Fast computation for point-to-point shortest path based on four closest nodes in higher level road network

Abstract

The point-to-point shortest path computation is one of the hot research topics today. One straight forward application is to find the optimal driving directions. To solve the difficulties in shortest path computation for large scale graph, an efficient approximation algorithm was proposed based on road network hierarchies. Four closest nodes in higher level road network to starting node and four closest nodes to ending node were computed first along with 8 corresponding shortest paths. For subgraph T which consists of only higher level roads, 8 edges corresponding to the previously computed 8 shortest paths were then added to T and results in a graph T′. In graph T′, search for the shortest path from starting node to ending node, which completed the task. This design demonstrates that the proposed algorithm is suitable to solve large scale problems. An error bound is provided for approximation shortest path. It is also possible to preprocess the data first. In real application, the computational results are quite competitive, which shows that the proposed algorithm is effective.