On using broadcast index for efficient execution of shortest path continuous queries

Abstract

Various efficient indexing techniques have been proposed for formulating broadcast index to minimize the tune-in cost at mobile clients to support shortest path search in a road network. However, none of them is designed for shortest path continuous queries (SPCQ). Due to frequent updates of traffic data, an SPCQ has to be invoked periodically while the client is following the suggested shortest path to its destination. This could result in heavy path searching cost at the client. To reduce the tune-in and path searching costs at the clients, in this paper, we propose the progressive approach (PA) in which we apply the next region (NR) approach together with a hierarchical network approach to formulate the local networks at the clients for shortest path navigation. Since the shortest path may change due to changing traffic, to minimize the path searching cost, we aim for an approximate path including shortcuts of shortcut networks (SN) from each invocation of an SPCQ to navigate the client to its destination step-by-step. In formulating the client local network, higher level shortcut networks are chosen if the regions are farther away from the current region of the client. An important property of the approximation mechanism adopted in PA is that traffic updates may be pruned if they will not significantly affect the shortest path information to be distributed to the clients. This can reduce the index re-generation cost at the traffic server. Extensive performance evaluation experiments have been conducted to investigate how PA reduces the tune-in cost from data broadcast and path searching cost at the clients with just small delays in the arrival times to their destinations.