Toward Incentive-Compatible Vehicular Crowdsensing: An Edge-Assisted Hierarchical Framework
Xinxin Yang, Bo Gu, Bing-Kun Zheng, Beichen Ding, Yu Han, Keping Yu
Abstract
Vehicular crowdsensing (VCS) is a promising paradigm that utilizes the mobility of vehicles to collect city-scale environmental data for monitoring and management purpose. However, the dramatic growth of traffic load, lack of incentive mechanism, and sensing cost heterogeneity bring considerable challenges in achieving a successful VCS system in the real world. To address these issues, we present an edge-assisted hierarchical VCS framework to achieve efficient vehicle recruitment and data collection. In particular, a Stackelberg game is for-mulated to analyze the interactions between edge servers and vehicles. Then, a deep reinforcement learning-based incentive mechanism is detailed for motivating vehicles to participate in sensing activities and contribute high-quality data. Intensive simulations are conducted to verify the efficiency of the proposed mechanism. Finally, we present several open issues and directions for future research.