Mobility and handoff management in vehicular networks: a survey

Transcription

1 WIRELESS COMMUNICATIONS AND MOBILE COMPUTING Wirel. Commun. Mob. Comput. 2011; 11: Published online 5 October 2009 in Wiley Online Library (wileyonlinelibrary.com)..853 SPECIAL ISSUE PAPER Mobility and handoff management in vehicular networks: a survey Kun Zhu 1, Dusit Niyato 1, Ping Wang 1, Ekram Hossain 2 and Dong In Kim 3 1 School of Computer Engineering, Nanyang Technological University, 50 Nanyang Avenue, , Singapore 2 Department of Electrical and Computer Engineering, University of Manitoba, Winnipeg, Manitoba R3T 5V6, Canada 3 School of Information and Communication Engineering, Sungkyunkwan University (SKKU), Suwon , Korea ABSTRACT Mobility management is one of the most challenging research issues for vehicular networks to support a variety of intelligent transportation system (ITS) applications. The traditional mobility management schemes for Internet and mobile ad hoc network (MANET) cannot meet the requirements of vehicular networks, and the performance degrades severely due to the unique characteristics of vehicular networks (e.g., high mobility). Therefore, mobility management solutions developed specifically for vehicular networks would be required. This paper presents a comprehensive survey on mobility management for vehicular networks. First, the requirements of mobility management for vehicular networks are identified. Then, classified based on two communication scenarios in vehicular networks, namely, vehicle-to-vehicle (V2V) and vehicle-toinfrastructure (V2I) communications, the existing mobility management schemes are reviewed. The differences between host-based and network-based mobility management are discussed. To this end, several open research issues in mobility management for vehicular networks are outlined. Copyright 2009 John Wiley & Sons, Ltd. KEYWORDS mobility management; vehicular networks; host mobility; network mobility * Correspondence Ekram Hossain, Department of Electrical and Computer Engineering at University of Manitoba, Winnipeg, Manitoba R3T 2N2, Canada. ekram@ee.umanitoba.ca 1. INTRODUCTION In recent years, there have been significant interest and progress in the field of intelligent transportation system (ITS) from both industry and academia. Typical ITS applications can be categorized into safety, transport efficiency, and information/entertainment applications (i.e., infotainment) [1]. Vehicular ad hoc networks (VANETs) are emerging ITS technologies integrating wireless communications to vehicles. Different Consortia (e.g., Car-to-Car Communications Consortium (C2C-CC) [2]) and standardization organizations (e.g., IETF) have been working on various issues in VANETs. C2C-CC aims to develop an open industrial standard for inter-vehicle communication (IVC) using wireless LAN (WLAN) technology. For example, IEEE p or dedicated short range communications (DSRC) is an extension of standard for IVC by IEEE working group. IETF has standardized NEtwork MObility Basic Support (NEMO BS) [3] for network mobility in VANETs. VANET involves vehicle-to-vehicle (V2V) and vehicleto-infrastructure (V2I) communications. V2V refers to the direct or multihop communications among vehicles in VANET. V2V is efficient and cost effective due to its short range bandwidth advantage and ad hoc nature. V2I refers to the communication between vehicles and infrastructure of roadside unit (RSU), e.g., base station and access point (AP) connected with Internet. V2I communications can be used for Internet access. A typical VANET scenario is shown in Figure 1. VANET is a special type of mobile ad hoc networks (MANETs) [4] with unique characteristics. Due to the high mobility of vehicles, topologies of VANETs are highly dynamic. Also, the density of VANETs varies dramatically. Another major difference between VANETs and traditional MANETs is that power consumption is not a major concern in VANETs. Instead, the efficiency of VANETs protocols is the most significant issue. Originating from cellular networks, mobility management has been an important and challenging issue to support seamless communication. Mobility management includes Copyright 2009 John Wiley & Sons, Ltd. 459

2 Mobility and handoff management in vehicular networks K. Zhu et al. 2. OVERVIEW OF MOBILITY MANAGEMENT IN VEHICULAR NETWORKS In this section, we discuss the mobility management issues in vehicular networks for V2I and V2V communications scenarios Mobility Management for V2I Communications Figure 1. General model of vehicular networks. location management and handoff management [5]. Location management has the functions of tracking and updating current location of mobile node (MN). Handoff management aims to maintain the active connections when MN changes its point of attachment. Mobility management is essential for providing highspeed and seamless services for vehicular networks since MNs change their points of attachment frequently and network topology can change abruptly. Due to the differences between V2I and V2V communications, corresponding mobility management schemes can be designed differently to achieve optimal performance. Since V2I communication needs data exchange with Internet, for compatibility and interoperability reasons, most mobility management solutions for V2I communication are designed based on Internet mobility management protocols (e.g., Mobile IPv6). For V2V communication, mobility management mainly focuses on route discovery, maintenance, and recovery which are similar to those in MANETs [6]. A review of the current state of the art on mobility management for both V2V and V2I-based VANETs will be provided in this paper. The rest of this paper is organized as follows. In Section 2 we provide an overview of mobility management schemes for both V2I and V2V communications. Host mobility and network mobility solutions designed at different open system interconnection (OSI) layers are introduced in Sections 3 and 4, respectively. Then, mobility management for VANET with heterogeneous access and simultaneous movement scenario is discussed in Section 5. The open issues for mobility management in vehicular networks are outlined in Section 6. Conclusion is given in Section 7. In this paper, the related terminologies of mobility management are consistent with those in RFC3753 [7] and RFC4885 [8]. In a V2I communications scenario, some ITS applications require Internet access [9] through an infrastructure or Internet gateway. The Internet gateway can provide global addressability and bidirectional Internet connectivity to the mobile nodes in a VANET [10]. In a VANET, mobile nodes can be far away from an Internet gateway, and their traffic can be relayed through intermediate mobile nodes. This is referred to as multihop communications. However, in such a scenario, traditional MIPv6-based mobility management solutions cannot be applied directly since they require a direct connection between a mobile node (MN) and the infrastructure. Therefore, when integrating MIPv6-based solutions into VANETs, many issues arise (e.g., movement detection and handoff decision). To support ITS applications, vehicular area network (VAN) can be established (e.g., fixed vehicle sensors, passengers mobile devices, or even personal area network (PAN) attached to the mobile routers located in the vehicle). In this scenario, network mobility (NEMO) basic support (BS) protocol [3] was introduced to support mobility in a VAN. NEMO is an efficient and scalable scheme since the mobility management is transparent to the mobile devices (i.e., mobile device does not send and receive signaling message directly). However, route optimization was not considered in NEMO BS. Triangular routing in MIPv4 becomes quadrangular routing in NEMO BS. Much work has been done to address the route optimization problem. In addition, a vehicular network can be heterogeneous in which different wireless technologies are integrated into one service. This will enable seamless and high-speed connection, since the MN can select the most suitable network for data transmission [11]. Mobility management should guarantee the reachability to correspondent nodes (CNs) in the Internet as well as the global reachability to mobile nodes in a vehicular network. Therefore, the mobility management has to meet the following requirements [12,13]: (i) Seamless mobility: Mobility of vehicles should be seamless. Accessability and service continuity should be guaranteed regardless of vehicle s location and wireless technology. (ii) Fast and vertical handover: Fast handover is needed for delay-sensitive ITS applications (e.g., safetyrelated). Fast handover is also a crucial requirement for wireless networks with small coverage area 460 Wirel. Commun. Mob. Comput. 2011; 11: John Wiley & Sons, Ltd.

3 K. Zhu et al. Mobility and handoff management in vehicular networks (e.g., WiFi network), since the vehicle spends only short period of time at each point of attachment (e.g., access point). In a heterogeneous wireless environment, vertical handover of the mobile users connections among different wireless technologies must be supported to achieve seamless service. (iii) IPv6 support: The global reachability requires a permanent globally routable IP address for each mobile node. With large address space, IPv6 can support a unique address for each sensor or mobile device in the vehicles. In addition to the advantage in address space, IPv6 also has better support of security and quality of service (QoS) which are the essential requirements of ITS applications. (iv) Multihop communication support: Multihop communication can extend the transmission range of the mobile nodes to reach the destination. Mobility management schemes for vehicular networks need to consider the multihop communications requirements, and therefore, need to be optimized accordingly. (v) Scalability and efficiency: VANETs may be large in size which can consist of hundreds of vehicles and thousands of devices in one network. Furthermore, due to the high frequency of change of the point of attachment, the mobility management scheme must be scalable and efficient to support different types of traffic. In traditional infrastructure-based mobile networks (e.g., cellular system), mobility management can be classified according to the following criteria: (i) Network structure: Mobility management can be classified into mobility management in homogeneous networks [5] and in heterogeneous networks [14]. (ii) Users roaming area: Mobility management can be classified into macro-mobility and micro-mobility management solutions which provide global and local mobility management, respectively. Due to the hierarchical design of global and local management, performance of mobile users can be improved. For macro-mobility management, mobile IPv4 [15] and mobile IPv6 [16] were introduced. For micro-mobility management, fast handover for MIPv6 (FMIPv6) [17], Hierarchical MIPv6 (HMIPv6) [18], Cellular IP [19], HAWAII [20], and Proxy MIPv6 (PMIPv6) [21] were proposed. (iii) Mobile host signaling: Mobility management can be classified into host mobility and NEMO management depending on whether or not the mobile host is involved in signaling for mobility management. If the signaling of mobility management is sent or received by the mobile host, it is called host mobility management, and NEMO management, otherwise. (iv) OSI layers: The type of mobility management can be identified by the OSI layer to which the mobility management belongs. Mobility management can be implemented in data link, network, transport, application layer, or in cross layer fashion. For the review of mobility management for vehicular networks, more than one criterion will be combined to better characterize the schemes. In this case, the mobility management schemes for vehicular networks are first categorized into host mobility and NEMO. Then, the protocol layer criterion is applied for further classification Mobility Management for V2V Communications For VANETs, mobility is managed through route discovery, maintenance, and recovery [6]. Efficient management of vehicular mobility is composed of topology control, location management, and handoff management. (i) Topology management: Topology management can be proactive and reactive. Proactive schemes periodically send signaling messages to explore the topology information. On the other hand, reactive schemes obtain the topology information only when it is needed (e.g., when there is a new mobile node to join the network). Since VANETs can be very large, purely hostbased topology control does not scale well in such networks. The cluster-based topology control can solve this limitation. In this cluster-based topology control, vehicles are grouped into multiple clusters. Head of each cluster is responsible for intra-topology management. These cluster heads coordinate among each other to manage the entire ad hoc network topology. However, due to the high speed and constrained mobility (e.g., moving along a straight road) of vehicles, current clustering schemes developed for MANETs cannot achieve the optimal performance in VANETs and the clusters could be unstable. To address this problem, clustering for open inter-vehicle communication (IVC) networks (COIN) was proposed [22]. The cluster head election is based on mobility information and driver intentions. Besides, COIN can accommodate the oscillatory nature of intervehicle distances. In Reference [23], a prediction-based reactive topology control was proposed. The basic concept of this scheme is to increase the topology maintenance interval and to reduce the periodic beaconing process by mobility prediction. Updates are only needed when the predicted topology information is incorrect. A location-aware framework, i.e., kinetic graph, was introduced to support the use of standard ad hoc network protocols. With kinetic graph, the standard ad hoc protocols can perform efficiently in VANETs. Wirel. Commun. Mob. Comput. 2011; 11: John Wiley & Sons, Ltd. 461

4 Mobility and handoff management in vehicular networks K. Zhu et al. (ii) Location management: With unique mobility characteristics of VANETs, basic ad hoc routing protocols cannot be directly applied to VANETs due to the large latency and overhead [23]. However, geographic routing was shown to be efficient and effective for VANETs. Using geographic routing (e.g., greedy perimeter stateless routing (GPSR) [24], geographical routing algorithm (GRA) [25]), communicating nodes are required to have the location information of each other. Therefore, location management scheme, which deals with the storage, maintenance, and retrieval of mobile node location information, is needed in VANETs [26]. It is worth noting that, the location here refers to geographical location which is not the same as the addressing location in Internet [6]. Location management in VANET can be classified into flooding-based and rendezvous-based approaches [27]. Using a flooding-based approach, the source floods the location query to the entire network which incurs huge overhead. On the other hand, in a rendezvous-based approach, location servers are responsible for locations management. Nodes update their location and query the location of destination from location servers. Many schemes were proposed for location management in MANETs. For example, region-based location service management protocol (RLSMP) which supports both scalability and locality awareness was proposed for VANETs [28]. In RLSMP, message aggregation with the enhancement from geographical clustering was used for both location updating and querying to improve scalability. For locality awareness, local search was used to locate the destination node. (iii) Handoff management: Handoff management in vehicular ad hoc networks is performed by rerouting to construct a new path to the destination. When a mobile node moves, a group of neighbors change and hence the new route of data transfer needs to be established quickly for better handoff performance. Figure 2. Taxonomy of mobility management solutions for vehicular network. Handoff management can be proactive and reactive which uses the same concepts as those in MANET routing. A survey of routing schemes in VANETs can be found in Reference [29]. A simple taxonomy of mobility management solutions for vehicular network is shown in Figure HOST MOBILITY SOLUTIONS FOR VEHICULAR NETWORKS Host mobility management is the scheme in which the mobility of each MN is managed individually. Host mobility management can be designed and implemented at different OSI layers [30,31]. In the following, mobility management schemes implemented at different layers are reviewed and their suitability to vehicular networks (i.e., for V2I communications) is discussed Link Layer When the MN moves between access points (APs) within a common subnet, mobility is managed by link layer protocol [32]. Using WLAN as an example, the movement of mobile nodes between two APs connecting to the same access router (AR) is handled by WLAN specific link layer handoff scheme. In Reference [33], an enhanced linklayer handover scheme in which MN can receive real-time downlink service from base station during handover process was proposed. MAC management message, i.e., Fast DL- MAP-IE, was also defined in Reference [33] to support downlink traffic reception during handoff process and to reduce the downlink transmission delay Network Layer When an MN moves to a different subnet, the original home IPv6 address will be topologically invalid. As a result, mobility management scheme in network layer is required. MIPv6 is the fundamental network layer protocol for host mobility support standardized by IETF. MIPv6 is independent of lower layer and also transparent to upper layer protocols. However, the shortcomings of MIPv6 are the long handoff latency and high packet loss. Besides, MIPv6 is not scalable. With the increasing number of mobile nodes, the signaling overhead increases dramatically. In this case, MIPv6 can be used as the location and path update protocol rather than a handover management protocol [34]. To address these limitations, extensions of MIPv6 (e.g., Fast handover for Mobile IPv6 (FMIPv6) and Hierarchical Mobile IPv6 (HMIPv6)) Mobility Management were proposed. The MN with MIPv6 has a permanent home address (PHA) and a care-of-address (CoA). PHA is used to identify, while CoA is used to locate the MN. When 462 Wirel. Commun. Mob. Comput. 2011; 11: John Wiley & Sons, Ltd.

5 K. Zhu et al. Mobility and handoff management in vehicular networks Figure 3. Illustration of MIPv6. an MN is attached to its home network, conventional IP routing mechanisms can be used to forward packets to mobile nodes. If an MN moves to a foreign network, movement detection is performed by receiving periodic router advertisement from a new AR. A new CoA is obtained according to the advertised foreign subnet prefix through stateful or stateless IPv6 autoconfiguration mechanisms. Then, duplication address detection (DAD) is performed to ensure the uniqueness of the mobile node s local link address as well as its new CoA. Once the address configuration is done, the MN sends a binding update (BU) message to its home agent (HA) to register the new address. Then, packets from correspondent nodes (CNs) with the destination to the mobile node s home address will be intercepted by the HA of the MN and then tunneled to its current CoA. The process of MIPv6 is illustrated in Figure 3. The BU message can also be sent to the MIPv6-enabled CN for direct communication with MN without involving the HA. This is referred to as route optimization. Based on the above mechanism of MIPv6, handoff latency is composed of link layer latency and network layer latency [35]. Link layer latency is the delay due to air link migration from current AP to new AP. Network layer latency is the delay due to movement detection, network authentication, CoA configuration, DAD, and BU messaging. Packets sent to the MN during the handoff period will be lost. Due to large handoff latency and high packet loss, MIPv6 is not suitable for V2I communication especially for real-time services. Furthermore, due to heavy signaling overhead caused by fast moving vehicles, MIPv6 is also not sufficiently scalable for V2I communications. To reduce the packet loss and the handoff latency in MIPv6, FMIPv6 (i.e., fast handover for Mobile IPv6) was proposed. This FMIPv6 addresses the following problems: how to allow an MN to send packets as soon as it detects a new subnet link, and how to deliver packets to an MN as soon as its attachment is detected by the new AR. FMIPv6 uses link layer triggers to predict network layer handoff [17]. FMIPv6 relies on the prediction whose accurate result is, however, difficult to obtain for fast and randomly moving mobile nodes. Many efforts have been devoted to improving reliability of the handoff prediction. In FMIPv6, when an MN receives a link layer trigger, several messages will be exchanged among MN, old AR, and new AR. However, fast moving nodes, i.e., vehicles, may cross the boundary of adjacent cells quickly such that signaling message may not be completely exchanged. To address this problem, an early binding fast handover (EBFH) scheme was proposed for high-speed mobile nodes [36]. In EBFH, a fast moving node can detect the new network by monitoring router advertisement and initiate early BU with its current access router. This scheme can improve the reliability of the prediction for high speed mobile nodes at the cost of larger overhead. To reduce the amount of signaling among the MN, CNs, and HA, HMIPv6 was introduced. In HMIPv6, mobility anchor point (MAP) located in the new network is used as a local HA for mobile nodes. With HMIPv6, an MN has two CoAs, i.e., MIPv6 CoA and regional CoA. A regional care-of-address (RCoA) has a similar role to home address. RCoA has the same subnet prefix as MAP. If an MN moves across subnetworks but within a MAP domain, the MN only needs to register its new CoA with MAP while RCoA does not change. HMIPv6 can also reduce handoff latency due to smaller signaling and shorter path. Besides, as a simple extension to MIPv6, FMIPv6 can be combined with HMIPv6 (FHMIPv6) to further minimize or eliminate the intra MAP domain handover latency. However, FMIPv6 does not support inter MAP domain movement. Therefore, it is not suitable for real-time services in fast moving vehicles. To address this problem, improved fast handover protocol using HMIPv6 (IFHMIPv6) based on IEEE e was proposed [37]. Layer 3 handover messages of the FHMIPv6 are embedded into the layer 2 handover messages so that multiple handover procedures can be performed simultaneously Upper Layers To avoid the change of current Internet architecture, much work has been done towards supporting mobility management in upper layers. For example, at transport layer, mobile Stream Control Transmission Protocol (msctp) [38] was proposed. Due to the multi-homing feature, msctp can be used for Internet mobility support without changing Internet architecture. However, interfaces between transport layer and application layer need to be modified and transport layer of CNs needs to be changed. Another approach is to manage mobility at application layer, for example, Session Initiation Protocol (SIP) [39] and its extensions [40]. However, due to large overhead and long latency, these upper layer mobility management schemes are not suitable for vehicular networks. Wirel. Commun. Mob. Comput. 2011; 11: John Wiley & Sons, Ltd. 463

6 Mobility and handoff management in vehicular networks K. Zhu et al Cross-layer Information from multiple layers can be effectively exchanged to improve performance of mobility management schemes. FMIPv6 is such a cross-layer design which uses link layer information for handover in network layer. In Reference [41], a new cross-layer design for fast IPv6 handover support over IEEE e was proposed. The prediction in FMIPv6 utilizes the information from link layer and physical layer protocols. The proposed scheme provides an interaction between the IP layer and the MAC layer which can improve the performance of FMIPv6 in IEEE e environment. A summary of the host mobility solutions for V2I communications is shown in Table I. The above schemes are used to manage the mobility of a single MN (i.e., vehicle) that can communicate with the Internet gateway directly. However, when vehicles form a VANET connecting to the Internet, the issues related to the integration of MIPv6- based mobility management schemes into VANET become challenging. In a VANET, some vehicles may need multihop communication to the gateway, e.g., vehicles B and C shown in Figure 4. MIPv6 cannot be used directly in such a scenario since a direct connection between gateway and vehicles is not always available [12]. The architecture of Internet and VANET is also different in terms of topology, routing protocols, and mobility management. For example, Internet uses Mobile IP to handle host mobility while in VANET the mobility is managed by ad hoc routing protocols. Some research works have been done for the integration of MANETs and Internet. A survey of this topic can be found in Reference [42]. Since VANET is a special type of MANET, we review the solutions for an integration of MIPv6 with MANETs. The related issues are as follows [43]. Figure 4. Multihop vehicular ad hoc networks (VANETs). (i) Internet gateway discovery: A mobile node detects a foreign network by monitoring router advertisement (RA) from the new Internet gateway. In IPv4 networks, RA message containing foreign subnet prefix is broadcast to the mobile nodes in coverage range of AP. Since IPv6 does not support broadcasting, all-nodes multicast address will be used instead. However, due to the unicast nature of MANET routing protocols, intermediate nodes cannot forward RA messages to the mobile nodes which have more than one hop to reach AP. As a result, these mobile nodes cannot detect their movement and cannot construct CoA. An example is shown in Figure 4. When vehicle D moves from VANET2 to VANET1, the previous link with IGW2 Table I. Host mobility solutions. Scheme in [27] MIPv6 FMIPv6 EBFH FHMIPv6 IFHMIPv6 msctp SIP Scheme in [35] Protocol layers L2 L3 L3 L3 L3 L3 L4 L5 Cross layer Route NO YES YES NO NO NO YES YES NO optimization support Signaling LOW HIGH HIGH HIGH LOW LOW LOW HIGH HIGH overheads Handover latency LOW LARGE LOW LOW LOW LOW LARGE LARGE LOW and packet loss Change to current NO HA HA Generic HA and MAP NO NO HA architecture link layer MAP Change to current YES YES YES YES YES YES YES NO YES protocol stack Technique YES NO NO NO NO YES NO NO YES specific Cross layer information need NO NO YES YES YES YES NO NO YES HA: Home Agent, MAP: Mobility Anchor Point. 464 Wirel. Commun. Mob. Comput. 2011; 11: John Wiley & Sons, Ltd.

7 K. Zhu et al. Mobility and handoff management in vehicular networks is broken. In this case, vehicle D should be aware of the service from IGW1 and establish a new CoA according to the subnet prefix advertised by IGW1. However, vehicle A cannot forward the RA message to other vehicles. Therefore, vehicle D cannot find the service of IGW1 and cannot obtain correct IP address. (ii) Routes to CoA in MANETs: In MIPv6, after acquiring a new CoA, a mobile node (e.g., vehicle D in Figure 4) sends BU message to HA which returns a binding acknowledgement message to mobile node. A bidirectional route between vehicle D and HA is then established. The BU message can be routed from vehicle D to gateway IGW1 using standard ad hoc routing protocol, e.g., optimized link state routing (OLSR) [44], and subsequently routed to HA via IP routing. However, as the CoA assigned by IGW1 is not a routable address in MANET, for routing binding acknowledgement from IGW1 to the CoA of vehicle D, no standard method can be used. (iii) Redundant routes: Using MIPv6, packets with the destination to a mobile node outside home network are intercepted by its HA and then forwarded/tunneled to the current CoA. However, though suitable paths between two mobile nodes within a MANET exist, mobile nodes still communicate through the HAs which causes redundant route and latency. To reduce latency, the mobile nodes can communicate with each other directly without the involvement of HAs. To address the Internet gateway discovery problem in MANET, several approaches were proposed. (i) Ad hoc routing extensions: The ad hoc routing protocol was extended to support mobile IP in MANET. There are two solutions. One solution is to modify the routing protocol to support IP broadcast. The main idea is to use IP broadcast to discover the Internet gateway instead of local link range broadcast used in Mobile IP. However, such extension is not suitable for VANET since it cannot support MIPv6. Also, this extension is not scalable due to the broadcasting of agent advertisements and agent solicitations. The other solution is to extend ad hoc routing protocol to support multicast. The main idea is to use IP multicast in ad hoc network to discover Internet gateway. The problem is that the use of multicast in ad hoc networks is less efficient and scalability is limited [12,45]. (ii) Service discovery: Service discovery protocols can be used for mobile nodes within an ad hoc network to identify and register to Internet gateway. However, since vehicles move at high speed, the performance of service discovery degrades severely in VANETs. Also, this solution is not scalable [12,45]. Due to the limitations of the existing solutions for MANET, Internet gateway discovery solutions for VANETs were proposed with many improvements. DiscoveRy of Internet gateways from VEhicles (DRIVE) based on Service Location Protocol (SLP) was developed [45] with scalability and efficiency enhancement. In addition, it has the ability to select the most suitable Internet gateways among multiple available choices. In Reference [43], the OLSR control packets are used to carry the foreign subnet prefix (i.e., RA/OLSR) and to distribute them to all mobile nodes in the VANET. OLSR control packets are used since they can be flooded to all VANET nodes. The ability of OLSR to construct routes to CoAs was also exploited in Reference [43]. With OLSR, an MN can have its own multiple routable addresses. Once acquiring a new CoA, an MN advertises its CoA as multiple routable addresses. In this way, a route to CoA in VANET can be established using OLSR. To solve the redundant route problem stated above, [43] proposed a new route optimization scheme. The general idea is to add the home address (HoA) of the MN into routing table as a host route. Similar to the route construction mechanism to CoAs, the work in Reference [43] uses OLSR to enable the routability of HoA in MANET. After that, packets to HoA in the same MANET are transmitted to correspondent node (CN) directly. Although this scheme was primarily designed for MANET, it is also applicable to VANET. VANET mobility management scheme (i.e., MMIP6) was proposed in Reference [12] for integration with Internet. This MMIP6 was optimized to be scalable and efficient. The key idea is to combine a proactive service discovery protocol with an optimized mobility management protocol. Although it was designed to support IPv6, MMIP6 is based on the principles of MIPv4 in terms of using both HA and foreign agent. An important feature in MMIP6 is that it only uses a permanent and global IPv6 address rather than CoA when an MN moves into a foreign network. Using Figure 4 as an example, the communication based on MMIP6 works as follows. A CN wants to communicate with vehicle D in VANET1. Packets from CN to vehicle D s home address are received by D s HA. The home agent then tunnels these packets to IGW1 which acts as vehicle D s foreign agent. Finally, IGW1 delivers the decapsulated packets to vehicle D using VANET routing protocol. 4. NETWORK MOBILITY SOLUTIONS FOR VEHICULAR NETWORKS With the proliferation of embedded and portable communication devices, a mobile network can be established in which the vehicles move as a group. NEMO is referred to as the situation in which the mobile network dynamically changes its point of attachment to the Internet. Compared with host mobility, NEMO is more effective and efficient for vehicular networks due to reduced handoff Wirel. Commun. Mob. Comput. 2011; 11: John Wiley & Sons, Ltd. 465

8 Mobility and handoff management in vehicular networks K. Zhu et al. and complexity [46]. Therefore, NEMO management is important for vehicular networks. NEMO BS protocol was standardized by IETF to provide basic NEMO support. In this section, scenarios and characteristics of NEMO are firstly introduced. Then, the requirements and advantages of NEMO are discussed. NEMO BS and some extended NEMO solutions are also reviewed. In addition, we will illustrate two NEMO route optimization solutions and discuss the problem of integrating NEMO with VANET Scenarios and Characteristics of Network Mobility A mobile network consists of one or more mobile IPsubsets formed by one or more mobile routers (MR). This MR, which can change its point of attachment, provides Internet connectivity to mobile network nodes (MNNs) within the network. IETF defines three types of MNNs, i.e., local fixed node (LFN), local mobile node (LMN), and visiting mobile node (VMN). LFN is a fixed node belonging to mobile network without mobility support. LMN and VMN are MNNs with mobility support. The home link of LMN belongs to the mobile network while that of VMN does not. An MNN can be either a host or a router and either fixed or mobile. A mobile network is called nested if there is another attached mobile network inside. The nested mobility is unique for NEMO [47]. For NEMO, there is no size limitation for mobile networks. The simplest network may only consist of a mobile router and MNNs. A mobile network can also consist of hundreds of MR and several nested mobile networks. The common mobile network scenarios NEMO are as follows [48]. (i) Personal area network (PAN): As shown in Figure 5(a), the personal area network (PAN) consists of a mobile phone with cellular and bluetooth interfaces. Using bluetooth, a PDA and a laptop connect to mobile phone which acts as the mobile router to provide Internet connectivity. (ii) Public transportation mobile network: Mobile hotspots deployed in public transportation (e.g., bus or train) can provide Internet connectivity to IP devices (e.g., PDA and laptop). Besides, such a mobile network may be nested if passengers PANs are attached. As shown in Figure 5(b), on the bus, a passenger with PAN is attached to the mobile router (MR) via WLAN interface of the laptop. In this case, the laptop performs as the sub mobile router for PAN which is the nested mobile network. The MR is called the root mobile router. (iii) Intra-vehicle embedded mobile network: To support ITS applications, vehicles are usually equipped with sensors, GPS, and embedded devices as shown in Figure 5(c). These devices can connect to mobile router located in the vehicle for Internet connectivity. Figure 5. Three scenarios of mobile networks. The characteristics of mobile networks in NEMO are as follows: (i) Group of nodes move as a unit: From Internet perspective, the entire mobile network changes its reachability in relation to the fixed Internet topology as a group or unit [3]. (ii) Various sizes and moving speeds: Mobile networks have various sizes and moving speeds. For example, pedestrian with PAN may walk at speed of 5 km/h while access networks accommodating hundreds of devices in a train may move at speed of 100 km/h. (iii) Various mobile network nodes: Mobile network nodes have various types, i.e., mobile host and mobile router, local nodes and visiting nodes, mobility aware nodes (e.g., MIPv6-enabled nodes) and mobility unaware nodes (e.g., standard IPv6 nodes) [49]. (vi) Arbitrary nested level: The mobile network can be nested with arbitrary number of levels. (v) Mobility transparency to mobile network nodes: In most cases, the internal topology of mobile network is relatively stable [48]. For example, a laptop attached to a mobile router in a moving bus will not change its point of attachment frequently. Therefore, the link layer connection of the laptop and the mobile router can be maintained even when the mobile router changes its point of attachment to the Internet. The mobile network nodes do not need to be aware of location change with respect to the Internet Advantages and Requirements of Network Mobility With NEMO, once attached to a mobile network, the mobility management for the MNNs is fully performed by the mobile router. In particular, the mobility management is transparent to the mobile network nodes. The advantages 466 Wirel. Commun. Mob. Comput. 2011; 11: John Wiley & Sons, Ltd.

9 K. Zhu et al. Mobility and handoff management in vehicular networks can be summarized as follows [49]: (i) Scalability: A mobile network may consist of hundreds of MNNs. Without a NEMO solution, these MNNs have to handle mobility independently. For one node, several signaling messages need to be exchanged with the point of attachment. On the other hand, using basic NEMO solutions, mobility is handled only by the mobile router and hence the signaling overhead can be reduced significantly. (ii) Reduced handoff: Due to the relatively stable internal topology of mobile network (e.g., topology among mobile router and MNNs), mobile network nodes do not change their points of attachment and hence can avoid link layer handoff. (iii) Reduced complexity: Mobile network can provide mobility support to standard IPv6. The IP addresses of MNNs will not change even if the mobile router changes its point of attachment. Therefore, the complexity of software and hardware used in MNNs can be reduced. The requirements of NEMO solution can be summarized as follows [48]. (i) Global reachability and session continuity of MNN: This is the fundamental requirement for NEMO. Mobile network nodes must be globally reachable given a permanent IP address. During the movement of mobile router, ongoing sessions of MNNs must be maintained. (ii) Minimum changes: For basic network mobility support, no modifications should be required to any entities other than mobile router and its HA. (iii) Support for different nodes: Basic NEMO solutions must support all types of mobile network nodes mentioned above. (vi) Compatibility: The solutions must be compatible with existing Internet standards. For example, it should not affect the operation of MIPv6 or standard IP addressing and routing schemes. (v) Nested mobility support: The solutions should support mobile network nodes which are located in nested mobile networks at different levels. (vi) Internal configuration transparency: The internal configurations (e.g., topology) should be transparent to the solutions. In other words, the solutions can be applied to mobile networks with arbitrary internal topologies. (vii) Scalability: To support large mobile networks, the solution needs to be scalable. (viii) Security: The solutions must have sufficient protection from the attack Network Mobility Solutions Similar to host mobility, solutions for NEMO can be designed and implemented in different layers. In the following, we mainly focus on network layer and application layer solutions Network layer solutions. Network Mobility Basic Support (NEMO BS) protocol was proposed by IETF to provide basic NEMO support. To minimize the change to existing architecture and to maintain backward compatibility, NEMO BS was designed based on MIPv6 with minimal extensions. Similar to mobile host in MIPv6, mobile router has home address and HA (i.e., HA- MR). NEMO BS specifies operation of mobile router and HA, while the details of mobile network nodes are the same as that in MIPv6 [50]. In NEMO BS, when a visiting MNN connects to the mobile network using MIPv6, the MNN will receive a subnet prefix (i.e., network prefix (MNP)) advertised by the mobile router. Then, the MNN establishes new CoA based on MNP. Once the address configuration is done, the MNN sends a BU message to its HA. The home agent sends binding acknowledge back to finish the location update procedure. When the mobile router changes its point of attachment, it also acquires a CoA from the visiting network and updates the binding cache of its HA. Since the CoAs of MNNs remain unchanged, location update messages do not need to be sent to the HA of the MNNs. Once the binding procedure is completed, a bidirectional tunnel between mobile router and HA is established based on IP-in-IP encapsulation [3]. However, route optimization is not considered in NEMO BS due to the security and incompatibility issues. All packets to and from mobile network nodes need to be tunneled by the HA of the mobile router. Packets from the MNNs to the CNs are encapsulated by mobile router and then tunneled to HA of mobile router. Then the HA decapsulates these packets and forwards them to the destinations. In the opposite direction, packets from CNs to MNNs will be first received by the HA of the mobile router. The home agent tunnels these packets to mobile router which then forwards them to mobile network nodes. However, the binding cache of HA only contains home address of mobile router. The addresses of mobile network nodes are not binded with current CoA of mobile router. As a result, packets cannot be tunneled to the mobile router correctly. To solve this problem, prefix scope binding update (PSBU) [51] was proposed. Using PSBU, the mobile router sends BU message to HA associating with mobile network prefix rather than the home address with current CoA. Having the prefix information, the HA can tunnel packets to the correct mobile router. The handoff performance, signaling and routing overhead of NEMO BS were analyzed in Reference [52]. The results show that NEMO BS itself is not sufficient for seamless handover, and optimization of the protocol is necessary. The components to support handoff in NEMO BS are shown in Figure 6. These components are similar to those in MIPv6. To reduce handoff delay in network attachment Wirel. Commun. Mob. Comput. 2011; 11: John Wiley & Sons, Ltd. 467

10 Mobility and handoff management in vehicular networks K. Zhu et al. Figure 6. Handoff components. process, in Reference [52], fast RA mechanism [53] was adopted to remove the random delay. Optimistic duplication address detection (ODAD) [54] was used to reduce DAD delay. However, the handoff performance of NEMO BS with above optimization is still not sufficient for QoS-sensitive applications. Latency of link layer handoff and NEMO signaling overhead affect the overall performance of mobility management significantly. Novel make-before-break (MBB) handoff scheme was proposed in Reference [52] to reduce handoff delay and packet loss. To enable MBB, two interfaces are needed for simultaneously listening to multiple APs. Besides, a method to minimize the overhead in route optimization was proposed for further performance enhancement [55]. However, this extended scheme can only be applied to the mobile network nodes with host mobility support. A reactive handoff optimization was proposed in Reference [56]. Compared with proactive schemes which detect and predict the movement before the current link is broken, reactive solutions are simpler to implement and more robust. Many limitations are eliminated in reactive solutions (i.e., possible erroneous movement, moving speed limitation, and high signaling overhead). A new cross-layer optimized movement detection procedure and a new DAD scheme were also proposed. A novel reactive handoff procedure combining the above two new schemes was designed in which the movement detection and DAD are performed simultaneously. Compared with existing reactive handoff solutions, the solution in Reference [56] does not rely on prediction information, buffering, bicasting, and soft handover. In NEMO BS, the BU traffics are minimized at the cost of tunneling overhead. To reduce the tunneling overhead, an adaptive NEMO support protocol based on hierarchical mobile IPv6 (HMIPv6) was proposed in Reference [57]. The general idea is to tradeoff between BU traffic and tunneling overhead adaptively. An adaptive BU strategy was deployed based on the session-to-mobility ratio (SMR). This SMR with a predefined threshold was compared with different BU procedures. An optimal threshold for adaptive binding updates was also derived in Reference [57] Application layer solutions. To reduce the deployment cost and avoid the suboptimal routing problems of NEMO BS, SIP-based Figure 7. The system architecture of SIP-NEMO. network mobility (SIP-NEMO) was proposed for NEMO management in application layer. The system architecture of SIP-NEMO is shown in Figure 7. Three types of SIP entities are employed in SIP-NEMO, i.e., SIP network mobility server (SIP-NMS), SIP home server (SIP-HS), and SIP foreign server (SIP-FS). Similar to mobile router in NEMO BS, SIP-NMS is used as a gateway between mobile network and Internet. This SIP-NMS also manages the mobility of entire mobile network. In SIP-NEMO, the mobile network nodes can be both SIP clients and SIP- NMS. Therefore, the concept of nested mobile network in SIP-NEMO is similar to that in NEMO BS. SIP-HS plays a similar role to HA in NEMO BS. The SIP-FS is used for handoff management. SIP-FS will send requests to the corresponding nodes according to Universal Resource Identifier (URI) list. This list is maintained by SIP-NMS when the mobile network changes the point of attachment. When the user agent client (UAC) of SIP moves into a foreign network, a new CoA will be constructed. Then, this CoA registers to the SIP-NMS to obtain a new contact address according to the domain name of SIP-NMS. To perform location update, UAC sends a REGISTER message with its new contact address to its SIP home server. This REGISTER message is translated by SIP-NMS and then forwarded to the SIP home server of UAC. Similar to the BU of mobile router in NEMO BS, when SIP-NMS changes its point of attachment, SIP-NMS sends a REGISTER message with its new CoA in the contact field to its SIP home server. Since mobility management schemes in network layer and higher layers have their own advantages and limitations, an alternative way is to deploy them together in a proper way. HarMoNy, a scheme integrating the host identity protocol (HIP) with NEMO was proposed in Reference [58]. HIP introduces a public key-based host identity name. Since the control signaling and data delivery are separated in SIP. An explicit session establishment 468 Wirel. Commun. Mob. Comput. 2011; 11: John Wiley & Sons, Ltd.

11 K. Zhu et al. Mobility and handoff management in vehicular networks procedure is required in SIP-NEMO. Using Figure 7 as an example, if SIP client UA1 in a mobile network wants to communicate with a corresponding SIP user agent client UA2, it first sends an INVITE message to the SIP-NMS. After translation, the SIP-NMS sends the INVITE message to the SIP home server of UA2 which finally forwards the message to UA2. The dash line in Figure 7 shows the outgoing session establishment. If the session is initiated by UA2, it sends an INVITE message to the SIP home server of UA1. Since the SIP-HS of UA1 registers its current location, the INVITE message is then redirected to the SIP home server of SIP-NMS. Also, SIP home server of SIP-NMS forwards the message to the current CoA of SIP-NMS which forwards the message to UA1. The dotted line in Figure 7 shows this incoming session establishment. Unlike NEMO BS, SIP-NEMO routes the packet directly between SIP clients [59]. In addition, as an application layer solution, SIP-NEMO has the advantage since SIP-NEMO can be deployed without modifications to the Internet architecture. However, the handoff delay in SIP-NEMO can be large due to longer message length. A comparative study of NEMO BS and SIP-NEMO was presented in Reference [50]. A summary of NEMO solutions is shown in Table II Solutions for Route Optimization Since basic support protocol of NEMO does not consider route optimization, all packets need to be tunneled by the HA of mobile router even when a shorter path exists. This scheme causes suboptimal route problems especially for multilevel nested mobile network [60]. NEMO BS uses IP-in-IP tunneling. In particular, packets are encapsulated by all upper level MR and then tunneled to their HAs. Therefore, the delay and overhead become large with increasing number of nested levels. Besides, since all packets of mobile network nodes must go through HAs of upper MR, this HA may be congested and become the bottleneck. Also, if the HA is unavailable, the entire path will be cut, and the communication becomes unreachable. Using three level nested mobile networks shown in Figure 8 as an example, MNN is a visiting mobile node with MIPv6 support. This node is attached to MR3 and the CN is a standard IPv6 node. MR1, the parent-mr of Figure 8. Nested mobile network. MR2, is the top level mobile router of the mobile network. MR3 is attached to MR2. The packets from CN to MNN are first sent to home agent of MNN (HA-MNN). HA- MNN encapsulates the packets and then tunnels them to the home agent of MR3 (i.e., HA-MR3). HA-MR3, HA-MR2, and HA-MR1 handle the packets in a similar way. While receiving these multilevel encapsulated packets, MR1, MR2, and MR3 decapsulate them sequentially. Finally, MR3 forwards the decapsulated packets to MNN. The main purposes of route optimization for NEMO are to avoid data packets passing through the HA of mobile router and to reduce the number of additional IPv6 headers added to the original packets. MIPv6 route optimization does not work in mobile networks due to the compulsory MR-HA tunneling. A simple extension of MIPv6 route optimization for NEMO is to use NEMO prefix option to inform the CN about the location of the mobile network prefix (MNP) [61]. However, security is a major problem and the modification to the CNs is required. In the following, two NEMO route optimization solutions are reviewed. It is worth noting that, although route optimization schemes can yield some benefits, the tradeoffs (i.e., additional signaling overhead, increased protocol complexity, and processing load) need to be taken into account [62]. Mobile IPv6 route optimization for NEMO (MIRON) was proposed in Reference [61]. MIRON combines two different operation modes applied to different types of mobile network nodes. Specifically, for the nodes without host mobility support (e.g., standard IPv6 nodes), the Table II. Network mobility solutions. NEMO BS Scheme in [45] Scheme in [49] Scheme in [50] SIP-NEMO Protocol layers L3 L3 L3 L3 L5 Route optimization support NO YES YES YES YES Signaling overheads HIGH LOW LOW Adaptive HIGH Handover latency and packet loss HIGH LOW LOW LOW HIGH Change to current architecture HA HA HA and MAP HA and MAP NO Change to current protocol stack YES YES YES YES NO Technique specific NO NO NO NO NO Cross layer information need NO NO YES NO NO HA: Home Agent, MAP: Mobility Anchor Point. Wirel. Commun. Mob. Comput. 2011; 11: John Wiley & Sons, Ltd. 469

12 Mobility and handoff management in vehicular networks K. Zhu et al. mobile router which works as Proxy-MR is responsible for all the mobility and route optimization management. While for the nodes with standard MIPv6 support, MIRON uses a PANA (Protocol for carrying Authentication for Network Access) and Dynamic Host Configuration Protocol (DHCP) based address delegation mechanism to enable self management of mobility and route optimization. Such combination guarantees route optimization for all types of nodes and network topology [61]. MIRON has a deployment advantage in that modification is needed only in MR. Evaluation results show that MIRON can significantly improve the performance over NEMO BS in terms of larger TCP throughput and smaller overhead. The Route Optimization solution for nested mobile networks using Tree Information Option (ROTIO) was proposed in Reference [63]. ROTIO extends the NEMO BS by modifying BU and router advertisement messages. Two BU messages are used by nested mobile router. One message sent to the Top Level Mobile Router (TLMR) contains routing information of TLMR, and the other sent to the HA of nested mobile router contains the home address of TLMR. Therefore, only two extra entities, i.e., home agents of the mobile router and TLMR, are required in the path from CN to the MNNs. Locations of MNN and CN are incorporated in ROTIO. Basic ROTIO scheme is used to optimize the route between MNN and CN which are not located in the same mobile network. The extended ROTIO scheme is used for intra-nemo routing optimization. Besides, location privacy and mobility transparency are guaranteed in ROTIO. Many other related works were introduced for NEMO route optimization [64--72]. An analytical framework with performance metrics (e.g., transmission latency, memory usage, and BU s occurrence number) was proposed in Reference [73]. Detailed classification, evaluation, and analysis can be found in Reference [74] VANETs with Network Mobility Similar to MIPv6, NEMO BS is designed for mobile network having direct communication link with an Internet access infrastructure. However, multihop communication is not supported in this scheme. Mobile routers in vehicles may form VANETs. To guarantee the consistent reachability to Internet from mobile networks via both direct and indirect links, it is necessary to integrate VANETs with NEMO. In addition, such integration can also be used for route optimization [75]. The works in Reference [76,77] realized route optimization in terms of delay and bandwidth by switching from NEMO to MANET. In Reference [75], MANET with NEMO was evaluated and the results show performance improvement. From security perspective, [78] proposed VARON, a Vehicular Ad hoc Route Optimization solution for NEMO using MIPv6 security concept to provide the same level of security of current Internet. MANEMO is the concept to integrate the MANET with NEMO [79], which combines the advantages of both the schemes. Since the schemes for integrating VANET with NEMO are inherited from MANEMO, it is worth reviewing MANEMO solutions. The MANEMO solutions can be designed based on MANET or NEMO, i.e., MANET-centric approaches and NEMO-centric approaches, respectively. According to the definitions in Reference [9], in MANETcentric solution, NEMO techniques (e.g., NEMO BS) is applied directly to MANETs. Similar to the aforementioned multihop communication methods in the former section, foreign subnetwork prefix advertising problems were addressed by specific extensions of MANET routing protocols. The main idea of this NEMO-centric solution is to use at least one intermediate mobile router along the multihop path between the mobile router and the infrastructure for relaying packets. In NEMO-centric solutions, NEMO techniques are used to provide and maintain Internet connectivity while MANET protocols are used to optimize the routing within a mobile network [9]. NEMO-centric solutions can only be used in networks with hierarchical topologies (e.g., nested mobile networks). Considering a VANET scenario, a comparison of MANET-centric and NEMO-centric approaches with respect to VANET specific requirements was also performed in Reference [9] based on economic, functional and performance criteria. The conclusion is that the MANET-centric approach outperforms the NEMO-centric approach for VANET in terms of complexity, routing performance, and cost. In the following, two MANETcentric solutions are reviewed. (i) Unified MANEMO architecture (UMA) is a protocol combining the functionality of the optimized link state routing (OLSR) protocol and NEMO BS [79]. In UMA, every UMA-enabled mobile router with direct connection to the Internet is required to establish a MR-HA bidirectional tunnel and acts as a gateway mobile router in the MANET. Then, the gateway mobile router advertises its reachability to the Internet via OLSR host and network association (HNA) messages. Therefore, once receiving such HNA messages, the mobile router can start BU process via the gateway mobile router. (ii) A solution applying NEMO BS to VANETs was proposed in Reference [80]. The network layer is divided into two sublayers. In this solution, topology-based routing or geographical routing protocol is used as VANET routing protocol while NEMO BS runs on top of this protocol to support mobility without any modification. This solution was designed specifically for IEEE based VANET based on Car2Car Communication Consortium (i.e., C2C-CC [2]) system architecture. The laboratory measurements show the effectiveness of the solution for highly dynamic vehicular networks. The above solutions are based on NEMO BS. Terminal mobility with network mobility support (PTEN), a 470 Wirel. Commun. Mob. Comput. 2011; 11: John Wiley & Sons, Ltd.

13 K. Zhu et al. Mobility and handoff management in vehicular networks terminal-assisted NEMO management scheme proposed in Reference [81], does not use MIPv6-based NEMO BS schemes. Instead, PTEN uses URI to locate a mobile network node. Directory service is used to map URI to IP address. An IP-IP address mapping scheme, an IPv6 header extension on IP packets, and an IP address redirection scheme on the MRs were implemented in Reference [81]. 5. MOBILITY MANAGEMENT FOR HETEROGENEOUS WIRELESS ACCESS Current mobile nodes or mobile routers in vehicles can be equipped with multiple radio access interfaces for different wireless networks (e.g., 3G, WiMAX, and WiFi). This is referred to as heterogeneous access. For session continuity and better wireless access performance, seamless vertical handoff (i.e., handoff among different wireless technologies) should be performed. Besides, for load balancing purpose, mobile vehicular nodes should be able to access multiple networks simultaneously. To achieve the optimal performance, efficient mobility management schemes are required for vehicular networks in presence of heterogeneous wireless access. Recent research has taken advantages of multihoming, which enables a mobile node to use multiple access networks simultaneously to perform smooth vertical handoff. An analysis of multihoming in network mobility support can be found in Reference [82]. Many works were done for host mobility in heterogeneous wireless networks. However, little attention has been paid on NEMO. The main challenge is due to the heterogeneity of access scenarios in mobile networks. A mobile network may have one mobile router with multiple access interfaces, or the heterogeneity may arise from several MRs in a mobile network each with a different access interface. A solution for multiple MRs was proposed in Reference [83]. This solution consists of mobile DHCPv6 agents and a handoff management center (HMC). Location management and forward loss recovery were implemented based on mobility prediction. Cooperative mobile router-based handover (CoMoRoHo) was proposed in Reference [84]. CoMoRoHo uses multihoming techniques to reduce handoff latency and packet loss for long-vehicular multihomed mobile networks. Multiple MRs connected to different access networks can also cooperate during handoff to reduce packet loss due to handoff latency and overlapped reception. In Reference [85], Mobile IP based mobility management architecture for highly mobile users and vehicular networks was proposed. This architecture focuses on network selection and timely handoff. The handoff decisions are based on network layer metrics and the frequency of BU messages is dynamically adjusted according to the speed of the mobile node. The simultaneous mobility is referred to the situation when both mobile nodes move to other networks simultaneously [86]. Due to the high handoff rate caused by highly mobile vehicles, simultaneous mobility will occur frequently in vehicular networks. The mobility management schemes with route optimization that sends location binding updates to CNs (e.g., MIPv6, SIP-NEMO) are vulnerable to such simultaneous mobility problem. In particular, when two communicating mobile nodes change their points of attachment at the same time, they both send BU messages to each other. However, these two BU messages are sent to their outdated addresses and the messages will be lost. An analytical framework and solution of simultaneous mobility were proposed in Reference [86]. Besides, Reference [86] also proposed and compared different solutions to support simultaneous mobility for MIPv6, SIP based mobility management, and MIP location registration. The solutions for simultaneous mobility are broadly divided into receiver-side and sender-side mechanisms based on the entity which is responsible for sending a particular BU message. Both receiver-side and sender-side mechanisms can be further categorized into timer-based retransmissions, forwarding, pro-active forwarding, redirecting, and proactive redirecting. Details can be found in Reference [86]. Network mobility solutions also have to take simultaneous mobility issue into account. A proxy-aided simultaneous handover (PASH) mechanism for mobile networks in vehicles was proposed in Reference [87]. This PASH mechanism aims to solve addressing problem resulting from simultaneous handover in SIP-NEMO. Besides, a Fast Route/local route re-establishment (FREE) algorithm was developed. The basic concept is to improve the speed of reestablishment process of the optimized routing path and to ensure that the signaling messages will be successfully received. 6. OPEN RESEARCH ISSUES Future vehicular networks will provide seamless services to the mobile users. Despite the existing research efforts, there are still many open research issues related to mobility management for vehicular networks. (i) Quality of service (QoS) issues: QoS requirements of vehicular applications pose great challenges to mobility management design. Safety applications have higher priority than non-safety applications, and such priority should be guaranteed even if handoff is performed. For multimedia applications, handoff latency should be minimized. For vertical handoff, a QoS mapping scheme for different wireless technologies may be required. Besides, for both horizontal and vertical handoff, a resource allocation mechanism for handed over sessions is needed to meet QoS requirements. Scalability and resource utilization are important factors when designing such resource allocation mechanisms. (ii) Access selection issues: Vehicular mobile nodes with multiple access interfaces need to perform Wirel. Commun. Mob. Comput. 2011; 11: John Wiley & Sons, Ltd. 471

14 Mobility and handoff management in vehicular networks K. Zhu et al. access selection in heterogeneous environment. Many factors (e.g., cost, bandwidth, and delay) and their weights for decision need to be defined. Access selection is also related to handoff decision. If multiple access networks are selected simultaneously, an efficient load balancing scheme is desirable. Besides, when integrating VANETs with Internet, multiple Internet gateways (e.g., a direct Internet gateway and an indirect Internet gateway) may be available for some nodes. Internet gateway selection is also required for the vehicular nodes. (iii) Issues related to mobility model: Performance evaluation is required for both designing new protocols and applying extensions of existing protocols for vehicular networks. Accurate mobility model is required for performance evaluation of vehicular networking protocols. Traditional mobility models (e.g., random waypoint) are not suitable for vehicular networks since they assume a random direction selection and random speed. However, mobility of vehicles is constrained by prebuilt roads, vehicle speed, and driving regulations. A flexible but realistic mobility model for vehicular networks is needed. (iv) Ad hoc routing issues: Mobility was not considered in ad hoc routing protocols. For both V2I and V2V mobility management solutions, the handoff performance degrades severely with increasing number of hops. Mobility-aware vehicular ad hoc routing protocol is required to facilitate fast handoff. (v) Transport and application-layer performance issues: Performance of transport and application layer protocols (e.g., TCP and UDP) need to be optimized for vehicular networks. Effects of mobility management schemes on transport and application layer performances are worth investigation. 7. CONCLUSIONS In this paper, we have presented a comprehensive survey of mobility management solutions for vehicular networks. We have classified the mobility management solutions for vehicular networks based on vehicle-to-vehicle (V2V) or vehicle-to-infrastructure (V2I) communications. The traditional Internet and MANET mobility management techniques and their suitability to vehicular networks have been discussed. Existing works for both V2I and V2V mobility management have been reviewed. Several open research issues have been also outlined. ACKNOWLEDGMENTS This work was done in the Centre for Multimedia and Network Technology (CeMNet) of the School of Computer Engineering, Nanyang Technological University, Singapore. This work was supported in part by the AUTO21 NCE research grant for the project F303-FVT, and the MKE (Ministry of Knowledge Economy), Korea under the ITRC (Information Technology Research Center) support program supervised by the IITA (Institute of Information Technology Assessment) 2009-C REFERENCES 1. Hartenstein H, Laberteaux KP. A tutorial survey on vehicular ad hoc networks. IEEE Communications Magazine 2008; 46(6): Manifesto for the car-to-car communication consortium, September Devarapalli V, Wakikawa R, Petrescu A, Thubert P. Network mobility (NEMO) basic support protocol. RFC 3963, January Jiang H, Wang P, Poor HV, Zhuang W. Voice service support in mobile ad hoc networks. In Proceedings of IEEE GLOBECOM, 2007; Akyildiz IF, McNair J, Ho JSM, Uzunalioglu H, Wang W. Mobility management in next-generation wireless systems. Proceedings of the IEEE 1999; 87(8): Xie J, Wang XD. A survey of mobility management in hybrid wireless mesh networks. IEEE Network 2008; 22(6): Manner J, Kojo M. Mobility related terminology. RFC 3753, June Ernst T, Lach HY. Network mobility support terminology. RFC 4885, July Baldessari R, Festag A, Abeille J. Nemo meets vanet: a deployability analysis of network mobility in vehicular communication. In Proceedings of ITST, 2007; Perkins CE, Malinen JT, Wakikawa R, Nilsson A, Tuominen A. Internet connectivity for mobile ad hoc networks. Wireless Communications and Mobile Computing 2002; 2(5): Gustafsson E, Jonsson A. Always best connected. IEEE Wireless Communications Magazine 2003; 10(1): Bechler M, Wolf L. Mobility management for vehicular ad hoc networks. In Proceedings of VTC 2005-Spring, Vol. 4, 2005; Ernst T, Uehara K. Connecting automobiles to the internet. In Porceedings of ITST: 3rd International Workshop on ITS Telecommunications, Akyildiz IF, Xie J, Mohanty S. A survey of mobility management in next-generation all-ip-based wireless systems. IEEE Wireless Communications Magazine 2004; 11(4): Perkins C. IP Mobility support for IPv4. RFC 3220, January Wirel. Commun. Mob. Comput. 2011; 11: John Wiley & Sons, Ltd.

15 K. Zhu et al. Mobility and handoff management in vehicular networks 16. Johnson D, Perkins C, Arkko J. Mobility support in IPv6. RFC 3775, June Koodli R. Fast Handovers for Mobile IPv6. RFC 4068, July Soliman H, Castelluccia C, Malki KE, Bellier L. Hierarchical mobile IPv6 mobility management (HMIPv6). RFC 4140, August Campbell AT, Gomez J, Kim S, Valko AG, Wan CY, Turanyi ZR. Design, implementation, and evaluation of cellular ip. IEEE Personal Communications 2000; 7(4): Ramjee R, Varadhan K, Salgarelli L, Thuel SR, Wang SY, La Porta T. Hawaii: a domain-based approach for supporting mobility in wide-area wireless networks. IEEE/ACM Transactions on Networking 2002; 10(3): Gundavelli S, Leung K, Devarapalli V, Chowdhury K, Patil B. Proxy mobile IPv6. RFC 5213, August Blum J, Eskandarian A, Hoffman L. Mobility management in ivc networks. In Proceedings of IEEE Intelligent Vehicles Symposium, 2003; Harri J, Bonnet C, Filali F. Kinetic mobility management applied to vehicular ad hoc network protocols. Computer Communications 2008; 31(12): Karp B, Kung HT. Gpsr: greedy perimeter stateless routing for wireless networks. In Proceedings of MOBICOM, 2000; Jain R, Puri A, Sengupta R. Geographical routing using partial information for wireless ad hoc networks. IEEE Personal Communications 2001; 8(1): Haas ZJ, Liang B. Ad hoc mobility management with uniform quorum systems. IEEE/ACM Transactions on Networking 1999; 7(2): Cheng H, Cao JN, Chen HH, Zhang HK. Grls: groupbased location service in mobile ad hoc networks. IEEE Transactions on Vehicular Technology 2008; 57(6): Saleet H, Langar R, Basir O, Boutaba R. Proposal and analysis of region-based location service management protocol for vanets. In Proceedings of IEEE GLOBE- COM, 2008; Li F, Wang Y. Routing in vehicular ad hoc networks: a survey. IEEE Vehicular Technology Magazine 2007; 2(2): Banerjee N, Wu W, Das SK. Mobility support in wireless internet. IEEE Wireless Communications Magazine 2003; 10(5): Eddy WM. At what layer does mobility belong? IEEE Communications Magazine 2004; 42(10): Montavont N, Noel T. Handover management for mobile nodes in ipv6 networks. IEEE Communications Magazine 2002; 40(8): Choi S, Hwang G, Kwon T, Lim AR, Cho DH. Fast handover scheme for real-time downlink services in ieee e bwa system. In Proceedings of IEEE VTC Spring, Vol. 3, 2005; Dimopoulou L, Leoleis G, Venieris IO. Fast handover support in a wlan environment: challenges and perspectives. IEEE Network 2005; 19(3): Mussabbir QB, Yao WB, Niu ZY, Fu XM. Optimized fmipv6 using ieee mih services in vehicular networks. IEEE Transactions on Vehicular Technology 2007; 56(6): Kim H, Kim Y. An early binding fast handover for highspeed mobile nodes on mipv6 over connectionless packet radio link. In Proceedings of Seventh ACIS International Conference on Software Engineering, Artificial Intelligence, Networking, and Parallel/Distributed Computing, 2006; Koo G, Yu K, Noh M, Mun Y. Improved Fast Handover Protocol Using HMIPv6 Based on IEEE e Network. Springer, Koh SJ, Chang MJ, Lee M. msctp for soft handover in transport layer. IEEE Communications Letters 2004; 8(3): Rosenberg J, Schulzrinne H, Camarillo G, et al. SIP: session Initiation Protocol. RFC 3261, June Schulzrinne H, Volz B. Dynamic host configuration protocol (DHCPv6) options for session initiation protocol (SIP) servers. RFC 3319, July Han YH, Jang H, Choi JH, Park BJ, McNair J. A crosslayering design for ipv6 fast handover support in an ieee e wireless man. IEEE Network 2007; 21(6): Ammari HM. A survey of current architectures for connecting wireless mobile ad hoc networks to the internet: research articles. International Journal of Communication Systems 2007; 20(8): Kazuya1 M, Hiroki1 S, Junji Y, et al. Integration of Mobile IPv6 into Mobile Ad-Hoc Network Systems. Springer, Clausen T, Jacquet P. Optimized link state routing protocol (OLSR). RFC 3626, October Bechler M, Wolf L, Storz O, Franz WJ. Efficient discovery of internet gateways in future vehicular communication systems. In Proceedings of IEEE VTC 2003-Spring, Vol. 2, 2003; Ernst T. The information technology era of the vehicular industry. SIGCOMM Computer Communications Review 2006; 36(2): Lach HY, Janneteau C, Petrescu A. Network mobility in beyond-3g systems. IEEE Communications Magazine 2003; 41(7): Ernst T. Network mobility support goals and requirements. RFC 4886, July2007. Wirel. Commun. Mob. Comput. 2011; 11: John Wiley & Sons, Ltd. 473

16 Mobility and handoff management in vehicular networks K. Zhu et al. 49. Perera E, Sivaraman V, Seneviratne A. Survey on network mobility support. SIGMOBILE Mobile Computer Communications Review 2004; 8(2): Pack SH, Shen XS, Mark JW, Pan JP. A comparative study of mobility management schemes for mobile hotspots. In Proceedings of IEEE WCNC, 2007; Ernst T, Olivereau A, Castelluccia C, Lach HY. Mobile networks support in mobile ipv6 (prefix scope binding updates). IETF Draft, March Petander H, Perera E, Lan KC, Seneviratne A. Measuring and improving the performance of network mobility management in ipv6 networks. IEEE Journal on Selected Areas in Communications 2006; 24(9): Kempf J, Khalid M, Pentland B. Ipv6 fast router advertisement. IETF Draft, June Moore N. Optimistic duplicate address detection for ipv6. IETF Draft, December Perera E, Seneviratne A, Sivaraman V. Optinets: an architecture to enable optimal routing for network mobility. In Proceedings of International Workshop on Wireless Ad-Hoc Networks, 2004; Han YH, Choi J, Hwang SH. Reactive handover optimization in ipv6-based mobile networks. IEEE Journal on Selected Areas in Communications 2006; 24(9): Pack S, Kwon T, Choi Y, Park E. An adaptive network mobility support protocol in hierarchical mobile ipv6 networks. IEEE Transactions on Vehicular Technology 2009; 58(7): Herborn S, Haslett L, Boreli R, Seneviratne A. Harmony - hip mobile networks. In Porceedings of IEEE VTC Spring, Vol. 2, 2006; Huang CM, Lee CH, Zheng JR. A novel sip-based route optimization for network mobility. IEEE Journal on Selected Areas in Communications 2006; 24(9): Ng C, Thubert P, Watari M, Zhao F. Network mobility route optimization problem statement. RFC 4888, July Calderon M, Bernardos CJ, Bagnulo M, Soto I, de la Oliva A. Design and experimental evaluation of a route optimization solution for nemo. IEEE Journal on Selected Areas in Communications 2006; 24(9): Ng C, Zhao F, Watari M, Thubert P. Network mobility route optimization solution space analysis. RFC 4889, July Cho H, Kwon T, Choi Y. Route optimization using tree information option for nested mobile networks. IEEE Journal on Selected Areas in Communications 2006; 24(9): Wakikawa R, Koshiba S, Uehara K, Murai J. Orc: optimized route cache management protocol for network mobility. In Proceedings of 10th International Conference on Telecommunications, Vol. 2, 2003; Thubert P, Molteni M. Ipv6 reverse routing header and its application to mobile networks. IETF Draft, February Ng C, Hirano J. Securing nested tunnels optimization with access router option. IETF Draft, January Na J, Cho S, Kim C, Lee S, Kang H, Koo C. Secure nested tunnels optimization using nested path information. IETF Draft, March Cho H, Paik EK, Choi Y. R-bu: recursive binding update for route optimization in nested mobile networks. In Proceedings of VTC 2003-Fall, Vol. 3, 2003; Ohnishi H, Sakitani K, Takagi Y. HMIP based route optimization method in a mobile network. IETF Draft, April Lee D, Lim K, Kim M. Hierarchical route optimization for nested mobile network. In Proceedings of 18th International Conference on Advanced Information Networking and Applications, Vol. 1, 2004; Thubert P, Kniveton TJ. Mobile network prefix delegation. IETF Draft, August Droms R, Thubert P, Dupont F, Haddad W. DHCPv6 prefix delegation for nemo. IETF Draft, March Lim HJ, Lee DY, Kim TK, Chung TM. A model and evaluation of route optimization in nested nemo environment. IEICE Transactions on Communications 2005; E88-B(7): Lim HJ, Kim M, Lee JH, Chung TM. Route optimization in nested nemo: classification, evaluation and analysis from nemo fringe stub perspective. IEEE Transactions on Mobile Computing 2009; 99(1): Tsukada M, Mehani O, Ernst T. Simultaneous usage of nemo and manet for vehicular communication. In Proceedings of the 4th International Conference on Testbeds and Research Infrastructures for the Development of Networks & Communities, ICST (Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering), 2008; Lorchat J, Uehara K. Optimized inter-vehicle communications using nemo and manet. In Proceedings of the 3rd Annual International Conference on Mobile and Ubiquitous Systems - Workshops, 2006; Wakikawa R, Okada K, Koodli R, Nilsson A. Design of vehicle network: mobile gateway for manet and nemo converged communication. In Proceedings of the 2nd ACM International Workshop on Vehicular Ad Hoc Networks, 2005; Wirel. Commun. Mob. Comput. 2011; 11: John Wiley & Sons, Ltd.

17 K. Zhu et al. Mobility and handoff management in vehicular networks 78. Bernardos CJ, Soto I, Calderón M, Boavida F, Azcorra A. Varon: vehicular ad hoc route optimisation for nemo. Computer Communications 2007; 30(8): McCarthy B, Edwards C, Dunmore M. Using nemo to extend the functionality of manets. In Proceedings of IEEE ICC Workshops, 2008; Roberto B, Zhang WH, Festag A, Long L. A manetcentric solution for the application of nemo in vanet using geographic routing. In Proceedings of the 4th International Conference on Testbeds and Research Infrastructures for the Development of Networks & Communities, 2008; Lim TM, Tantra JW, Lee BS, Yeo CK, Xia Y. Terminalassisted network mobility management. In Proceedings of IEEE WCNC, 2007; Ng C, Ernst T, Paik E, Bagnulo M. Analysis of multihoming in network mobility support. RFC 4980, October Paik EK, Choi Y. Seamless mobility support for mobile networks on vehicles across heterogeneous wireless access networks. In Proceedings of IEEE VTC, Vol. 4, 2003; Kafle VP, Kamioka E, Yamada S. Comoroho: cooperative mobile router-based handover scheme for long-vehicular multihomed networks. IEICE Transactions on Communications 2006; E89-B(10): Andersson K, Ahlund C, Gukhool BS, Cherkaoui S. Mobility management for highly mobile users and vehicular networks in heterogeneous environments. In Proceedings of the 33rd IEEE Conference on Local Computer Networks, 2008; Wong K, Dutta A, Schulzrinne H, Young K. Simultaneous mobility: analytical framework, theorems and solutions. Wireless Communications and Mobile Computing 2007; 7(5): Chiang WK, Chang WY, Liu LY. Simultaneous handover support for mobile networks on vehicles. In Proceedings of IEEE WCNC, 2008; AUTHORS BIOGRAPHIES Kun Zhu is a Ph.D. student in School of Computer Engineering, Nanyang Technological University, Singapore. He received the B.Eng. and M.Eng. degrees both in Computer Engineering from Beijing Jiaotong University, Beijing, China, in 2005 and 2007, respectively. His research interests are in the area of mobility management and network selection in heterogeneous wireless networks. Dusit Niyato is currently an assistant professor in the School of Computer Engineering, at the Nanyang Technological University, Singapore. He obtained his Bachelor of Engineering in Computer Engineering from King Mongkut s Institute of Technology Ladkrabang (KMITL), Bangkok, Thailand. He received Ph.D. in Electrical and Computer Engineering from the University of Manitoba, Canada. His research interests are in the area of radio resource management in cognitive radio networks and broadband wireless access networks. Ping Wang (M 09) received the Ph.D. degree in electrical engineering in 2008 from the University of Waterloo, Canada. She is currently an assistant professor at School of Computer Engineering, Nanyang Technological University, Singapore. Her current research interests include QoS provisioning and resource allocation in multimedia wireless communications. She was a corecipient of a Best Paper Award from IEEE ICC She is an Editor of EURASIP Journal on Wireless Communications and Networking, International Journal of Communication Systems, and International Journal of Ultra Wideband Communications and Systems. Ekram Hossain is currently an Associate Professor in the Department of Electrical and Computer Engineering at University of Manitoba, Winnipeg, Canada. Dr Hossain s current research interests include design, analysis, and optimization of wireless communication networks and cognitive radio systems. He is a co-author/co-editor for the books Dynamic Spectrum Access and Management in Cognitive Radio Networks (Cambridge University Press, 2009), Heterogeneous Wireless Access Networks (Springer, 2008), Introduction to Network Simulator NS2 (Springer, 2008), Cognitive Wireless Communication Networks (Springer, 2007), and Wireless Mesh Networks: Architectures and Protocols (Springer, 2007). Dr Hossain serves as an Editor for the IEEE Transactions on Mobile Computing, the IEEE Transactions on Wireless Communications, the IEEE Transactions on Vehicular Technology, IEEE Wireless Communications, IEEE Communications Surveys and Tutorials and several other international journals. He served as a guest editor for the special issues of IEEE Communications Magazine (Cross-Layer Protocol Engineering for Wireless Mobile Networks, Advances in Mobile Multimedia) and IEEE Wireless Communications (Radio Resource Management and Protocol Engineering for IEEE ). He served as a technical program cochair for the IEEE Globecom 2007, IEEE WCNC 2008, Wirel. Commun. Mob. Comput. 2011; 11: John Wiley & Sons, Ltd. 475

18 Mobility and handoff management in vehicular networks K. Zhu et al. IEEE VTC 2008-Fall, and QShine 2008: International Conference on Heterogeneous Networking for Quality, Reliability, Security, and Robustness. Dr Hossain served as the technical program chair for the workshops on Cognitive Wireless Networks (CWNets 2007) and Wireless Networking for Intelligent Transportation Systems (WiN- ITS 2007) held in conjunction with QShine 2007 during August 14 17, in Vancouver, Canada, and the First IEEE International Workshop on Cognitive Radio and Networks (CRNETS 2008) in conjunction with IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC 2008). He served as the technical program co-chair for the Symposium on Next Generation Mobile Networks (NGMN 06), NGMN 07, NGMN 08, NGMN 09 held in conjunction with International Wireless Communications and Mobile Computing Conference (IWCMC 06), IWCMC 07, IWCMC 08, and IWCMC 09. Dr Hossain has several research awards to his credit which include Lucent Technologies Research Award for contribution to IEEE International Conference on Personal Wireless Communications (ICPWC 97), and the Best Student-paper Award in IWCMC 06. He is a registered Professional Engineer (P.Eng.) in the Province of Manitoba, Canada. Dr Hossain is a Senior Member of the IEEE. Dong In Kim received the B.S. and M.S. degrees in Electronics Engineering from Seoul National University, Seoul, Korea, in 1980 and 1984, respectively, and the M.S. and Ph.D. degrees in Electrical Engineering from University of Southern California (USC), Los Angeles, in 1987 and 1990, respectively. From 1984 to 1985, he was a Researcher with Korea Telecom Research Center, Seoul. From 1986 to 1988, he was a Korean Government Graduate Fellow in the Department of Electrical Engineering, USC. From 1991 to 2002, he was with the University of Seoul, Seoul, leading the Wireless Communications Research Group. From 2002 to 2007, he was a tenured Full Professor in the School of Engineering Science, Simon Fraser University, Burnaby, BC, Canada. From 1999 to 2000, he was a Visiting Professor at the University of Victoria, Victoria, BC. Since 2007, he has been with Sungkyunkwan University (SKKU), Suwon, Korea, where he is a Professor and SKKU Fellow in the School of Information and Communication Engineering. Since 1988, he is engaged in the research activities in the areas of wideband wireless transmission and access. His current research interests include cooperative relaying and base station (BS) cooperation, multiuser cognitive radio networks, advanced transceiver design, and cross-layer design. Dr Kim was an Editor for the IEEE Journal on Selected Areas in Communications: Wireless Communications Series and also a Division Editor for the Journal of Communications and Networks. He is currently an Editor for Spread Spectrum Transmission and Access for the IEEE Transactions on Communications and an Area Editor for Transmission Technology III for the IEEE Transactions on Wireless Communications. He also serves as Co-Editor-in- Chief for the Journal of Communications and Networks. 476 Wirel. Commun. Mob. Comput. 2011; 11: John Wiley & Sons, Ltd.