A Fair Reputation-Based Incentive Mechanism for Cooperative Crowd Sensing

Abstract

Crowd Sensing (CS) is a paradigm empowered by the pervasiveness of mobile smart devices, in which crowds of device owners cooperate to provide information about their surrounding environment. In this paper, we introduce the Data and Participant Assessment and Remuneration Scheme (DPARS) for cooperative CS applications. DPARS implements a three-stage procedure to estimate a fair reputation-based payoff for CS participants. We achieve this by first applying a consensus- based outlier detection technique on the received data. The output of this technique is used to statistically evaluate participants' reputations based on the Dirichlet process. Consequently, a fair payoff for every participant is determined by treating participants as coalitions of players in a cooperative game. Performance results indicate that our proposed scheme efficiently detects misbehaving participants, and decreases the amount of incentives allocated to them.