A Stream Data Service Framework for Real-Time Vehicle Companion Discovery

Because of the large-scale and complexity nature of vehicle trajectory data, existing methods have struggled to guarantee the efficiency and effectiveness of vehicle companion discovery in real-time. This paper proposes a stream data service framework to real -time discover vehicle companion. It relaxes the spatial and temporal constraints of vehicle companion definition to find more potential companion vehicles. And to ensure the performance, we make use of flexible service collaboration to process data generated by related monitoring sites selectively, and employ a stream partition strategy to realize service collaboration in parallel. Experiments conducted with real and simulated data demonstrate that our method can identify a greater number of potential vehicle companion sets while requiring less transmission and processing resources.