Privacy-Aware Participant Recruitment in Opportunistic Device to Device Networks

In most of the existing mobile applications for data collection and data analytics, either the privacy issue is frequently neglected or the privacy options are not configurable by the participants. This paper proposes configurable privacy level by potential crowdsourcing participants who are able to choose their desirable privacy level and get paid based on the quality of data they provide to the originator in Device to Device network (D2D). Combining with the encryption technique, not only the user's privacy is protected, but also the encryption complexity is reduced. We first formulate the participant recruitment process into an optimization problem from the participants' perspective by considering the competition and collaboration among existing and candidate participants in order to achieve the best utility. Then we design a distributed approximate scheme that relies on participants' local knowledge to complete the overall recruitment task. We implement the approximate approach in Dell Streak tablets and carry out a campus-scale experiment for 21 days, plus run simulations for more extensive and detailed evaluation under various task settings. The results demonstrate the efficiency of the proposed approaches and disclose valuable insights for practical considerations in D2D based crowdsourcing.