Researches On Data Lifecycle Security And Privacy Preservation Of Mobile Crowdsensing

The development of wireless communication technology and smart terminal equipment,promotes participatory sensing feasible in large-scale mobile crowds.The sensing tasks in mobile crowdsensing(MCS)are outsourced to ordinary users with mobile smart devices(such as smart phones,tablets,smart wearable devices,on-board units).MCS uses crowd intelligence and mobile devices as basic sensing units to complete sensing tasks through the Internet of things and wireless sensor networks.The collected and aggregated sensing data is used to extract analysis and provide personalized services that enable large-scale social sensing,which has creative significance in the expansion of smart city.However,data security and privacy issues restrict the development of MCS,and hinder the optimal utilization of resources.Existing research focuses on the privacy preservation with data reliability evaluation,incentive mechanism and task allocation,the following problems need to be discussed:(1)the existing privacy-preserving deduplication schemes are designed for digital data and single type data,without heterogeneous data,meanwhile,the frequent interactions and high latency cannot be directly applied to MCS;(2)most existing privacy-preserving task allocation schemes less consider task attributes,user preferences,and reputation level privacy leakage in the process of task matching;(3)the existing privacy-preserving data searching schemes fail to simultaneously achieve image data quality evaluation and support multi-requester/multi-user query;(4)for privacy leakage in MCS application scenarios,existing schemes can hardly meet the requirements of efficient data aggregation and verification.Therefore,we explore the theory and key technologies of data security and privacy preservation for each stage of the data lifecycle in MCS,and conduct relevant research in four aspects:secure heterogeneous data deduplication,privacy-preserving multi-task allocation and data aggregation,privacy-preserving image filtering and searching,and MCS application of passenger flow privacy query.In the data collection phase,a privacy-preserving secure heterogeneous data deduplication scheme(SHD)is proposed to address the large amount of data redundancy generated by multiple participants,which affects the quality evaluation,and occupies large amount of storage space and communication bandwidth.The decryption capability of the sensing task is delegated to a specific fog node,and a proxy re-encryption method is used to guarantee the confidentiality of the task content.Based on lightweight two-party random mask and polynomial aggregation techniques,a privacy-preserving cosine similarity calculation protocol is constructed to achieve heterogeneous data deduplication with fog-based and protect the reports,while effectively improving data quality and reducing traffic load.The security and privacy are demonstrated under the defined threat model.The performance evaluation shows the computational overhead and communication overhead,and the experimental results demonstrate the efficiency of SHD scheme on real data sets.In the data aggregation phase,a privacy-preserving multi-task allocation and data aggregation scheme(PMTA)is proposed to address the privacy leakage of task attributes and user preferences,which affects the participation of users while failing to achieve secure data aggregation.A secure and efficient grouping mechanism is designed using K-means clustering and matrix multiplication to divide users into different task groups based on the similarity of user preferences and task attributes,which obtains a high-quality and accurate target set with privacy preservation.Based on the short group signature algorithm and 0-1 encoding technology,a privacy-preserving matching mechanism is constructed to guarantee anonymous authentication and achieve the matching of task requirements and user reputation levels in a privacy-preserving manner.The PMTA scheme ensures secure and accurate multi-task allocation and anonymous authentication,and improves user participation and task completion rates.A formal security analysis is conducted under a threat model.The performance evaluation and experimental results demonstrate the feasibility and efficiency of PMTA scheme.In the data service phase,a privacy-preserving image filtering and searching scheme(PIFS)is proposed for the leakage of data content privacy and user query privacy with the multirequester/multi-user data query,while the low-quality image data filtering and image content privacy-preserving.Considering factors such as light intensity,motion blur state,shooting angle and recognition accuracy that affect image quality,a comprehensive scoring algorithm is defined to measure image quality and filter out high-quality image data.An efficient non-interactive,encryption matching method is proposed to achieve image query under a multi-requester/multiuser model,while eliminating the hidden risk of user identity and query information leakage.Performance evaluation and experimental results on real datasets demonstrate the effectiveness of the PIFS scheme.In the data service phase,a privacy-preserving data query and verification scheme(PDQV)is proposed to address the user identity privacy and query privacy leakage in MCS application.A pseudonym mechanism is used to protect the user’s identity and location privacy,the data tuple is verified and aggregated based on the Paillier cryptographic algorithm,meanwhile,a secure kNN algorithm is used to protect the user’s query privacy.In addition,malicious users who upload invalid data to interfere with aggregation results are pursued through digital signatures mechanism.The experimental results show that PDQV scheme achieves the privacy-preserving data query and verification.