Research On Reliable Participant Selection Problem In Mobile Crowdsensing

With the advances of sensing,communication,and mobile computing,mobile crowdsourcing has become a hot technology nowadays.As a new paradigm,mobile crowdsensing,attracts a crowd of active participants to execute spatial-temporal sensing task surrounding them.Utilizing some incentive mechanisms(e.g.monetary reward,virtual credit,entertainment),the process of crowdsensing could take place of deploying traditional static sensors and save a lot of manpower,financial and material resources.There are also exist many problems unaddressed in crowdsensing systems.Participant selection problem is one of the major challenges,which is a process of selecting appropriate participants to tackle sensing tasks about the target areas during the period.Most works focus on how to select reliable participants for task allocation to assure task quality and minimize incentive cost.Conventional methods are mainly based on historical reputation,which is a statistical result of participant behaviors during a past period.However,these methods could cause unreliable assignment that reduces the quality of sensing task,since historical reputation cannot exactly reflect the current state of participants.In this paper,we propose a Reliable Execution Crowdsensing framework,named RECrowd.In RECrowd,a two-stage assignment strategy associated with human-in-the-loop model is designed to achieve reliable participant selection.We introduce a term,truthful willingness to represent participant exact willingness of contributing sensing data to the present task.In actual scenarios,the truthful willingness of participants could be a support for estimating current state.Hence,RECrowd could achieves efficient task assignment via participants' truthful willingness.Besides,we formulate an optimization problem with the objective of minimizing incentive cost while ensuring task quality and design a two-stage online greedy algorithm with pre-assignment step.The two-stage assignment consists of pre-assignment and assignment process.Based on reputation getting from willingness and feedback at pre-assignment,the framework chooses participants by Top-K method,while the framework leverages participants' execution willingness to truly allocate tasks at assignment.In this process,RECrowd is public to participants,which could inspire participants to respond truthful willingness.We take responding truthful willingness as dominate strategy inthis paper.Thus,we exert the proactive impact of participants in platform to increase reliability of assignment,but do not add additional workload when they have known the sensing task in advance.During the participant selection,we also consider participants position privacy.In order to break the spatial-temporal correlation,we adopt cloaking location and time delay strategies to preserve participants' position privacy,in which the truthful willingness could eliminate the uncertainty effect of cloaking location.However,it may introduce a new challenge.The latency of sensing results in temporal domain and the uncertainty of participants' location in spatial domain could bring the potential security issues including authentication and data integrity.To avoid this,we take encryption measures to insure authentication and data integrity.Finally,the experimental results show that compared with other participant selection methods,our algorithm has advantages in reliability,which can meet the requirements of task quality and minimize incentive cost in mobile crowdsensing.