A trajectory-based recruitment strategy of social sensors for participatory sensing

Abstract

Participatory sensing, a promising sensing paradigm, enables people to collect and share sensor data on phenomena of interest using mobile devices across many applications, such as smart transportation and air quality monitoring. This article presents a framework of participatory sensing and then focuses on a key technical challenge: developing a trajectory-based recruitment strategy of social sensors in order to enable service providers to identify well suited participants for data sensing based on temporal availability, trust, and energy. To devise a basic recruitment strategy, the Dynamic Tensor Analysis algorithm is initially adopted to learn the time-series tensor of trajectory so that the users' trajectory can be predicted. To guarantee reliable sensing data collection and communication, the trust and energy factors are taken into account jointly in our multi-objective recruitment strategy. In particular, friend-like social sensors are also defined to deal with an emergency during participatory sensing. An illustrative example and experiment are conducted on a university campus to evaluate and demonstrate the feasibility and extensibility of the proposed recruitment strategy.