Developing Map Matching Algorithm for Transportation Data Center

Abstract

Map matching (MM), pins the drifting position data to the correct road link on which a vehicle is travelling, is a crucial step needed by many industrial or research ITS projects which rely on post-hoc analysis of trajectories. To address the unprecedented challenge of massive GPS data processing in urban transportation data center nowadays, this paper proposed an improved parallel topological map-matching algorithm that aims to achieve highest efficiency as well as guaranteed accuracy. The main contributions of this work include: I) a weighting scheme based on cost-effectiveness ratio to reduce candidate path set in low time cost, II) a novel leapfrog method to omit the redundant GPS points that are not needed in path determination, III) parallelized processing using Map Reduce paradigm. Experiment show that these improvements greatly reduced algorithm's running time when compare to the state of the art.