TVD-RA: A Truthful Data Value Discovery-Based Reverse Auction Incentive System for Mobile Crowdsensing

Han Wang, Anfeng Liu, Neal N. Xiong, Shaobo Zhang, Tian Wang

Abstract: Emerging crowdsensing paradigm enables a large number of sensing applications, where much attention is drawn to the fundamental problems for maximizing the system utility and improving the data quality. However, the existing works only consider one of the above issues. In fact, the truthful value of data lies in its veracity. The false data submitted by malicious workers brings no benefits and even causes losses to the system. Therefore, the utility calculations without considering the veracity of data in the previous incentive mechanisms are inaccurate. While current truth discovery (TD) methods cost a lot and cannot improve the quality of data at source. To overcome these problems, we design a truthful value discovery and reverse auction (TVD-RA) framework. First, a TD and trust inspection approach are proposed to find the truth and get the trust degree of workers. This combination can capture the truthful value of data and identify malicious workers. In addition, an incentive mechanism based on reverse auction is proposed. Discovered malicious workers are excluded when selecting task performers, thereby improving the quality of data. In addition, tasks are redundantly assigned in the initial period, creating opportunities for trust inspections. Although this entails losses, the influence of malicious workers can be ruled out. Therefore, we gain long-term utility at the expense of current benefits. Theoretical analysis demonstrates that the proposed incentive mechanism is incentive compatible and individual rational. We finally carry out extensive evaluations, where results demonstrate the superiority of our mechanism over the state-of-the-art approaches.