Passport: parallel land route planning software

In geographical information systems and spatial-based decision-support systems several computational challenges stem from the need to handle large quantities of spatial data, and the need to analyze, interact and visualize complex data sets. We concentrate on algorithms for land route planning, which is used in applications such as mobility analysis and rescue route determination. The land route planning process, which makes use of terrain topography and its attribute data (i.e. road networks, streams, forests, type of soil and vegetation cover), consists of determining optimal land routes that a moving object can follow to reach a destination. We propose an algorithm that considers a combination of terrain features along with operator-selected heuristics. The algorithm performs speculative parallel route planning using the bidirectional Dijkstra algorithm. A fast heuristic (threshold) is used to expand nodes along the wave-front that have the greatest promise or being along the optimal path. The algorithm described has been implemented on a shared memory symmetric multiprocessor (SMP) computer. Performance results are provided along with an evaluation of our heuristic approach.