GRAIM: Game and Reverse Auction based Incentive Mechanism in Mobile Crowd Sensing

In mobile crowd sensing (MCS), most of the research work does not consider the dropout situation of the workers, resulting in a lower completion rate of the tasks. To decrease the performance loss caused by the dropout of workers, one solution is to prevent workers from dropping out in advance. However, this solution cannot be widely applied to some emergent scenarios and avoid dropouts strictly. In this paper, we consider another solution to respond to worker dropouts actively and innovatively propose a two-stage incentive framework, namely the Game and Reverse Auction based Incentive Mechanism (GRAIM), which aims at motivating dropout workers to submit perceived data and effectively recruiting high-quality idle workers to complete all unfinished tasks, including the tasks left behind by the dropout workers and other incomplete tasks. In the first stage, a bargaining game-based reward allocation method (BGRA) is proposed to incentivize the perceived data submission of dropout workers with a reward equilibrium between the platform and the dropout workers. In the second stage, worker recruitment is modeled as a multi-armed bandits (MAB) issue in a reverse auction, and the extended upper confidence bound (EUCB) algorithm is proposed for the platform to recruit high-quality idle workers to perform unfinished tasks. The experimental results show that compared with the state-of-art mechanisms, our proposed GRAIM increases social welfare by about 15%~ 23%, reduces average reward by about 16% ~ 25%, and increases task completion rates by about 8% ~ 20%.