Enhancing network access for highly mobile nodes

With the proliferation of mobile devices, the demand for mobile computing has called for better mobile network access technology. In particular, network access for highly mobile nodes such as vehicles is envisioned to have great significance to future transportation systems and help to foster many attractive applications from disseminating safety information among vehicles to downloading of multimedia infotainment content from the fixed network infrastructure. Mobile nodes can form Mobile Ad hoc Networks (MANETs) on demand and communicate among themselves without infrastructure support. On the other hand, they can also exchange information with the surrounding network infrastructure while moving. However, the high node mobility causes frequent topology changes among mobile nodes and brief contacts between mobile nodes and roadside infrastructure. This poses great challenges to communication within the MANET and the data exchange between mobile nodes and roadside infrastructure in highly mobile networks such as vehicular networks. The author first focuses on mitigating the impact of high node mobility on the routing in highly dynamic mobile ad hoc networks. Existing literatures propose to use mobility based clustering schemes to form stable communication structure among mobile nodes by exploiting group mobility. To understand the degree of group mobility in common mobility models, particularly Manhattan mobility model which represents urban vehicle movements, the author first proposes a group mobility metric based on the underlying network topology to measure the degree of group mobility for various mobility models. The author observes that in some mobility models, including Manhattan model, no significant group mobility is demonstrated. This suggests mobility based clustering can be ineffective in these models. To find an alternative solution for scenarios in which significant group mobility is absent, the author continues to investigate the effectiveness of deploying mobile relays with stable uplinks. The simulation results show that the mobile relays can significantly improve network connectivity. The author compare the effectiveness of deploying mobile relays against adding more user nodes. The simulation results from both Manhattan and Random Waypoint mobility models suggest that mobile relays are much more effective than adding more user nodes. It is also observed that the path duration is usually brief in highly mobile networks even with mobile relays. This suggests that the clustering and routing algorithms should incur minimum delay to maximally utilize the established paths. Based on the above observation, the author proposes a new cluster based routing protocol FASTR which utilizes mobile relays as backbones to mitigate the impact of node mobility for dense networks with high node mobility and low group mobility. The proposed scheme eliminates the delay caused by cluster head election and enables nodes to start communication immediately after joining a cluster. Through simulation and analysis, the protocol is shown to possess good scalability, incur lower control overhead and achieve higher packet delivery ratio than existing routing protocol. The control overhead of FASTR is shown to be independent of node mobility and consumes less network resources. On the other hand, mobile Internet access through roadside Access Points (APs) has emerged as an alternative to cellular networks due to its high bandwidth and low cost. However, due to the limited range and sparse deployment of roadside APs, several issues could arise when vehicles roam between APs. Besides the brief contact between mobile nodes and roadside infrastructure caused by high node mobility, other issues include intermittent network connectivity and changes in IP address. To address these issues, the author proposes a mobile network access protocol (mNAP) for highly mobile nodes. mNAP introduces a Terminal Local Proxy (TLP) to shield application connections from connectivity disruptions and change of IP address such that the application connections can be preserved across network disruption when roaming between different APs. By enabling cooperative relaying, mNAP exploits opportunistic contacts for additional data transfer. Through simulation, mNAP is shown to be able to exploit both direct and indirect contact opportunities and delivers more than existing scheme. In addition, when vehicles roam across several APs with overlapping coverage, it is important to provide fast and transparent mobility management for the vehicles. To this end, the author proposes a network based local mobility management scheme COAP for 802.11 wireless mesh networks. COAP makes use of 802.11s to forward traffic to the correct location after mobile nodes have roamed to a different AP. Furthermore, COAP proposes a cooperative DHCP service to ensure mobile node always obtain the same IP address within the mesh network. It has several advantages compared to existing schemes. COAP does not require modification on the mobile node and the mobility management is entirely handled by the network. In addition, unlike existing Proxy Mobile IPv6 (PMIPv6) protocol, COAP allows network based mobility across different operator domains. In summary, COAP facilitates easy deployment and helps to enable transparent roaming between roadside APs for mobile nodes.Doctor of Philosophy (SCE