Hybrid networks: cellular-relay architecture

Summary form only given. Inter-vehicle communication, an important component of intelligent transportation system, performs crucial functions in collision avoidance, road-hazard notification, and coordinated driving system. An inter-vehicle communication network is an instantiation of a mobile ad hoc network, which is characterized by multi-hop wireless communications among mobile nodes without any centralized control infrastructure. However, automotive ad hoc networks behaves in fundamentally different ways than the models that predominate mobile ad hoc network research. Driver behavior, constraints on mobility and high speeds create unique characteristics in such networks. They are characterized by rapid but somewhat predictable topology changes, with frequent fragmentation, a small effective network diameter, and redundancy that is limited temporally and functionally. The motivation of this research is to propose an effective approach to topology management under these constraints. Apparently flat structure is not suited to vehicular ad hoc networks since it is impossible for every node to acquire the global topology information. All participating nodes must be organized hierarchically so that only local information is required to exchange and global information is hidden for single node. Clustering is an important technique to implement hierarchical structure. Several heuristic clustering techniques have been proposed to choose cluster heads in an ad hoc network. These are lowest-ID, highest-degree and node-weight heuristics. However they cannot be deployed directly in ad hoc vehicular networks since their design objectives are not for highly mobility vehicular networks. The metrics used in these proposals are frequently changed. In this paper, we propose a new clustering technique, position-based hierarchical clustering, which incorporates position information into a novel hierarchical clustering technique. Each node knows its own position through a Global Positioning System. The cluster structure is determined by the geographic position of nodes. Each cluster has one node as the cluster head. Not only is the election of cluster heads based on nodes' position, but also the association and dissociation of clusters are determined by each node's position. The predefined maximum distance between the cluster head and its members then controls the cluster size. It enables nodes to move during cluster setup and maintenance. Furthermore, it allows asynchronous operation for cluster head election. A vehicular ad hoc network can be considered as a one-dimension network by taking the number of lanes into account. We give some mathematical analysis about the performance under some assumptions. Our simulation results show that it gains better stability of the cluster structure, and needs smaller communication overhead for maintaining the cluster structure than the existing approaches do.