Research And Implementation Of Context-aware In Crowd Sensing Based On Mobile Terminals

With the proliferation of smart phones, Internet of things and mobile Internet isdeveloping by leaps and bounds. Wi-Fi, GPS and other equipment are widely deployed inmajor cities. A large number of digital fingerprints are generated when mobile devices interactto geographic location. Under such a background, we can quickly collect large amounts ofuser-generated data. Context-aware computing is into the crowd sensing time from individualperception. Large amounts of data with heterogeneous, multi-source, multi-mode, qualitydifferences in characteristics brings many new scientific research as well as challenges incrowd sensing.Firstly, this paper analyzes the existing crowd sensing application architecture. Secondly,focus on problems of non-scalability and unable to reuse the data, designing an mobile crowdsensing application architecture which supports for the data collection controlling, data reusingand sharing. Thirdly, to processing the heterogeneous, multi-source, quality different contextinformation, this paper proposes a new multi-layer information fusion method based oncommunity detection algorithm, data layer and decision layer information fusion algorithm.This method is divided into two phases: client analytics and server information fusion. Onclient side, through the improved fuzzy logic algorithm, modifying the membership functionsto compute the membership size, and then computing the probability of all the membershipsize to suit more complex situations. The output result of client is a sub decision computedthrough the value of three axises acceleration. On server side, information filtered bycommunity detection, and information fusion based on data layer batch aggregation arithmeticand decision layer rule theory algorithm. Lastly, proved through the crowd sensing experimentin teaching building that, this method can effectively processes the data in crowd sensing andinfers the specified context.