| With popularization of mobile terminal devices,Mobile Crowdsensing(MCS),as an emerging data collection paradigm,has received extensive attention from academia and industry.It uses smart device carried by mobile user as perception carrier,forms an interactive perception network through user’s movement.It uses smart device perception module to perform data perception,analysis,and sharing operations,and collaborate to complete complex tasks that are difficult for individuals to solve.Compared with wireless sensor network,MCS network has the advantages of low perception cost,strong flexibility and wide coverage.At the same time,it has a wider range of application scenarios in the fields of environmental monitoring,medical care,and intelligent transportation.Among them,task allocation mechanism has always been a key issue in MCS network research.In actual perception scenarios,task requirements are complex and changeable,and the applicability of existing task allocation strategies is not high.This paper studies the task allocation problem in two application scenarios,aiming to obtain better sensing quality and reduce the cost of perception.The main work and innovations of this paper are as follows:(1)Aiming at the task assignment problem in opportunistic perception scenario in MCS network,this paper takes minimizing the average completion time of all tasks as objective function,and proposes a multi-task assignment algorithm based on the reputation update mechanism.By introducing data quality evaluation methods and participant reputation evaluation methods,a reputation update mechanism is constructed,that is,the timeliness,completeness and accuracy of tasks are used to measure data quality.The historical reputation is determined by combining the willingness of participants on this basis,and use logistic regression function to update the credibility of current participants.In the task allocation process,participants with high reputation are selected first,and tasks are assigned to participants with the smallest expected perception time in turn.At the same time,an incentive mechanism is introduced to allow participants with higher reputation to get higher rewards when performing tasks.Simulation analysis is carried out in offline and online scenarios,which proves that compared with other algorithms,the algorithm in this paper can reduce the average completion time of all tasks by more than 10% under the premise of ensuring quality of task data.(2)In view of the task assignment problem in participatory perception scenario in MCS network,the task usually belongs to the form of spatial distribution,and the heterogeneity of the task,the execution ability of the participants,and the data quality of task need to be considered.Generally,higher data quality corresponds to higher perceived cost,so how to weigh data quality and perceived cost is a difficult point in task assignment problem related to location.In order to deal with these problems,this paper proposes a location-based multi-task assignment algorithm.The algorithm introduces the participant’s perceived credibility evaluation method and incentive mechanism.The former can effectively evaluate the participant’s perceived credibility to complete task and meet the data quality requirements of task performed by the participant,and the latter can motivate participants to complete the perception task.On this basis,the task allocation problem is transformed into an optimization problem.Under the premise of ensuring quality of task data,two task allocation methods are designed with the goal of minimizing the perceived cost.One is the Reward First Algorithm based on the greedy algorithm,and the other is the hybrid task assignment algorithm(Genetic and Greedy Algorithm for Task Assignment,GGA-TA)that combines greedy algorithm and genetic algorithm.Simulation experiments show that the GGA-TA algorithm can improve the performance of the comparison algorithm by more than 20% under the premise of ensuring quality of task data. |
PDF Full Text Request