V2PSense: Enabling Cellular-based V2P Collision Warning Service Through Mobile Sensing

Abstract

The C-V2X (Cellular Vehicle-to-Everything) technology is developing in full swing. One of its mainstream services can be the Vehicle-to-Pedestrian (V2P) service. It can protect pedestrians who are mostly vulnerable on the road. In this work, we seek to enable a V2P service that can identify which pedestrians may be nearby a dangerous driving event and then notify them of warning messages. To enable this V2P service, there are two major challenges. First, a low-latency V2P message transport is required for this infrastructure-based service. Second, the pedestrian's smartphone requires an energy-efficient outdoor positioning method instead of power-hungry GPS due to its limited battery life. We thus propose a novel solution, V2PSense, which trades off positioning precision for energy savings while achieving low-latency message transport with LTE high-priority bearers. It does a coarse-grained positioning by leveraging intermittent GPS information and mobile sensing data, which includes step count from the pedometer and cellular signal strength changes. Though the V2PSense's positioning is not as precise as the GPS, it can still ensure that all the pedestrians nearby dangerous spots can be notified. Our results show that it can achieve the average precision ratio 92.6% for estimating where the pedestrian is while saving 20.8% energy, compared with the GPS always-on case.