Research On Efficient Data Collection Scheme In Vehicular Crowdsensing System

With the rapid increase of intelligent connected vehicles and the continuous development of intelligent transportation system,vehicular crowdsensing has become a novel way to conduct large-scale sensing task.In recent years,mobile edge computing architecture deploys storage computation resources to the edge end nearby the users,which significantly reduces the latency of data transmission and computation.With the aid of mobile edge computing,vehicular crowdsensing system is able to collect real-time data more effectively.However,with the increase of sensed data,the data redundancy and the communication cost of vehicular crowdsensing system sharply raises and the data processing and storage cost of edge server also increases,which affects the efficiency of crowdsensing.In addition,in vehicle recruitment,it is difficult for the recruited vehicles to cover all the sensing tasks,which affects the reliability of the vehicular crowdsensing system.And recruiting a large number of connected vehicles further increases the overhead of crowdsensing system.Therefore,it is an urgent problem to be solved that how to design a reasonable data collection scheme to improve the efficiency and effect of crowdsensing.To this end,we research and design an efficient data collection scheme for edge-assisted vehicular crowdsensing and the major achievements include:In view of the problem of data redundancy in data collection,we propose a low redundancy oriented grid selection and parameter adjustment scheme to reduce communication overhead,control data redundancy and improve data quality.The scheme adopts a sensed data preprocessing mechanism,which takes advantage of the computation ability of vehicles to preprocess sensed data.Then we formulate the grid selection problem as an information entropy based optimization problem.According to the submodular property of the objective function,we design an greedy based grid selection algorithm in order to solve the optimization problem.Finally,we propose an online collecting parameter adjustment algorithm,which adjusts collecting parameters reasonably.Extensive simulation results show that the scheme significantly reduces communication overhead and controls data redundancy,which improves the efficiency of data collection in vehicular crowdsensing.In view of the problem of coverage in data collection,we design a high coverage oriented vehicle recruitment scheme and incentive mechanism to improve the coverage rate of sensing tasks and enhance the initiative of vehicles to participate sensing.The scheme firstly constructs a vehicle reputation model to comprehensively evaluate the reputation of vehicles from many aspects.Next,we formulate the vehicle recruitment problem as a double objective optimization problem,and adopt a genetic based recruitment algorithm to solve the optimization problem.Besides,we propose a reputation based vehicle incentive mechanism,which pays additional rewards to the recruited vehicles according to the vehicle reputation.Extensive simulation results show that the scheme can balance the sensing coverage rate and recruitment cost while enhancing the vehicle initiative,which improves the effect of data collection in the vehicular crowdsensing.