Mobile crowdsensing (MCS) is a compelling paradigm that enables a crowd of individuals to cooperatively collect and share data to measure phenomena or record events of common interest using their mobile devices. Pairing with inherent mobility and intelligence, mobile users can collect, produce and upload large amounts of data to service providers based on crowdsensing tasks released by customers, ranging from general information, such as temperature, air quality and traffic condition, to more specialized data, such as recommended places, health condition and voting intentions. Compared with traditional sensor networks, MCS can support large-scale sensing applications, improve sensing data trustworthiness and reduce the cost on deploying expensive hardware or software to acquire high-quality data.

Despite the appealing benefits, however, MCS is also confronted with a variety of security and privacy threats, which would impede its rapid development. Due to their own incentives and vulnerabilities of service providers, data security and user privacy are being put at risk. The corruption of sensing reports may directly affect crowdsensing results, and thereby mislead customers to make irrational decisions. Moreover, the content of crowdsensing tasks may expose the intention of customers, and the sensing reports might inadvertently reveal sensitive information about mobile users. Data encryption and anonymization techniques can provide straightforward solutions for data security and user privacy, but there are several issues, which are of significantly importance to make MCS practical. First of all, to enhance data trustworthiness, service providers need to recruit mobile users based on their personal information, such as preferences, mobility pattern and reputation, resulting in the privacy exposure to service providers. Secondly, it is inevitable to have replicate data in crowdsensing reports, which may possess large communication bandwidth, but traditional data encryption makes replicate data detection and deletion challenging. Thirdly, crowdsensed data analysis is essential to generate crowdsensing reports in MCS, but the correctness of crowdsensing results in the absence of malicious mobile users and service providers become a huge concern for customers. Finally yet importantly, even if user privacy is preserved during task allocation and data collection, it may still be exposed during reward distribution. It further discourage mobile users from task participation.

In this thesis, we explore the approaches to resolve these challenges in MCS. Based on the architecture of MCS, we conduct our research with the focus on security and privacy protection without sacrificing data quality and users' enthusiasm. Specifically, the main contributions are, i) to enable privacy preservation and task allocation, we propose SPOON, a strong privacy-preserving mobile crowdsensing scheme supporting accurate task allocation. In SPOON, the service provider recruits mobile users based on their locations, and selects proper sensing reports according to their trust levels without invading user privacy. By utilizing the blind signature, sensing tasks are protected and reports are anonymized. In addition, a privacy-preserving credit management mechanism is introduced to achieve decentralized trust management and secure credit proof for mobile users;​ ii) to improve communication efficiency while guaranteeing data confidentiality, we propose a fog-assisted secure data deduplication scheme, in which a BLS-oblivious pseudo-random function is developed to enable fog nodes to detect and delete replicate data in sensing reports without exposing the content of reports. Considering the privacy leakages of mobile users who report the same data, the blind signature is utilized to hide users' identities, and chameleon hash function is leveraged to achieve contribution claim and reward retrieval for anonymous greedy mobile users;​ iii) to achieve data statistics with privacy preservation, we propose a privacy-preserving data statistics scheme to achieve end-to-end security and integrity protection, while enabling the aggregation of the collected data from multiple sources. The correctness verification is supported to prevent the corruption of the aggregate results during data transmission based on the homomorphic authenticator and the proxy re-signature. A privacy-preserving verifiable linear statistics mechanism is developed to realize the linear aggregation of multiple crowdsensed data from a same device and the verification on the correctness of aggregate results;​ and iv) to encourage mobile users to participating in sensing tasks, we propose a dual-anonymous reward distribution scheme to offer the incentive for mobile users and privacy protection for both customers and mobile users in MCS. Based on the dividable cash, a new reward sharing incentive mechanism is developed to encourage mobile users to participating in sensing tasks, and the randomization technique is leveraged to protect the identities of customers and mobile users during reward claim, distribution and deposit.