Approximate shortest path algorithms for network-tree model

The problem addressed in this paper is how to efficiently compute the approximate shortest paths of large-scale network. We propose the wholly new network-tree model (NTM) constructed from arbitrary network and associated approximate algorithms for high performance computation of shortest path in large-scale network. For the tradeoffs between the computational speed and errors, three approximate network-tree shortest path (ANTSP) algorithms are presented: nearest ANTSP, heuristic ANTSP and sub-optimal ANTSP algorithms. The experiment results based on grid network show that all ANTSP algorithms achieve much higher computational efficiency than Dijkstra algorithm, and the average errors of heuristic ANTSP and sub-optimal ANTSP algorithms are significantly reduced.