Wireless Pers Commun DOI 10.1007/s11277-015-2592-y

Mobility Management for Delay-Sensitive Urban Vehicular Networks Xiaonan Wang1 • Deguang Le1 • Hongbin Cheng1 • Yufeng Yao1

 Springer Science+Business Media New York 2015

Abstract This paper proposes a mobility management scheme for delay-sensitive urban vehicular networks in order to reduce the mobility handover delay. The architecture based on road domains is presented. Based on this architecture, the intra-RD and inter-RD mobility handover algorithms are proposed. During the intra-RD mobility handover process, a vehicle’s care-of address keeps unchanged, so the intra-RD mobility handover process does not include the care-of address configuration. As a result, the handover cost and delay are reduced. During the inter-RD mobility handover process, a vehicle can receive the data destined for both its old care-of address and new care-of address from the same border access point, so the packet loss is substantially reduced. This paper evaluates the performance of this scheme, and the data results show that this scheme reduces the mobility handover cost and delay. Keywords

Mobility management  IPv6  Vehicular network  Road domain  Handover

1 Introduction Over the last few years, there has been significant progress in the development of urban vehicular networks and the role of newly emerging infotainment applications has rapidly taken an important place. Since these infotainment applications require vehicular networks to support delay-sensitive Internet services in fast-moving vehicles, it is important to reduce mobility handover delay in vehicular networks [1, 2]. In vehicular networks, a vehicle moves in high speed, so it frequently changes an attachment point. As a result, a vehicle has to frequently change its care-of address (CoA) and perform the mobility handover in order to keep communication continuity. However, the

& Xiaonan Wang [email protected] 1

Changshu Institute of Technology, Changshu 215500, Jiangsu, China

123

X. Wang et al.

typical mobility management standards, such as MIPv6, HMIPv6, and FMIPv6 [3–5], are unsuitable for vehicular networks due to considerable packet loss and long latency [6]. Therefore, it needs further research on the mobility management for delay-sensitive urban vehicular networks in order to support high-quality Internet services in fast-moving vehicles. Based on our pervious work [2], this paper proposes a low-delay mobility management scheme for urban vehicle networks, and it has the following contributions: 1. 2.

3.

The architecture based on road domains is proposed in order to reduce the frequency of CoA configuration and registration. Based on this architecture, the intra-RD mobility handover algorithm is proposed. During the intra-RD mobility handover process, a vehicle’s CoA keeps invariable, so the intra-RD mobility handover process does not include the CoA configuration. As a result, the mobility handover cost and delay are reduced. Based on this architecture, the inter-RD mobility handover algorithm is proposed. During the inter-RD mobility handover process, a vehicle can receive the data destined for both its old CoA and new CoA from the same border access point (BAP), so the packet loss is substantially reduced. The differences between this scheme and our previous work [2] are as follows:

1.

2.

3.

The architectures are essentially different. In the architecture of our previous work, an access point (AP) is only connected to an access router (AR) and neighbor vehicles can directly communicate with each another. In the architecture of this scheme, a BAP is defined to improve the mobility handover performance, and it is connected to more than one AR. Moreover, in this scheme neighbor vehicles unnecessarily communicate with each another. The mobility handover algorithms based on the architectures are different. In the previous work, it is a vehicle’s neighbor vehicle that launches the mobility handover for that vehicle, so a vehicle must have at least one neighbor vehicle and it can directly communicate with its neighbor vehicles. In order to overcome this limitation in the previous work, in this scheme it is an AP that performs the mobility handover for a vehicle. Moreover, in order to reduce packet loss, this scheme defines a BAP so that during the mobility handover process a vehicle can receive data from a BAP. The different communication technologies are used to perform the mobility handover algorithms. In the previous work, tunneling technology is employed to achieve the mobility handover, so the delay and cost are increased to some extent. In order to reduce the mobility handover delay and cost, in this scheme the mobility handover is performed without tunneling technology.

The remainder of this paper is organized as follows. The related work on the mobility handover is discussed in Sect. 2, the intra-RD and inter-RD mobility handover algorithms for vehicular networks are presented in Sect. 3, and the performance of this scheme is evaluated in Sect. 4. We conclude the paper with a summary in Sect. 5.

2 Related Work When a mobile node moves between different IP domains, it has to perform the mobility handover in order to keep communication continuity. The Internet engineering task force (IETF) proposes various IP-based mobility management protocols, such as MIPv6,

123

Mobility Management for Delay-Sensitive Urban Vehicular…

HMIPv6, and FMIPv6. However, most of these protocols suffer from considerable packet loss and long latency [6], so PMIPv6 [7] is proposed to improve handover performance. PMIPv6 is originally designed for local mobility management, so it also suffers from large delay and packet loss when it is used to support global mobility management [8, 9]. PFMIPv6 [10] extends FMIPv6 to PMIPv6 to reduce such handover delay and packet loss in PMIPv6. When PFMIPv6 is used in vehicular networks, it does not improve handover performance because it does not fully take into account main characteristics of vehicular networks, such as high mobility [8]. At present, some mobility handover schemes are proposed to improve handover performance. Since a vehicular network is a special kind of mobile ad hoc network (MANET), the mobility handover schemes for MANET and vehicular network are discussed respectively.

2.1 Mobility Handover for MANET In Bag et al. [11], during the mobility handover process the dispatch types are employed to determine source and destination of a control message. As a result, intermediate nodes forwarding a message must identify all the dispatch types to determine the next hop, so the mobility handover delay is increased and the network scalability is limited. Moreover, this scheme adds a header structure between the adaptation layer and the IP layer, and this header also increases the transmission delay to some extent. In Islam and Huh [12], the network architecture and the message format are presented. The mobility handover performance parameters, including the mobility handover cost and energy consumption, are evaluated. The results show that the proposed scheme reduces the energy consumption significantly. In Wang et al. [13], the control messages used for the mobility handover are exchanged in the link layer, so the mobility handover delay and cost are reduced to some extent. In Ha et al. [14], each mobile node is equipped with a partner which stores the information on personal area network (PAN) coordinators in neighbor PANs. According to this information, a mobile node’s partner performs the pre-configuration process for the mobile node in order to shorten the handover delay. If a mobile node moves relatively fast or its mobile angle suddenly changes, then the pre-configuration process may fail. In Couto et al. [15], the mobility handover process between two neighbors is optimized. If a mobile node moves in high speed, then its neighbors frequently change. Therefore, this scheme is suitable for the low-speed networks. In addition, when two nodes have not yet realized that they are neighbors, this scheme may fail. In Wang et al. [16], a local mobility management scheme is proposed and it only deals with intra-domain mobility handover. This scheme depends on a location server to achieve the mobility handover. If the location server fails, then the entire system may collapse. In Amir et al. [17], the authors focus on the second-tier handover, and more than one AP is allowed to serve mobile nodes in order to support fast mobility handover. In this scheme, the link quality is monitored and the best link is selected to ensure the communication quality. In Denko [18], the mobility management scheme based on mobile gateways (MGs) is proposed and it adopts the buffering mechanism to achieve the mobility management in hybrid networks. The performance parameters of the mobility management schemes, including optimized handover, optimized handover with prediction and forced handover, are compared and evaluated. Based on Denko [18], in Denko and Wei [19] the authors integrate MANET into the Internet using multiple MGs. This scheme extends the ad hoc ondemand distance vector routing (AODV) and MIP to achieve the integration. The simulation results show that the hybrid gate discovery mechanism and MGs enhance the

123

X. Wang et al.

mobility handover performance. In Fan et al. [20], the localized mobility management in mesh networks is proposed, and it uses the multi-path routing to reduce the handover delay. However, this solution requires some special signaling costs to deal with mobile terminals, so the handover cost is increased to some extent. In Capone et al. [21], a mesh routing and mobility management scheme is proposed. The routing protocol is the optimized link state routing (OLSR), and it depends on flooding OLSR HNA messages to signal mobility events. Therefore, the network scalability is limited.

2.2 Mobility Handover for Vehicular Networks As a special kind of MANET, a vehicular network has its unique characteristics. Gerla et al. [22] discuss the development trend of vehicular networks, and present the key issues of mobility management for vehicular networks. They also expound the reasons why the existing technologies can not solve these key issues very well. In Park and Chun [1], HMIPv6 is extended to support the mobility handover in vehicular networks, and the architecture for supporting multiple tunnels is proposed. Both the mathematical analysis and simulation are done for performance evaluation, and the simulation results that the mobility handover performance is improved. However, maintaining multiple tunnels consumes network resources to some extent. In Chiu et al. [23] a cross-layer fast handover scheme is proposed. In this scheme, the attributes in the physical layer are shared with the link layer to reduce the handover delay. This scheme is based on WiMAX multi-hop relay technique that allows inter-vehicle communications to access the Internet via a relay vehicle. However, this scheme does not discuss the IP mobility. In Lee et al. [9], an intermediate-mobile access gateway (iMAG) is used to perform the mobility handover for vehicular networks. Since iMAG must be geographically located between the home domain and foreign domain, this scheme cannot support the global mobility management. In enhanced PFMIPv6 (ePFMIPv6) [8], the serving MAG pre-establishes a tunnel with multiple candidate MAGs. In this way, when the serving MAG performs the mobility handover, the packets can be forwarded to the next MAG. ePFMIPv6 shortens the mobility handover delay and lowers the packet loss, but it increases mobility handover cost. In Wang and Qian [2], a mobility handover scheme for IPv6-based vehicular networks is proposed. In this scheme, it is a vehicle’s neighbor vehicle that launches the mobility handover for that vehicle, and the tunneling technology is employed to achieve the mobility handover. Inevitably, the tunneling increases the mobility handover cost and delay to some extent.

2.3 Our Solution In vehicular networks, a vehicle moves in relatively high speed, so it has to frequently perform the mobility handover to keep communication continuity. In order to support the high-quality Internet services in fast moving vehicles, it is important to improve the mobility handover performance in vehicular networks [1, 22]. However, from the above discussion, it can be seen that the existing handover schemes have some limitations. For example, some schemes [9, 23] do not support the global mobility management, and the mobility handover costs in Park and Chun, Wang and Qian, Kim et al. [1, 2, 8] are relatively high, etc. In order to overcome these limitations, this paper proposes a mobility management scheme for vehicular networks. Compared with the existing schemes, this scheme has the following contributions:

123

Mobility Management for Delay-Sensitive Urban Vehicular…

1.

2. 3.

The architecture based on road domains is proposed in order to reduce the frequency of CoA configuration and registration. In this way, the mobility handover cost and delay are reduced. During the intra-RD mobility handover process, a vehicle does not need to be configured with a new CoA, so the mobility handover cost and delay are reduced. This scheme defines a BAP so that during the inter-RD mobility handover process a vehicle can receive the data destined for both its old CoA and new CoA from a BAP. In this way, the packet loss is substantially reduced. Moreover, the inter-RD mobility handover provides the global mobility support.

3 Mobility Management 3.1 Architecture A vehicular network is made up of ARs, APs and vehicles. An AR is connected to the routing backbone of the IPv6 Internet, an AP is connected to an AR (or multiple ARs), and a vehicle achieves the communication with the IPv6 Internet through an AR and an AP. All the APs cover the entire vehicular network. The area covered by an AP is called a road segment (RS), and the area covered by the APs connected to one AR is called a road domain (RD). An AP at the junction of multiple RDs is called a BAP which simultaneously belongs to these RDs, and for each RD it has an IPv6 address, as shown in Fig. 1. When a vehicle joins a vehicular network, it acquires a home address (HoA). The RD where a vehicle acquires a HoA is called the vehicle’s home RD. When a vehicle enters a new RD, it obtains a CoA to ensure the routing correctness. In the home RD, a vehicle’s HoA and CoA are the same. When a vehicle moves within a RD, its HoA and CoA keep invariable. When a vehicle enters a new RD, it acquires a new CoA.

IPv6 Internet 3E01:1:1:1:2::/80 AP

3E01:1:1:2::/64 AR1

3E01:1:1:1:1::/80

3E01:1:1:2:3::/80

3E01:1:1:1::/64

AP

AR2

3E01:1:1:2:2::/80 AP

BAP 3E01:1:1:1:3::/80 3E01:1:1:2:1::/80

RS RD

AP

Vehicle HA:3E01:1:1:1:2::1/80 CA:3E01:1:1:2:1::1/80

Fig. 1 Architecture for vehicular network

123

X. Wang et al.

3.2 Address Structure Based on the architecture in Fig. 1, the hierarchical address structure for vehicular networks is proposed, as shown in Table 1. In Table 1, an IPv6 address is made up of three parts. The first part is the RD ID which is the global routing prefix and uniquely identifies an RD. The RD IDs of all APs in one RD are the same, and the RD IDs of the addresses acquired in one RD are also identical. The value is equal to the RD ID of the AR in the same RD. The second part is the RS ID which uniquely identifies an RS and is i-bit long. The RS IDs of the addresses acquired in one RS are identical and the value is equal to the RS ID of the AP in the same RS. The third part is V ID which uniquely identifies a vehicle and is j-bit long. The RS ID and V ID of an AR are zero, and the V ID of an AP is zero, as shown in Fig. 1. The values of i and j are determined by the size of a vehicular network and the density of vehicles. Taking the generality into account, the scheme sets i to 16 and j to 48.

3.3 Mobility Handover In a vehicular network, an AP periodically broadcasts a dedicated short range communications (DSRC) message-BasicSafetyMessage [24]. The payload of a DSRC message broadcasted by a BAP includes all the addresses in the RDs it belongs to. Each AR has a routing table which stores the relationship between a vehicle in the same RD and the RS which this vehicle belongs to. One entry in the table includes three fields: CoA, RS and life time. Among them, the CoA field only records the RS ID and V ID of a vehicle’s CoA because the RD IDs of all vehicles’ CoAs in one RD are the same. Similarly, the RS field only records the RS ID of the corresponding AP because the RD IDs of all APs in one RD are the same. The life time attenuates with the machine clock. When the life time attenuates to zero, the corresponding entry is removed from the table.

3.3.1 Intra-RD When a vehicle moves from the current RS to the new RS in the same RD, the intra-RD mobility handover is launched to ensure the communication continuity. This scheme defines a Pre-RS message to achieve the intra-RD mobility handover. The payload of a PreRS message is the RS ID of the AP in the new RS. When a vehicle detects that it is entering the new RS in the same RD, it sends a Pre-RS message to the AR in the same RD in order to update the routing information. It is assumed that vehicle V1 moves from RS RS1 to RS RS2, the AP in RS1 is AP1, the AP in RS2 is AP2 which is not a BAP, and RS1 and RS2 belong to RD RD1 where the AR is AR1. When V1 receives a DSRC message from AP2, it launches the following operations: 1. 2. 3.

V1 sends AP1 a Pre-RS message whose payload is AP2’s RS ID. After AP1 receives the Pre-RS message, it forwards this message to AR1. AR1 receives the Pre-RS message and checks whether there is an entry for V1 in its routing table. If not, AR1 adds an entry for V1 into its table. In the entry for V1, the RS field is updated with AP2’s RS ID.

Table 1 IPv6 address structure

123

(128-i-j) bits

i bits

j bits

RD ID

RS ID

V ID

Mobility Management for Delay-Sensitive Urban Vehicular…

4.

The intra-RD mobility handover ends, as shown in Fig. 2a, b.

During the above process, V1 is located within the overlapping area of RS1 and RS2. Before AR1 updates the routing table, V1 receives the data from AP1. After AR1 updates the routing table, V1 receives the data from AP2. In this way, the communication continuity is ensured. In Fig. 2a, b, at the time T1, V1 receives the data from AP1. Since V1 receives a DSRC message from AP2, it launches the intra-RD mobility handover process. At the time T2, the intra-RD mobility handover process ends and AR1 updates the routing table, so V1 receives the data from AP2.

3.3.2 Inter-RD 3.3.2.1 Inter-RD Mobility Handover It is assumed that vehicle V1 moves from RS RS1 to RS RS2, the AP in RS1 is AP1, the AP in RS2 is BAP BAP1 which belongs to both RD1 and RD2, RS1 belongs to RD1 where the AR is AR1, RS2 belong to both RD1 and RD2 where the AR is AR2, the AR in V1’s home RD is HAR1, and V1 is communicating with correspondent node CN1. When V1 receives a DSRC message from BAP1, it does the following operations: 1. 2. 3.

4.

5.

6.

V1 sends BAP1 a Pre-RD message. After BAP1 receives the Pre-RD message, it sends AR1 a Pre-RS message and begins to perform the CoA configuration for V1. After AR1 receives the Pre-RD message, it checks if there is one entry for V1 in its routing table. If not, AR1 adds one entry for V1. In the entry for V1, the RS field is updated with the RS ID of BAP1’s address in VD1. After BAP1 performs the CoA configuration for V1, it sends V1 an RD-Res message whose payload is the new CoA, sends AR2 a Pre-RS message whose payload is the RS ID and V ID of the new CoA, and sends HAR1/CN1 a Binding message whose payload is V1’s old CoA and new CoA. After V1 receives the RD-Res message, it acquires a new CoA. After AR2 receives the Pre-RS message, it adds an entry for V1 where the CoA field is V1’s new CoA and the RS field is the RS ID of BAP1’s address in RD2. After HAR1/CN1 receives the Binding message, it updates V1’s CoA with the new CoA. The inter-RD mobility handover ends, as shown in Fig. 2c, d.

During the above process, V1 always keeps the link connection with BAP1 and receives the data destined for both its old CoA and new CoA from BAP1, so the packet loss is greatly reduced and the communication continuity is ensured. As shown in Fig. 2c, d, at the time T1, V1 receives the data from AP1. At the same time, V1 receives a DSRC message from BAP1, so it launches the inter-RD mobility handover process. At the time T2, AR1 updates the routing table, so V1 receives the data destined for the current CoA from BAP1 and disconnects from AP1. At the time T3, V1 acquires a new CoA and the inter-RD mobility handover process ends, so V1 receives the data destined for the new CoA from BAP1. 3.3.2.2 CoA Configuration In this scheme, a BAP belongs to multiple RDs and has more than one IPv6 address. Therefore, before a BAP assigns a new CoA to a vehicle, it first determines the next VD the vehicle is going to enter. Then the BAP assigns a globally unique address in the next VD to the vehicle. After BAP BAP1 receives a Pre-RD message from vehicle V1, it assigns a new CoA to V1 according to the following steps:

123

X. Wang et al.

3E01:1:1:1::/64 Routing table at T1: CoA RS 1::1/16 2 Routing table at T2: CoA RS 1::1/16 3 AR1 AP1 3E01:1:1:1:2::/80

AP2 3E01:1:1:1:3::/80 RS1

RD1

RS2

V1 At T1 HA:3E01:1:1:3:1::1/80 CoA:3E01:1:1:1:1::1/80

V1 At T2 HA:3E01:1:1:3:1::1/80 CoA:3E01:1:1:1:1::1/80

(a)

V1

AP2

AP1

Data

Data

Pre-RS

Pre-RS

AR1

Update routing table Data

Data

(b) HAR1 3E01:1:1:1::/64 Routing table at T1: CoA RS 1::1/16 3 Routing table at T2: CoA RS 1::1/16 2

AP1 3E01:1:1:1:3::/80

IPv6 Internet

AR1

3E01:1:1:2::/64 Routing table at T3: CoA RS 1::1/16 1

AR2 BAP1 VD1: 3E01:1:1:1:2::/80 VD2: 3E01:1:1:2:1::/80

RS1

RS2

RD1

RD2 V1 At T1 and T2 HA:3E01:1:1:3:1::1/80 CoA:3E01:1:1:1:1::1/80

V1 At T3 HA:3E01:1:1:3:1::1/80 CoA:3E01:1:1:2:1::1/80

(c) Fig. 2 Mobility handover process

123

CN1

Mobility Management for Delay-Sensitive Urban Vehicular…

V1

AP1 Data

Pre-RD

BAP1 Data

HAR1, CN1

AR2

AR1 Pre-RS Update routing table

Data

Data CoA configuration In the next RD

RD-Res

Pre-RS Binding Data

Data

Data

(d) Fig. 2 continued

1. 2. 3.

BAP1 sends a Fresh message to V1. After V1 receives the Fresh message, it begins to periodically send a Fresh_Res message to BAP1. In this way, BAP1 can determine the next VD V1 is going to enter. BAP1 assigns a globally unique address in the next VD and sends V1 an RD-Res message whose payload is the assigned address.

In step 2, BAP1 calculates the relative position of V1 by measuring the received Fresh_Res messages with angle of arrival (AoA) and received signal strength indicator (RSSI) [25, 26], and then determines the next VD V1 is going to enter, as shown in Fig. 3. It is assumed that BAP1 belongs to K RDs and the kth (K C k C 1) RD is defined by the closed angle interval [ak, ak?1]. At the time T1, V1 enters the communication range of BAP1 from the kth VD. The relative angle between BAP1 and V1 is a1 (ak B a1 B ak?1) and the distance between them is d1. Through measuring the Fresh_Res messages from V1 with RSSI, BAP1 can calculate the distance from V1. If the distance tends to be zero, it means that V1 is entering a new VD. Then, through measuring the Fresh_Res messages from V1 with AOA, BAP1 can calculate the relative angle of V1. At the time T2, BAP1 detects that the relative angle becomes a2 (aj B a2 B aj?1). Therefore, BAP1 can determine that V1 enters the jth (K C j C 1) VD.

Fig. 3 Process of determining the next VD

0 αK

VDK

α1

VDK-1

VD1

α2

αK-1 (α1, d1)

... αj+1

V1

... BAP1 (α 2, d 2)

... ... V1

... ...

αj T2

αk

αk+1 T1

123

X. Wang et al.

4 Performance Evaluation 4.1 Analysis The intra-RD mobility handover model is shown in Fig. 4a. Vehicle V1 moves from RS RS1 to RS RS2. RS1 and RS2 belong to one RD where the AR is AR1, the AP in RS1 is AP1, and the AP in RS2 is AP2. When V1 detects that it enters RS2, it launches the intraRD mobility handover process. The intra-RD mobility handover cost CIntra-RD is made up of the mobility detection cost CDetection and the registration costCIntra-reg, as shown in formula (1). CIntra-RD ¼ CDetection þ CIntra-reg

ð1Þ

In formula (1), CDetection is generally a constant [27]. According to Fig. 2b, CIntra-reg is shown in formula (2) where DV1-AP1 is the distance between V1 and AP1, DAP1-AR1 is the distance between AP1 and AR1, cPre-RS is the cost of transmitting a Pre-RS message between two neighbors, and CRouting is the routing cost. CIntra-reg ¼ cPr e-RS  DV1-AP1 þ cPr e-RS  DAP1-AR1 þ CRouting

ð2Þ

The mobility handover operation is made up of three parts [28], namely, the mobility detection, the CoA configuration and the location registration. One characteristic of this scheme is that in the intra-RD mobility handover process a vehicle does not need to be configured with a new CoA. Therefore, the intra-RD mobility handover delay TIntra-RD is made up of the mobility detection delay TDetection and the registration delay TIntra-reg, as shown in formula (3) where TDetection is the link-layer handover delay and is a constant [8]. Fig. 4 Mobility model

AR1

AP2

AP1

RS1

V1

RS2

(a) AR1

RS1

RD1

AR2

AP2 RS2

AP1

V1

RD2

(b)

123

Mobility Management for Delay-Sensitive Urban Vehicular…

According to Fig. 2b, TIntra-reg is shown in formula (4) where tPre-RS is the delay of transmitting a Pre-RS message between two neighbors, and TRouting is the routing delay. TIntra-RD ¼ TDetection þ TIntra-reg

ð3Þ

TIntra-reg ¼ tPr e-RS  DV1-AP1 þ tPr e-RS  DAP1-AR1 þ TRouting

ð4Þ

Before the intra-RD mobility handover is launched, V1 receives the data from AP1. After the intra-RD mobility handover ends, V1 receives the data from AP2. Therefore, the maximum packet loss PIntra during the mobility handover process is shown in formula (5) where k is the packet arrival rate. PIntra ¼ TIntra-RD  k

ð5Þ

The inter-RD mobility handover model is shown in Fig. 4b. Vehicle V1 moves from RS RS1 to RS RS2. The AP in RS1 is AP1 and RS1 belongs to RD RD1 where the AR is AR1. The AP in RS2 is AP2 and RS2 belongs to RD RD2 where the AR is AR2. When V1 detects that it enters RS2, it launches the inter-RD mobility handover.The inter-RD mobility handover cost CInter-RD is made up of the mobility detection cost CDetection, the CoA configuration cost CCoA and the registration cost CInter-reg, as shown in formula (6). CInter-RD ¼ CDetection þ CCoA þCInter-reg

ð6Þ

According to Fig. 2d, CInter-reg is shown in formula (7). In formula (7), DV1-AP2 is the distance between V1 and AP2, DAP2-AR1 is the distance between AP2 and AR1, DAP2-AR2 is the distance between AP2 and AR2, DAP2-HAR/DAP2-CN is the distance between AP2 and HAR/CN, and cPr e-RD =cRD-Res =cBinding is the cost of transmitting a(n) Pre-RD/RD-Res/ Binding message between two neighbors. CInter-reg ¼ cPr e-RD  DV1-AP2 þ cPr e-RS  DAP2-AR1 þ CRouting þ cRD-Res  DV1-AP2 þ cPr e-RS  DAP2-AR2 þ cBinding  ðDAP2-HAR þ DAP2-CN Þ

ð7Þ

According to Sect. 4.2.2, CCoA is shown in formula (8) where cFresh =cFreshRes is the cost of transmitting a Fresh/Fresh-Res message between two neighbors, and N is the number of transmitted Fresh-Res messages. CCoA ¼ cFresh  DV1-AP2 þ N  cFresh-Res  DV1-AP2

ð8Þ

The inter-RD mobility handover delay TInter-RD is made up of the mobility detection delay TDetection, the CoA configuration delay TCoA and the registration delay TInter-reg, as shown in formula (9). Since the CoA configuration and the routing update are performed at the same time, the routing update delay is not included in formula (9). TInter-RD ¼ TDetection þ TCoA þ TInter-reg

ð9Þ

According to Fig. 2d, TInter-reg is shown in formula (10). In formula (10), tPr e-RD =tRD-Res =tBinding is the delay of transmitting a(n) Pre-RD/RD-Res/Binding message between two neighbors. TInter-reg ¼ tPr e-RD  DV1-AP2 þ tPr e-RS  DAP2-AR1
þ max tRD-Res  DV1-AP2 ; tPr e-RS  DAP2-AR2 ; tBinding  DAP2-HAR ; tBinding  DAP2-CN ð10Þ

123

Fig. 5 a Mobility handover cost comparison. b Mobility handover delay comparison. c Packet loss comparison

Mobility handover cost(packet)

X. Wang et al.

(a) 50 40

Intra-RD Inter-RD

30 20 10 0

2

4

6

8

10

8

10

80

100

D AP2-HAR Mobility handover delay(ms)

(b) 250 200

Intra-RD Inter-RD

150 100 50 0

2

4

6

D AP2-HAR

(c) Packet loss(packet)

5 4

Intra-RD Inter-RD

3 2 1 0

20

40

60

λ (packet/s)

According to Sect. 4.2.2, TCoA is shown in formula (11) where tFresh =tFreshRes is the delay of transmitting a Fresh/Fresh-Res message between two neighbors. TCA ¼ tFresh  DV1-AP2 þ N  tFreshRes  DV1-AP2

ð11Þ

After V1 enters RS2, it can still receive the data destined for both its old CoA and new CoA from BAP AP2. Therefore, the packet loss PInter tends to be zero. This scheme adopts the message cost to measure the mobility handover cost, so the cost of transmitting a message is equal to 1. During the mobility detection process and the routing update process, the message costs Cdetection and CRouting are usually smaller than 6, so they are set to 6 [2, 29, 30]. The distance DAP2-CN between an AP and a CN is generally smaller than 10 hops, so it is set to 10. The intra-RD and inter-RD mobility handover cost comparison is shown in Fig. 5a. During the mobility detection process, the messages are

123

Mobility Management for Delay-Sensitive Urban Vehicular…

transmitted between two neighbors. In general, the delay of transmitting a message between two neighbors is stable and tends to be constant [27], so TDetection is set to 10 ms [8]. During the routing update process, the messages are transmitted between two nodes which are multi-hop away, so TRouting is usually more than TDetection and is set to 15 ms [8]. The intra-RD and inter-RD mobility handover delay comparison is shown in Fig. 5b, and the packet loss comparison is shown in Fig. 5c. In this scheme, during the intra-RD mobility handover process, a vehicle does not need to be configured with a new CoA, so the intra-RD mobility handover cost and delay are not affected by DAP2-HAR. During the inter-RD mobility handover process, a vehicle is configured with a new CoA, so the inter-RD mobility handover cost and delay grow with DAP2HAR, as shown in Fig. 5a, b. As shown in Fig. 5c, the packet loss in the intra-RD handover process is more than the one in the inter-RD handover process. The main reason is that during the intra-RD handover process a vehicle receives the data messages from different APs and during the inter-RD mobility handover process a vehicle always receives the data messages from a BAP. As a result, the packet loss in the intra-RD handover process grows with k while the packet loss in the inter-RD handover process tends to be zero. From Fig. 5a–c, it can be seen that the intra-RD mobility handover cost and delay are smaller, but the packet loss in the inter-RD mobility handover process is lower. The reasons are analyzed as follows: 1. 2.

The intra-RD mobility handover includes neither the CoA configuration nor the CoA registration, so the mobility handover cost and delay are lower. During the inter-RD mobility handover, a vehicle receives the data destined for both its old CoA and new CoA from a BAP, so the packet loss is smaller.

4.2 Simulation NS-2 is used as the simulation platform, and the simulation parameters are shown in Table 2. The urban vehicular network map is shown in Fig. 6 where four RDs are included. When a vehicle reaches an intersection, the probability of it moving forward/turning left/turning right is 1/3. MHVA [2] and ePFMIPv6 [8] are selected to compare with the proposed scheme due to the following reasons: 1. 2. 3.

MHVA and ePFMIPv6 are the lastest schemes which support the global mobility management for vehicular networks. MHVA and ePFMIPv6 are based on IPv6. MHVA and ePFMIPv6 outperform existing schemes [10, 31].

4.2.1 The Effect of Speed When the number of vehicles is 50, the mobility handover cost, delay and packet loss based on speed are shown in Fig. 7a–c. When the speed grows, the probability of a vehicle performing the inter-subnet/inter-RD mobility handover also does. Since the inter-subnet/inter-RD mobility handover cost and delay are more than the intra-subnet/intra-RD ones, the cost and delay grow with the speed, as shown in Fig. 7a, b. Moreover, the increase in speed can cause the growth in packet loss, so the retransmission of lost packets slightly increases the mobility handover cost and delay. MHVA and ePFMIPv6 use the tunnel technology to achieve the mobility handover,

123

X. Wang et al. Table 2 Simulation parameters

Parameters

Values

Simulation area

1km 9 1km

Number of RDs

4

Number of vehicles

50–200

Transmission range of an AP

300 m

Transmission range of a vehicle

300 m

Speed

5–20 m/s

k

100 pkt/s

MAC

IEEE 802.11p

Simulation time

500 s

1km AP

AR AR

BAP

AR

0

AR

1km

Fig. 6 Urban vehicular network

so they have more costs and delays than the proposed scheme. In ePFMIPv6, the serving MAG pre-establishes a tunnel with multiple candidate MAGs, so ePFMIPv6 has more cost and delay than MHVA. In the proposed scheme, during the inter-RD mobility handover process, a vehicle receives the data destined for both its old CoA and new CoA from one BAP, so the packet loss tends to be zero. As a result, the extra cost and delay caused by the retransmission of lost packets are greatly reduced. In the proposed scheme, with the increase in speed, the probability of a vehicle performing inter-RD mobility handover also increases. Since the inter-RD mobility handover cost and delay are more than the intra-RD ones, the cost and delay grow with the speed, as shown in Fig. 7a, b. Specially, when the speed ranges from 10 m/s to 15 m/s, there is a great increase in the mobility handover delay. The main reason is that the number of vehicles performing the inter-RD mobility handover has a great increase. After the speed is more than 15 m/s, the growth rate of the number of vehicles performing the inter-RD mobility handover is reduced and tends to be stable. As a result, the growth rate of the mobility handover delay is also reduced and tends to be steady. From Fig. 7b, it can be seen that the growth rate of this scheme approximates to the growth rate of MHVA. When the speed grows, the link stability weakens. As a result, the packet loss in MHVA and ePFMIPv6 grow with the speed. With the increase in speed, the probability of a vehicle

123

Fig. 7 a Mobility handover cost based on speed. b Mobility handover delay based on speed. c Packet loss based on speed

Mobility handover cost(packet)

Mobility Management for Delay-Sensitive Urban Vehicular…

(a) 16 Proposed scheme MHVA ePFMIPv6

12 8 4 0

5

10

15

20

Speed(m/s) Mobility handover delay(ms)

(b) 100 80 60 40 Proposed scheme MHVA ePFMIPv6

20 0 5

10

15

20

Speed(m/s)

Packet loss(packet)

(c) 8 Proposed scheme M HVA ePFM IPv6

6 4 2 0 5

10

15

20

Speed(m/s)

performing the inter-RD handover also grows. In the proposed scheme, during the inter-RD mobility handover process, a vehicle receives the data destined for both its old CoA and new CoA from the same BAP, so the packet loss tends to be zero. As a result, the packet loss decreases with the speed, as shown in Fig. 7c. Moreover, both the mobility handover delays and costs in MHVA and ePFMIPv6 are more than the ones in the proposed scheme, so the proposed scheme has lower packet loss than MHVA and ePFIMPv6.

4.2.2 The Effect of Vehicle Population From Fig. 7a–c, it can be seen that MHVA has better performance than ePFMIPv6, so we only compare the proposed scheme with MHVA. Section 4.2.1 discusses the effect of various speed on the mobility handover. The main aim of Sect. 4.2.2 is to evaluate the effect of various vehicle population on the mobility handover, so the speed is set to a

123

X. Wang et al.

constant. This scheme is for urban vehicular networks. In an urban scenario, the speed usually varies from 5 to 20 m/s [32], that is, the slow, intermediate and fast speed are 5, 15 and 20 m/s, respectively. Taking the generalization into account, this scheme sets the speed to the intermediate value 15 m/s. When the speed is 15 m/s, the mobility handover cost, delay and packet loss based on vehicle population are shown in Fig. 8a–c. When the number of vehicles grows, the probability of a vehicle performing the interRD/intra-RD mobility handover tends to be stable. Therefore, the number of the vehicles performing the inter-RD/intra-RD mobility handover also tends to be stable. As a result, the mobility handover cost and delay tend to be stable. Since the increase in vehicle population results in the growth in network traffics, the network performance is degraded. As a result, the packet loss increases, and the extra mobility handover cost and delay caused by the retransmission of lost packets also slightly grow, as shown in Fig. 8a–c.

(a) Mobility handover cost(packet)

Fig. 8 a Mobility handover cost based on vehicle population. b Mobility handover delay based on vehicle population. c Packet loss based on vehicle population

10 8 6 4

Proposed scheme MHVA

2 0 50

100

150

200

Vehicle population Mobility handover delay(m/s)

(b) 80 60 40

Proposed scheme MHVA

20 0

50

100

150

200

Vehicle population

Packet loss(packet)

(c)

5 4 Proposed scheme MHVA

3 2 1 0 50

100

150

Vehicle population

123

200

Mobility Management for Delay-Sensitive Urban Vehicular…

5 Conclusion This paper proposes the intra-RD and inter-RD mobility handover algorithms for urban vehicular networks. The performance parameters of the proposed scheme are evaluated. The data results show that the proposed scheme reduces the mobility handover cost, shortens the delay and lowers the packet loss. Acknowledgments This work is supported by Jiangsu Nature Science Foundation (BK20141230) and National Natural Science Foundation of China (61202440).

References 1. Park, J., & Chun, S. (2011). Fast mobility management for delay-sensitive applications in vehicular networks. Communications Letters, IEEE, 15(1), 31–33. 2. Wang, X., & Qian, H. (2013). A mobility handover scheme for IPv6-based vehicular ad hoc networks. Wireless Personal Communications, 70(4), 1841–1857. 3. Perkins, C., Johnson, D., & Arkko, J. (2011). Mobility support in IPv6, RFC 6275. 4. Koodli, R. (2009). Mobile IPv6 fast handovers. RFC 5568. 5. Soliman, H., Castelluccia, C., ElMalki, K., & Bellier. L. (2008). Hierarchical mobile IPv6 mobility management (HMIPv6). RFC 5380. 6. Zhu, K., Niyato, D., Wang, P., Hossain, E., & Kim, D. I. (2011). Mobility and handoff management in vehicular networks: A survey. Wireless Communications and Mobile Computing, 11, 459–476. 7. Gundavelli, S., Leung, K., Devarapalli, V., Chowdhury, K., & Patil, B (2008). Proxy mobile IPv6. IETF RFC 5213. 8. Kim, M. S., Lee, S. K., & Golmie, N. (2012). Enhanced fast handover for proxy mobile IPv6 in vehicular networks. Wireless Network., 18, 401–411. 9. Lee, K.-W., Seo, W.-K., Cho, Y.-Z., et al. (2010). Inter-domain handover scheme using an intermediate mobile access gateway for seamless service in vehicular networks. International Journal of Communication Systems, 23, 1127–1144. 10. Yokota, H., Chowdhury, K., Koodli, R., Patil, B., & Xia, F (2009). Fast handovers for PMIPv6. RFC 5949. 11. Bag, G., Raza, M., Kim, K., & Yoo, S. (2009). LoWMob: Intra-PAN mobility support schemes for 6LoWPAN. Sensors, 2009(9), 5844–5877. 12. Islam, M. M., & Huh, E. (2011). Sensor Proxy Mobile IPv6 (SPMIPv6)—A novel scheme for mobility supported IP-WSNs. Sensors, 11, 1865–1887. 13. Wang, X., Zhong, S., & Zhou, R. (2012). A mobility support scheme for 6LoWPAN. Computer Communications, 35, 392–404. 14. Ha, M., Kim, D., Kim., S. H., & Hong, S. (2010). Inter-MARIO: A fast and seamless mobility protocol to support inter-PAN handover in 6LoWPAN. In IEEE Communications Society subject matter experts for publication in the IEEE Globecom 2010 proceedings (pp. 1–6). New York: IEEE press. 15. Couto, L., Barraca, J., Sargento, S., & Aguiar, R. (2009). FastM in WMN: A fast mobility support extension for wireless mesh networks, 2009. In Second international conference on advances in mesh networks IEEE, 2009 (pp. 90–96). IEEE. 16. Wang, H., Huang, Q., Xia, Y., Wu, Y., & Yuan, Y. (2007). A network-based local mobility management scheme for wireless mesh networks. Wireless Communications and Networking Conference, IEEE, 2007, 3792–3797. 17. Amir, Y., Danilov, C., Hilsdale, M., Musaloiu-Elefteri, R., & Rivera, N. (2006). Fast handoff for seamless wireless mesh networks. In: Proceedings of the 4th international conference on mobile systems, applications and services, ACM, 2006 (pp. 83–95). ACM. 18. Denko, M. K. (2007). Mobility management in hybrid ad-hoc networks and the Internet environment. Concurrency and Computation, 19, 1193–1206. 19. Denko, M. K., & Wei, C. (2008). A multi-gateway-based architecture for integrating ad hoc networks with the internet using multiple foreign agents. International Journal of Ad Hoc and Ubiquitous Computing, 3(2), 99–109. 20. Fan, Y., Zhang, J., & Shen, X. (2008). Mobility-aware multi-path forwarding scheme for wireless mesh networks. Wireless Communications and Networking Conference, WCNC 2008, IEEE, 2008, 2337–2342.

123

X. Wang et al. 21. Capone, A., Napoli, S., & Pollastro, A. (2006). Mobimesh: an experimental platform for wireless mesh networks with mobility support. In: Proceedings of ACM QShine 2006 workshop on wireless mesh: Moving towards applications, Waterloo. 22. Gerla, M., & Kleinrock, L. (2011). Vehicular networks and the future of the mobile internet. Computer Networks, 55, 457–469. 23. Chiu, K., Hwang, R., & Chen, Y. (2011). Cross-layer design vehicle-assisted handover scheme in VANETs. Wireless Communications and Mobile Computing, 11(7), 916–928. 24. Dedicated short range communications (DSRC) message set dictionary. DRAFT SAE J2735, 2008. 25. Zhang, L., Cheng, Q., Wang, Y., & Zeadally, S. (2008). A novel distributed sensor positioning system using the dual of target tracking. IEEE Transactions on Computers, 57, 246–260. 26. Patwari, N., Ash, J., Kyperountas, S., Hero, A., Moses, R., & Correal, N. (2005). Locating the nodes: Cooperative localization in wireless sensor networks. IEEE Signal Processing Magazine, 22, 54–69. 27. Li, R., Li, J., Wu, K., & Xiao, Y. (2008). An enhanced fast handover with low latency for mobile IPv6. Wrl Ommnaon Ranaon on, 7, 334–342. 28. Han, Y.-H., Jang, H., & Choi, J. H. et al. (2006). A cross-layering design for IPv6 fast handover support in an IEEE 802.16e Wireless MAN. IEEE Network (pp. 54–62), Nov/Dec 2007. 29. Reaz, A. S., & Atiquzzaman, M. (2006). P-SIGMA: Paging in end to end mobility management. In Communications, 2006. ICC’06. IEEE international conference on (vol. 4, pp. 1813–1818). IEEE: Istanbul. 30. Reaz, A. S., Chowdhury, P. K., Atiquzzaman, M., & Ivancic. W. (2006). Signalling cost analysis of SINEMO: Seamless EndtoEnd network mobility. In: Procedings of the first ACM/IEEE international workshop on mobility in the evolving internet architecture (MobiArch 2006) (pp. 37–42), San Francisco. 31. Chen, Y., Hsu, C., & Cheng, C. (2013). Network mobility protocol for vehicular ad hoc networks. International Journal of Communication Systems,. doi:10.1002/dac.2525. 32. Marquez-Barja, J., Calafate, C. T., Cano, J., & Manzoni, P. (2010). Multi-layer performance evaluation of a content delivery framework for urban vehicular networks. In: Communications workshops (ICC), 2010 IEEE international conference on (pp. 1–5). IEEE. Xiaonan Wang received PHD from Nanjing University of Science and Technology. She is currently a full professor at Changshu Institute of Technology. Her research interests include Ad hoc networks, sensor networks, the next-generation network architecture and protocol, and the all-IP communication between IPv6 networks and wireless networks, etc.

123

Mobility Management for Delay-Sensitive Urban Vehicular… Deguang Le is currently working at Changshu Institute of Technology as an associate professor. His research interests include the network architecture and network security, etc.

Hongbin Cheng is currently working at Changshu Institute of Technology as an associate professor. His research interests include the low-power networks and 6LoWPAN, etc.

Yufeng Yao is currently working at Changshu Institute of Technology. His research interests include network architecture and clouds, etc.

123