Opportunistic Data Collection Protocol Based On Social Behavior In Mobile Crowd Sensing

For the past few years, with the popularity of smart phone and the research of wireless sensor network, crowdsourcing computing, social computing and opportunistic network, mobile crowd sensing has been proposed as a new sensing model. The theory and key technology of mobile crowd sensing have an opportunity for further breakthrough. The mobile crowd sensing constitutes a people-centered network, using people’s mobility and wisdom to perceive context data and react to human themselves. The social behavior analysis based data collection has become an important data collection method under this human centric network.This thesis surveyed the origin and development of mobile crowd sensing and summarized three closely related aspects of platforms, social attributes and opportunistic for social behavior analysis based data collection. Then we analyzed the relationship between them, and proposed the critical issues of future work.In existing data collection protocols of mobile crowd sensing service, multiple behavior property has not been considered yet. We extracted multi-behavior place attributes from user mobility trace and proposed multiple behavior places scaling model based on BM25, which enable mapping user social profile to multiple behavior space. Then this thesis created a user-cloud interaction model, which requires users to regularly submit user behavior data to cloud platform in order to calculate the weight vector of the behavior place group based on their behavioral characteristics and real-time data provided by cloud platform. The cloud platform also collect user data and build the user- place united matrix so as to calculate the BM25 related factors specified in the model according to the matrix.With the above basis, this thesis then proposed a multiple behavior place opportunistic data collection protocol, called MBPOP, which provides behavior-cast service in mobile crowd sensing applications. MBPOP firstly initializes the single behavior users, then diffuses the data by single behavior gradient ascend. Finally, the protocol multicasts the data to the recipients who satisfies multiple behavior place profile accurately. Comparing with other similar protocols, MBPOP is superior in energy cost and delivery ratio.