Reputation-based multi-auditing algorithmic mechanism for reliable mobile crowdsensing

Abstract

Mobile crowdsensing has become an efficient paradigm in which crowd workers are recruited to collect data by using their mobile smart phones. However, different workers may provide data with varied degrees of quality. Therefore, it is imperative to develop a reliable crowdsensing system that guarantees the quality of service (QoS) for each task. In this paper, we propose a Reputation-based Multi-Auditing algorithmic mechanism (RMA) by integrating Task-based Temporal Reputation mechanism (TTR) and Reputation-based PM truth inference algorithm (RPM). Further, Performance-Based Payments scheme (PBP) is adopted to promote truthful workers. Based on the past benefits, the behavior of a rational requester may vary over time. Particularly, reinforcement learning and (1-ϵ) accuracy algorithm are used to model the update policy of a requester’s strategy. Both rational and irrational workers are considered in this paper. Depending on whether a worker can perceive the benefits of other workers, K-armed bandits and neighborhood learning mechanism are respectively adopted to model the update policy of rational workers. By using Lyapunov stability theory, it is qualitatively proved that the trustful provision of sensed data provides an unique stable evolutionary equilibrium for each rational worker in our proposed system. Finally, extensive simulations and real data experiments illustrate that the RMA mechanism has an outstanding performance on discovering truth and achieving profits.