On auction design for crowd sensing

The recent paradigm of mobile crowd sensing which collects sensed data from pervasive mobile devices enables a broad range of large scale sensing tasks. In this paper, we aim to study one critical challenge in this paradigm, namely, design of compensation or incentive mechanisms for sensors to expend sufficient resources to take high quality measurements and transmit them to the central fusion unit. Inspired by the widespread use of the generalized second-price auction in search engine advertising business, we analyze the potential and performance of a reverse generalized second-price auction being used for this problem. Our main result is that reverse generalized second-price auction has a special subset of Nash equilibria that can achieve a desired level of error covariance at the central fusion unit with less payment than other common mechanisms such as reverse Vickrey-Clark-Groves or reverse generalized first-price auction.