Research On Privacy Preservation For Task Cycle Of Mobile Crowdsensing

Due to the technical development of wireless communication and sensor and the popu-larization of mobile smart devices,mobile crowdsensing(MCS)centered on participants ener-gizes wireless sensor network(WSN)centered on sensor devices.Sensing tasks in MCS are outsourced to a crowd of normal users owning smart devices by the way of crowdsourcing,and sensing data is collected and analyzed through the intelligence and advantages of crowds.The task cycle of MCS consists of task creation and assignment,task execution,and sensing data integration.Compared with the traditional WSN,MCS has advantages in organization mode,sensing contents,application scenarios,scalability,and development cost.MCS is considered a critical component of the Internet of Things(Io T)and has been applied in many fields of a smart city,such as indoor navigation,traffic analysis,environmental monitoring,services rec-ommendation,both behaviour and healthcare identification of crowds.Although MCS has excellent application prospects,there are still many technical challenges for its development.For example,how to set appropriate rewards to motivate the participation of more normal users,how to protect the privacy of task requester and task participant simul-taneously,and how to evaluate the reliability of heterogeneous sensing data.A large number of schemes focus on incentive design,task assignment,and data reliability evaluation of MCS,but the following deficiencies exist in previous solutions.(1)Existing privacy-preserving task assignment schemes fail to protect the privacy of both task requester and task participants while obtaining the minimum travel distance of task participants.(2)Most previous incentive mecha-nisms cannot resolve contradictions among privacy preservation,data reliability evaluation,and reward distribution and establish a relationship between data reliability evaluation and reward distribution.(3)For continuous sensing tasks,traditional incentive mechanism usually lacks strategies for real-time reward distribution.(4)Considering the heterogeneous sensing data col-lected by various sensors,previous data reliability evaluation schemes generally do not support privacy-preserving data evaluation for heterogeneous sensing data.To address the above prob-lems,this dissertation conducts fourfold research on three phases of the sensing task cycle.In the phases of task assignment,to protect the privacy both task requester and task par-ticipants and calculate the minimum travel distance of task participants,a bilateral privacy-preserving task assignment scheme for MCS is proposed in this dissertation,named i TAM.The privacy of both task requester and task participants is protected via the Paillier encryption.To support task assignment and distance calculation on ciphertext,a privacy-preserving comparison protocol(PCP)and a privacy-preserving minimum protocol(PMIN)are designed.For compet-itive and cooperation tasks,single participant and multiple participants selection problems are proposed.Whether k(k>1)task participants or one task participant is required by a sensing task,k(k?1)task participants who are nearest task location can be located in the proposed i TAM while protecting the privacy of both task requester and task participants.The results of experimental evaluations on a real-world and synthetic dataset show that the proposed i TAM is effective and efficient and certainly obtains the minimum travel distance of task participants.In the phases of task execution,to enable an incentive mechanism to protect the privacy and evaluate data reliability,simultaneously,a privacy-preserving and quality-aware incentive mechanism for MCS is proposed,called PACE.To resolve the contradiction between privacy protection and data reliability evaluation,zero-knowledge model of data reliability estimation and its construction are suggested.Moreover,a quantization function of data quality,which takes sensing data and a data centroid calculated by reliable data as inputs,is designed.Besides,a prepaid reward distribution mechanism is proposed to prevent dishonest task requester from refusing to pay rewards for task participants.The results of experimental evaluations on a real-world dataset show that the proposed PACE is feasible and efficient and provides more rewards for task participants providing higher quality.In the phases of task execution,although the proposed PACE achieves privacy-preserving and quality-aware reward distribution,it requires to preset the reliable requirement and sev-eral rounds of communication between a task participant and the sensing platform to estimate data reliability.To reduce the cost of both computation and communication of task partici-pants and realize real-time reward distribution,a privacy and reliability-aware real-time incen-tive mechanism for continuous crowdsensing is proposed,named PRICE.Specifically,a novel privacy-preserving truth discovery protocol(Pri TD)is designed to achieve privacy-preserving data reliable estimation.Different from existing privacy-preserving truth discovery solutions,the proposed Prin TD only needs one server and not requires several rounds of communication between a task participant and the sensing platform.Furthermore,a two-layer truth discovery mechanism is first proposed to solve the unfairness reward distribution brought by the single time slot failure.The results of experimental evaluations on real-world and synthetic datasets show that the proposed PRICE supports real-time reward distribution and solve the unfairness reward distribution.In the phases of data integration,to support heterogeneous data reliability evaluation and protect data privacy,a privacy-preserving reliability estimation scheme for heterogeneous data in MCS is proposed,named Iron M.In Iron M,heterogeneous data reliability estimation is trans-formed into the constraint test of range and equality.A privacy-preserving range matching scheme P~2RM is designed to achieve privacy-preserving reliability estimation of heterogeneous data,where the proposed P~2RM can evaluate the numeral data and textual data,simultaneously.Moreover,the one-wayness and distinction of the perceptual hashing function are adopted to achieve privacy-preserving reliability evaluation for multimedia data.The proposed Iron M does not require several rounds of communication between a task participant and the sensing plat-form,and multiple servers to estimate the reliability of data,and it can protect the privacy of both task requester and task participants and resist the guessing attack of the sensing platform.The results of experimental evaluations on real-world datasets show that the proposed Iron M supports privacy-preserving and real-time reliability estimation for heterogeneous data.