VTracer: When Online Vehicle Trajectory Compression Meets Mobile Edge Computing

Vehicles can be easily tracked due to the proliferation of vehicle-mounted global positioning system (GPS) devices. ${\sf VTracer}$ is a cost-effective mobile system for online trajectory compression and tracing vehicles, taking the streaming GPS data as inputs. Online trajectory compression, which seeks a concise and (near) spatial-lossless data representation before revealing the next vehicle's GPS position, is gradually becoming a promising way to alleviate burdens such as communication bandwidth, storing, and cloud computing. In general, an accurate online map-matcher is a prerequisite. This two-phase approach is nontrivial because we need to overcome the essential contradiction caused by the resource-constrained GPS devices and the heavy computation tasks. ${\sf VTracer}$ meets the challenge by leveraging the idea of mobile edge computing. More specifically, we offload the heavy computation tasks to the nearby smartphones of drivers (i.e., smartphones play the role of cloudlets), which are almost idle during driving. More importantly, they have relatively more powerful computing capacity. We have implemented VTracer on the Android platform and evaluate it based on a real driving trace dataset generated in the city of Chongqing, China. Experimental results demonstrate that VTracer achieves the excellent performance in terms of matching accuracy, compression ratio, and it also costs the acceptable memory, energy, and app size.