Metropolitan-scale taxicab mobility modeling

Taxicabs, as one of the major transportation platforms in metropolises, are of great interest in vehicular communications and networking research. The mobility of taxicabs, determined by driver behaviors and passenger destinations, has attracted a lot of attention in recent years. Among different mobility models, trace-driven taxicab mobility models preserve many details but often bring too much overhead, while simple, random mobility models largely miss the needed social and geographical features. In this paper, we follow a new approach to capture the social behaviors (e.g., transition among regions) and geographical features (e.g., hot spots) of taxicabs in a metropolis, and build a hierarchical taxicab mobility model to strike a better balance between fidelity and tractability. The performance study shows that the synthesized mobility model can well capture the original trace data while being simple and extensible. It also reveals the difference between taxicab and pedestrian mobility, the latter of which is also often used for vehicles in the literature.