Research On User-Behavior Aware Data Optimization Strategy For MCS

Currently,mobile smart devices such as smartphones,tablets and smart wearable devices integrate more and more sensors,which have more and more powerful computing,sensing and storage capabilities.So a new sensing way that combines crowdsourcing ideas with mobile sensing to capture sensing data is generated,called Mobile Crowd Sensing(MCS).In MCS,a large number of participants are required to collect sensing data to obtain sensing results.However,the data quality and the data coverage between participants are different because of the social attributes of participants' behaviors,mobility and other factors.Therefore,designing data optimization strategies to ensure the accuracy,timeliness and information comprehensiveness of sensing results becomes an urgent problem to be solved.Firstly,this thesis introduces the research background of MCS,and then the challenges of data quality optimization and data coverage optimization are analyzed and existing data optimization strategies are expounded.Secondly,aiming to the problem of data quality optimization,a Data Quality Optimization strategy based on Reputation is proposed.In DQOR,the comprehensive reputation is determined by assessing the direct and indirect reputation of the participants.The historical interaction behavior and the historical collaborative behavior of the participants are analyzed according to the comprehensive reputation,and the sensing data of the participants with higher comprehensive reputation is selected to ensure the reliability of the data.In addition,the quality of the received data is evaluated and then the current trust degree of participants will be dynamically update because of the influence of participants' attributes,social behaviors and other factors.Consequently,the accuracy of the sensing results is optimized based on the data items with higher quality and the current trust degree of the participants.Lastly,aiming to the problem of data coverage optimization,this thesis proposes a Data Coverage optimization strategy based on Service Quality under the scenario where the sensing task has higher real-time requirement.In DCSQ,the service quality is determined by the willingness and regional preference of participants,so as to analyze the real-time effectiveness of the data.Meanwhile,the data coverage is evaluated according to the number of target points covered by the participant during the execution of the sensing task.Furthermore,the efficiency of the participant is determined by service quality and data coverage when the platform budget is limited,and an iterative greedy algorithm is used to select the sensing data of the most efficient set of participants,so as to maximize the data coverage and ensure the real-time validity of the data.This thesis proposed corresponding models and methods from two aspects,data quality optimization and data coverage optimization.The effectiveness of them have been demonstrated by theoretical analysis and numerical results.The study in thesis provides a theoretical basis for data optimization of MCS,which has a certain reference value.