coSense

• DOI: 10.1145/3287074
• Publication Date: 2018-12-27T19:28:03Z
• Authors: Xiaoyang Xie; Yu Yang; Zhihan Fang; Guang Wang; Fan Zhang; Fan Zhang; Yunhuai Liu; Desheng Zhang
• Keywords: Taxis; Penetration rate

The real-time vehicle sensing at urban scale is essential to various services. To date, most existing approaches rely on static infrastructures (e.g., traffic cameras) or mobile services smartphone apps). However, these are often inadequate for the individual level because of their natures low penetration rates. In this paper, we design a system called coSense utilize commercial vehicular fleets taxis, buses, and trucks) scale, given (i) availability well-equipped other vehicles by onboard cameras peer-to-peer communication, (ii) an increasing trend connected autonomous with periodical status broadcasts safety applications. Compared solutions based smartphones, key features in its high rates transparent participating drivers. technical challenge addressed how recover spatiotemporal gaps considering mobility patterns deep learning. We evaluate preliminary road test large-scale trace-driven evaluation Chinese city Shenzhen, including 14 thousand 13 trucks, 10 regular vehicles. compare infrastructure cellphone-based approaches, results show that increase accuracy 10.1% 16.6% average.