Memory Efficient Index for Route Planning in Public Transportation Networks

Given a source stop and a destination stop in a transportation timetable, the route planning query returns a Pareto set of paths that optimized in both earlier arrival time and less transfer times. Current indexing methods require large runtime memory. This paper proposes a memory space efficient index method (a-)PAINT. (a-)PAINT pre-computes, for each stop v, a set of labels, such that, for any query from stop s to stop d, we can efficiently retrieve a path of the query result from the label sets of s and d. PAINT could return optimal paths for every query. a-PAINT incurs far less preprocessing overheads, but might return sub-optimal path that takes one more transfer. We experimentally conducted simulated queries on real timetables. PAINT incurs reasonable preprocessing overheads. a-PAINT requires far less preprocessing overheads, and it has accurate rate up to 90% in large timetables. © 2021, Chinese Institute of Electronics. All right reserved.