Traffic sensing is crucial to a numas traffic management and city road networkbuild a traffic sensing system with probe vehiitan scale traffic sensing. Each probe vehicle speed and position periodically and sensoryhicles can be aggregated for traffic sensing. Hcritical issue that the sensory data contain spcancies with no reports. This is a result of thedistribution of probe vehicles in both spatidimensions since they move at their own willsposes a new approach based on compressivescale traffic sensing in urban areas. We mine trace datasets of taxies in an urban environmecomponent analysis and reveal the existencetures with sensory traffic data that underpinsensing approach. By exploiting the hidden stcient algorithm is proposed for finding the becondition matrix by minimizing the rank of trix. With extensive trace-driven experimentsthat the proposed algorithm outperforms a nutive algorithms. Surprisingly, we show that oachieve an estimation error of as low as 20%than 80% of sensory data are not present.

PDF