover message
to R1.
3) After R1 receives the Handover message, it updates M1s CH
with C2, and returns a Handover-Ack message to B2. After B2
receives the Handover-Ack message, it sends C1 a Handover-Ack
message.
4) After C1 receives the Handover-Ack message, it sends M1 a Handover message whose payload is C2s working channel.
5) After M1 receives the Handover message, it uses C2s working
channel to directly switch to C2 and starts receiving the data
messages from C2, as shown in Fig. 5(a) and (b).

1) C1 sends B2 a Handover message whose payload is the addresses


of M1, C2 and R3.
2) After B2 receives the Handover message, it stores the relationship between M1 and C2, and forwards the Handover message to
R2. R2 stores the relationship between M1 and C2, and forwards
the Handover message to R3.
3) After R3 receives the Handover message, it establishes the tunnel
reaching R2 and returns a Handover-Ack message to R2. After R2
receives the Handover-Ack message, it sends a Handover-Ack
message to B2 which then sends a Handover-Ack message to
C1.
4) After C1 receives the Handover-Ack message, it sends M1 a Handover message whose payload is C2s working channel.
5) After M1 receives the Handover messages, it uses C2s working
channel to directly switch to C2 and starts receiving the data
messages from C2, as shown in Fig. 6(a) and (b).

In Fig. 5(a) and (b), at the time T1, M1 is leaving C1s communication range and entering the next cluster identied by C2, R1
and B1 store the associate relationship between M1 and C1, and C1
launches the CM inter-RS handover by sending a Handover message. At the time T2, M1 switches to C2 and marks C2 as its CH, and

In Fig. 6(a) and (b), at the time T1, M1 is leaving C1s communication range and entering the next cluster identied by C2, R1
and B1 store the associate relationship between M1 and C1, and C1
launches the CM inter-RD handover by sending a Handover message. At the time T2, M1 switches to C2 and marks C2 as its CH, and

1520

X. Wang et al. / Int. J. Electron. Commun. (AE) 69 (2015) 15141524

Fig. 6. CM inter-RD handover.

R2 and B2 store the associate relationship between M1 and C2. At


this stage, the CM inter-RD handover process ends.
In the CM intra-RS/inter-RS/inter-RD handover process, the L3
handover is launched before the L2 one. If C2 receives the data messages destined for M1 but it does not receive a DSRC message from
M1, then it stores these data messages. After M1 switches to C2, C2
forwards these data messages to M1.

4. Analysis
4.1. Handover delay
4.1.1. CH handover delay
Based on Fig. 2(b), the CH inter-RS mobility handover delay
TCH-RS is made up of the L3 handover delay TL3-CH-RS and the L2
handover delay TL2 , as shown in Eq. (2) where tHandover /tHandover-Ack
is the delay of transmitting a Handover/Handover-Ack message
between two neighbor nodes, DAR-BS is the distance between an
AR and a base station in the same RD, and DBS-CH is the distance
between a base station and a CH in the same RS.
TCH-RS = TL3-CH-RS + TL2
where
TL3-CH-RS = tHandover DAR-BS + tHandover-Ack DAR-BS

distance between the AR of the RD where a vehicle is located and


the AR where the vehicle acquires an address.
TCH-RD = TL3-CH-RD + TL2
where
TL3-CH-RD = (tHandover + tHandover-Ack ) DAR-BS

(3)

+ (tHandover + tHandover-Ack ) DAR-HR + tHandover DBS-CH


4.1.2. CM handover delay
The CM intra-RS mobility handover delay TCM-CH includes the L3
handover delay TL3-CM-CH and the L2 handover delay TL2 , as shown
in Eq. (4) where DCH-CM is the distance between a cluster member
and a cluster head in the same cluster.
TCM-CH = TL3-CM-CH + TL2
where
TL3-CM-CH = (tHandover + tHandover-Ack ) DBS-CH

(4)

+ tHandover DCH-CM
Based on Fig. 5(b), the CM inter-RS mobility handover delay
TCM-RS consists of the L3 handover delay TL3-CM-RS and the L2 handover delay TL2 , as shown in Eq. (5).
TCM-RS = TL3-CM-RS + TL2

(2)

+ tHandover DBS-CH
Based on Fig. 3(b), the CH inter-RD mobility handover delay
TCH-RD is made up of the L3 handover delay TL3-CH-RD and the
L2 handover delay TL2 , as shown in Eq. (3) where DAR-HR is the

where
TL3-CM-RS = (tHandover + tHandover-Ack ) DAR-BS

(5)

+ (tHandover + tHandover-Ack ) DBS-CH + tHandover DCH-CM


Based on Fig. 6(b), the CM inter-RD mobility handover delay
TCM-RD is made up of the L3 handover delay TL3-CM-RD and the L2
handover delay TL2 , as shown Eq. (6) where DAR-DR is the distance

X. Wang et al. / Int. J. Electron. Commun. (AE) 69 (2015) 15141524

between the AR of the RD where a vehicle is located and the AR of


the subnet where a destination IPv6 node is located.
TCM-RD = TL3-CM-RD + TL2
where
TL3-CM-RD = (tHandover + tHandover-Ack ) DAR-BS

(6)

+ (tHandover + tHandover-Ack ) DBS-CH


+ (tHandover + tHandover-Ack ) DAR-DR + tHandover DCH-CM
4.2. Packet loss
The real-world trafc traces and synthetic mobility models
demonstrate that it is a reasonable assumption that the vehicle
arrival follows Poisson distribution [24,25]. In this scheme, the
vehicle arrival follows Poisson distribution. It is assumed that the
position of a base station is 0, the transmission range of a base station is R,  is the vehicle density measured in vehicles per meter
(vpm), and the transmission range of a CH or CM is r which is less
than R. For a vehicle located at x in [0,R], p1 (x) is the probability
of a CH being directly connected to a base station, and p2 (x) is the
probability of a CM being connected to a base station through a
CH which is directed connected to a base station. Then, p1 (x) and
p2 (x) are shown in Eqs. (7) and (8) [26] where gBS (x) is the probability that a CH and a base station separated by a distance x are
directly connected, and gCH (x) is the probability that a CM and a CH
separated by a distance x are directly connected.
p1 (x) = gBS (x)
p2 (x) = 1 e

 2r
0

(7)
gCH (xy)p1 (y)dy

(8)

4.2.1. CH packet loss


During the CH inter-RS and inter-RD mobility handover processes, the L3 handover is performed before the L2 one, so during
the L3 handover a CH can still receive the data messages from the
original serving base station. As a result, the inter-RS packet loss
rate PCH-RS and inter-RD packet loss rate PCH-RD are shown in Eqs.
(9) and (10) where v is the average speed.
PCH-RS =

PCH-RD =

tHandover-Ack DAR-BS + tHandover DBS-CH + TL2


p1 (x)
R/v

(9)

tHandover-Ack (DAR-BS + DAR-HR ) + tHandover DBS-CH + TL2


R/v
p1 (x)

(10)

4.2.2. CM packet loss


During the CM intra-RS, inter-RS and inter-RD mobility handover processes, the L3 handover is performed before the L2 one,
so during the L3 handover a CM can receive the data messages from
the original CH. As a result, the intra-RS packet loss rate PCM-CH , the
inter-RS packet loss rate PCM-RS and the inter-RD packet loss rate
PCM-RD are shown in Eqs. (11)(13).
PCM-CH =

tHandover-Ack DBS-CH + tHandover DCH-CM + TL2


p2 (x)
r/v
(11)

PCM-RS =

tHandover-Ack (DAR-BS + DBS-CH ) + tHandover DCH-CM + TL2


r/v
p2 (x)

(12)

1521

Table 2
Simulation parameters.
Parameters

Values

1030 m/s
25 ms
12 km
200300 m
5 ms
1
4
10
0.95
500 s

TL2
R
r
tHandover /tHandover-Ack
DAR-BS /DBS-CH /DCH-CM
DAR-HR /DAR-DR
Rounds
Condence level
Simulation time

PCM-RD
=

tHandover-Ack (DAR-DR + DAR-BS + DBS-CH ) + tHandover DCH-CM + TL2


r/v

p2 (x)

(13)

Based on Eqs. (14) and (15) [26], PCH-RS , PCH-RD , PCM-CH , PCM-RS
and PCM-RD can be evaluated.

gBS (x) =

0; otherwise


gCH (x) =

1; x R

1; x r
0; otherwise

(14)

(15)

5. Simulation
NS-2 is used to evaluate the performance of the proposed
scheme. In this scheme, the link protocol adopts the IEEE 802.11p
standard which is dened by Ref. [27], so the simulation parameters
are set based on Ref. [27], as shown in Table 2. In the simulation,
after a node starts it periodically broadcasts a BasicSafetyMessage
at a rate of every 100 ms. The average speed of a vehicle ranges
from 10 m/s to 30 m/s, and the L2 handover delay is set to 25 ms.
The average data of 10 simulation rounds are used to evaluate the
handover delay and packet loss rate, and the simulation time for
one round is 500 s. The trafc model follows Poisson distribution.
In Poisson process, the number of events in a given interval follows
Poisson distribution [28,29]. In the trafc model, Poisson process is
used to describe the arrivals of vehicles in a given interval. That is,
in a given period the number of vehicles arriving follows Poisson
distribution.
5.1. The effect of speed
5.1.1. The effect of speed on CH
When R is 1 km, the CH handover delay and packet loss rate
based on speed are shown in Figs. 7 and 8.
With the increase in speed, the frequency of a CH performing the
inter-RS/inter-RD mobility handover increases and the link stability weakens, so both the network trafc and the packet loss rate
grow, as shown in Fig. 8. Since the retransmission of the lost packets causes the extra delay, there is a slight increment in the delay,
as shown in Fig. 7. From Figs. 7 and 8, it can be seen that both the
inter-RS handover delay and packet loss rate are lower than the
inter-RD ones.
5.1.2. The effect of speed on CM
When r is 250 m, the CM handover delay and packet loss rate
based on speed are shown in Figs. 9 and 10.

1522

X. Wang et al. / Int. J. Electron. Commun. (AE) 69 (2015) 15141524

Inter-RS-Ana
Inter-RD-Ana

Inter-RS-Ana
Inter-RD-Ana

Inter-RS-Sim
Inter-RD-Sim

120

100

100

80
Delay(ms)

Delay(ms)

80
60
40
20

60
40
20

0
10

15

20

25

30

Speed(m/s)

1000

1200

1400

Inter-RS-Ana
Inter-RD-Ana

1600

1800

2000

Fig. 7. CH handover delay based on speed.

Fig. 11. CH handover delay based on R.

Inter-RS-Sim
Inter-RD-Sim

Inter-RS-Ana
Inter-RD-Ana

5
4

Inter-RS-Sim
Inter-RD-Sim

Loss rate()

Loss rate()

Inter-RS-Sim
Inter-RD-Sim

2
1
0
10

15

20

25

1.5
1
0.5

30

Speed(m/s)

1000

Fig. 8. CH packet loss rate based on speed.

1200

1400

1600

1800

2000

Intra-RS-Ana
Inter-RS-Ana
Inter-RD-Ana

Intra-RS-Sim
Inter-RS-Sim
Inter-RD-Sim

100

With the increase in speed, the frequency of a CM performing


the intra-RS/inter-RS/inter-RD mobility handover increases and the
link stability weakens, so both the network trafc and the packet
loss rate grow, as shown in Fig. 10. Since the retransmission of the
lost packets causes the extra delay, there is a slight increment in
the delay, as shown in Fig. 9. From Figs. 9 and 10, it can be seen that
the intra-RS handover delay and packet loss rate are minimum, and
the inter-RD handover delay and packet loss rate are maximum.

Delay(ms)

80
60
40
20
0
10

15

20

25

30

Speed(m/s)
Fig. 9. CM handover delay based on speed.

Intra-RS-Ana
Inter-RS-Ana
Inter-RD-Ana

Intra-RS-Sim
Inter-RS-Sim
Inter-RD-Sim

10
8
Loss rate()

Fig. 12. CH loss rate based on R.

6
4
2
0
10

15

20

25

Speed(m/s)
Fig. 10. CM packet loss rate based on speed.

30

5.2. The effect of communication range


5.2.1. The effect of communication range on CH
When the average speed is 20 m/s, the CH handover delay and
packet loss rate based on R are shown in Figs. 11 and 12.
When R increases, the frequency of a CH performing the
inter-RS/inter-RD handover reduces. Therefore, the network trafc
caused by the CH handover decreases and the network performance
enhances. As a result, the packet loss rate and the additional delay
caused by the retransmission of the lost packets are reduced. Hence,
the packet loss rate and the handover delay are reduced, as shown
in Figs. 11 and 12.
5.2.2. The effect of communication range on CM
When the speed is 20 m/s, the CM handover delay and packet
loss rate based on r are shown in Figs. 13 and 14.
With the increased in r, the frequency of a CM performing
the intra-RS/inter-RS/inter-RD handover decreases, so the network
trafc caused by the CM handover reduces and the network performance enhances. Also, the packet loss rate and the additional delay
caused by the retransmission of the lost packets are reduced, so the
handover delay is decreased, as shown in Figs. 13 and 14.

X. Wang et al. / Int. J. Electron. Commun. (AE) 69 (2015) 15141524

Intra-RS-Ana
Inter-RS-Ana
Inter-RD-Ana

120

Intra-RS-Sim
Inter-RS-Sim
Inter-RD-Sim

Proposed

Standard

ePFM IPv6

90
80
Delay(ms)

100
Delay(ms)

1523

80
60

70
60

40

50

20

20

0
200

225

250

275

22

24

26

28

30

Speed(m/s)

300

Fig. 15. Delay comparison based on speed.

r
Fig. 13. CM handover delay based on r.

Proposed

Standard

ePFM IPv6

Intra-RS-Ana
Inter-RS-Ana
Inter-RD-Ana

Intra-RS-Sim
Inter-RS-Sim
Inter-RD-Sim

10

Loss rate()

8
6
4

Loss rate(%)

8
6
4
2
0
20

22

24

26

28

30

Speed(m/s)
0
200

225

250

275

300

Fig. 16. Loss rate comparison based on speed.

5.3. Comparison

Fig. 14. CM loss rate based on r.

From Fig. 7 to Fig. 14, it can be seen that the delay and packet loss
rate in the simulation do not totally match the ones in the analysis,
and the main reasons are analyzed as follows:

1) The analysis only focuses on the delay caused by the handover


process itself. That is, the handover delay in the analysis is the
interval from the time when the handover process is launched
to the time when the handover process is complete, and it does
not include the extra delay caused by the retransmission of the
lost packets. In the simulation, the handover delay is made up
of the delay caused by the handover process itself and the extra
delay caused by the retransmission of the lost packets. Therefore,
compared with the handover delay in the analysis, the handover
delay in the simulation has a slight increment which tends to be
equal to the extra delay caused by the retransmission of the lost
packets.
2) The analysis only focuses on the packet loss caused by the handover process itself, so the packet loss does not include the extra
packet loss caused by the link instability. In the simulation, the
packet loss is made up of the packet loss caused by the handover process itself and the extra packet loss caused by the link
instability. Therefore, compared with the packet loss rate in the
analysis, the packet loss rate in the simulation has a slight increment which tends to be equal to the extra packet loss rate caused
by the link instability.

In order to evaluate the performance of this scheme, we compare


our scheme with the standard [1] and the new handover scheme
ePFMIPv6 [19]. When R is 1 km and r is 250 m, the handover
delay and packet loss rate comparison based on speed are shown
in Figs. 15 and 16.
With the increase in speed, the link stability weakens and the
frequency of a vehicle performing the mobility handover increases.
Therefore, the network trafc caused by the handover grows and
the network performance degrades. As a result, the handover delays
and the packet loss rates in these three solutions all grow, as shown
in Figs. 15 and 16. In this scheme, the number of CMs is much more
than the number of CHs, so the overall performance tends to present
the performance of the CM handover. When the speed increases,
the frequency of a CM performing the Inter-RD handover grows.
Because the packet loss rate and the delay in the inter-RD handover
are more than the ones in both the intra-RS handover and inter-RS
handover, the packet loss rate and the delay grow with the increase
in speed.
Since the number of CMs is much more than the number of CHs,
the overall performance tends to present the performance of the
CM handover. Therefore, the handover delay and packet loss rate
based on r are evaluated. When the speed is 10 m/s and R is 1 km,
the handover delay and packet loss rate based on r are shown in
Figs. 17 and 18.
With the increase in r, the frequency of a vehicle performing the
handover decreases, so the network trafc caused by the handover
reduces and the network performance enhances. As a result, the
packet loss rate and the additional delay caused by the retransmission of the lost packets are reduced, so the handover delay is also
decreased, as shown in Figs. 17 and 18.

1524

X. Wang et al. / Int. J. Electron. Commun. (AE) 69 (2015) 15141524

Proposed

Standard

ePFM IPv6

100

This work is supported by Jiangsu Nature Science Foundation


(BK20141230) and National Natural Science Foundation of China
(61202440).

Delay(ms)

90
80

References

70
60
50
200

225

250

275

300

r
Fig. 17. Delay comparison based on communication range.

Proposed

Standard

ePFM IPv6

4
3
Loss rate(%)

Acknowledgements

2
1
0
200

225

250

275

300

r
Fig. 18. Loss rate comparison based on communication range.

From Fig. 15 to Fig. 18, it can be seen that the proposed scheme
has better performance than the standard and ePFMIPv6. In the
standard and ePMIPv6, a vehicle needs to be congured with a
new care-of address, so the frequent change of a vehicles address
increases the packet loss rate. In addition, a vehicle needs to scan
all channels to achieve the L2 handover, so the total handover delay
and packet loss rate grow. In the proposed scheme, the L3 handover
is performed before the L2 one. Through the L3 handover process, a
vehicle can achieve the L2 handover without scanning all channels,
so the L2 handover delay is shortened and the packet loss rate is
lowered. Moreover, during the L3 handover process, a vehicle does
not need to be congured with a care-of address, so the handover
delay in L3 is reduced and the packet loss is decreased.
6. Conclusion
This paper proposes a cross-layer mobility handover scheme for
IPv6-based vehicular networks. In this scheme, the L3 handover is
performed before the L2 one, and through the L3 handover process
a vehicle can achieve the fast L2 handover. Moreover, during the L3
handover process, a vehicle does not need to be congured with a
care-of address.
The performance of the proposed scheme is evaluated and the
data results show that this scheme improves the handover performance.

[1] IEEE 1609 WG.1609. 3-2010 IEEE Standard for Wireless Access in Vehicular
Environments (WAVE) Networking Services. IEEE Std 1609. 3-2010 (Revision
of IEEE Std 1609.3-2007), 2010.
[2] Gerla M, Kleinrock L. Vehicular networks and the future of the mobile internet.
Comput Netw 2011;55(2):45769.
[3] Park JT, Chun SM. Fast mobility management for delay-sensitive applications
in vehicular networks. Commun Lett IEEE 2011;15(1):313.
[4] Mishra A, Shin M, Arbaugh W. An empirical analysis of the IEEE
802.11 MAC layer handoff process. ACM SIGCOMM Comput Commun Rev
2003;33(2):93102.
[5] Perkins C, Johnson D, Arkko J. Mobility Support in IPv6. RFC 6275; 2011.
[6] Kun Zhu, Dusit Niyato, Ping Wang, Ekram Hossain, Dong In Kim. Mobility and
handoff management in vehicular networks: a survey. Wirel Commun Mob
Comput 2011;11(4):45976.
[7] Chiu K, Hwang R, Chen Y. Cross-layer design vehicle-assisted handover scheme
in VANETs. Wirel Commun Mob Comput 2011;11(7):91628.
[8] Shin S, Forte AG, Rawat AS, et al. Reducing MAC layer handoff latency in IEEE
802.11 wireless LANs. In: Proceedings of the second international workshop on
Mobility management & wireless access protocols. ACM; 2004. p. 1926.
[9] Ramani I, Savage S. SyncScan: practical fast handoff for 802.11 infrastructure
networks. In: INFOCOM 2005. 24th Annual Joint Conference of the IEEE Computer and Communications Societies. Proceedings IEEE, vol. 1. IEEE; 2005. p.
67584.
[10] Wu H, Tan K, Zhang Y, et al. Proactive scan: Fast handoff with smart triggers for
802.11 wireless LAN. In: INFOCOM 2007. 26th IEEE international conference on
computer communications, IEEE. IEEE; 2007. p. 74957.
[11] Islam MM, Huh EN. Sensor proxy mobile IPv6 (SPMIPv6) a novel scheme for
mobility supported IP-WSNs. Sensors 2011;11(2):186587.
[12] Wang X, Zhong S, Zhou R. A mobility support scheme for 6LoWPAN. Comput
Commun 2012;35(3):392404.
[13] Bag G, Raza MT, Kim KH, et al. LoWMob: intra-PAN mobility support schemes
for 6LoWPAN. Sensors 2009;9(7):584477.
[14] Denko MK, Wei C. A multi-gateway-based architecture for integrating ad hoc
networks with the internet using multiple foreign agents. Int J Ad Hoc Ubiquitous Comput 2008;3(2):99110.
[15] Fan Y, Zhang J, Shen X. Mobility-aware multi-path forwarding scheme for wireless mesh networks. In: Wireless communications and networking conference,
2008. WCNC 2008. IEEE. IEEE; 2008. p. 233742.
[16] Lee KW, Seo WK, Cho YZ, et al. Inter-domain handover scheme using an intermediate mobile access gateway for seamless service in vehicular networks. Int
J Commun Syst 2010;23(910):112744.
[17] Boukerche A, Zhang Z, Fei X. Reducing handoff latency for NEMO-based vehicular ad hoc networks. In: Global Telecommunications Conference (GLOBECOM
2011), 2011 IEEE. IEEE; 2011. p. 15.
[18] Wang X, Qian H. A mobility handover scheme for IPv6-based vehicular ad hoc
networks. Wirel Pers Commun 2013;70(4):184157.
[19] Kim MS, Lee SK, Golmie N. Enhanced fast handover for proxy mobile IPv6 in
vehicular networks. Wirel Netw 2012;18(4):40111.
[20] Wang X, Zhong S. Research on IPv6 address conguration for a VANET. J Parallel
Distrib Comput 2013;73(6):75766.
[21] Dedicated Short Range Communications (DSRC) Message Set Dictionary. SAE
J2735; 2009.
[22] Benslimane A, Barghi S, Assi C. An efcient routing protocol for connecting
vehicular networks to the Internet. Pervasive Mob Comput 2011;7(1):98113.
[23] Xiaonan Wang, Huanyan Qian. Constructing a VANET based on cluster chains.
Int J Commun Syst 2012, http://dx.doi.org/10.1002/dac.2484.
[24] Tan WL, Lau WC, Yue OC, et al. Analytical models and performance evaluation
of drive-thru internet systems. IEEE J Sel Areas Commun 2011;29(1):20722.
[25] Tonguz OK, Viriyasitavat W, Bai F. Modeling urban trafc: a cellular automata
approach. IEEE Commun Mag 2009;47(5):14250.
[26] Ng SC, Zhang W, Zhang Y, et al. Analysis of access and connectivity probabilities
in vehicular relay networks. IEEE J Sel Areas Commun 2011;29(1):14050.
[27] 802.11p-2010 IEEE Standard for Information technology Local and
metropolitan area networks Specic requirements Part 11: Wireless LAN
Medium Access Control (MAC) and Physical Layer (PHY) Specications Amendment 6: Wireless Access in Vehicular Environments; 2010.
[28] Cannizzaro F, Greco G, Rizzo S, et al. Results of the measurements carried
out in order to verify the validity of the Poisson-exponential distribution in
radioactive decay events. Int J Appl Radiat Isot 1978;29(11):64951.
[29] Frank A. Haight handbook of the Poisson distribution. New York: John Wiley &
Sons; 1967.

Potrebbero piacerti anche