Distributed K-Distance Indexing Approach for Efficient Shortest Path Discovery on Large Graphs

The emergence of large real life networks such as social networks, web page links, and traffic networks exhibits complex graph structures with millions of vertices and edges. Among many operations for exploiting these graphs, the shortest path discovery is a major and expensive one. Besides the in-memory approaches, many efficient shortest path computation methods have been developed on top of distributed and parallel platforms. Pregel, a bulk synchronous parallel framework, is one of them for processing large graphs. The known shortest path computation approach with Pregel is computation intensive and unable to target real-time services. In this paper, we propose a Pregel based efficient k-distance index technique that allows efficient single pair shortest path discovery. We reduce the network cost and unnecessary operations by transmitting more information in a single superstep. The extensive experiments on both real and synthetic datasets reveal the superiority of the proposed approach.