Over the past decades, more and more people would like to outsource their data to the cloud due to its low price, high availability, and scalability. However, data outsourcing services have recently incurred severe data privacy concerns because users may outsource their private personal data. Recent data breach incidents remind us that the data we store on cloud services is under threat. Although users can simply encrypt their uploaded data, it may invalidate the basic functionality of cloud services, such as data searching. Therefore, we need techniques that preserve the privacy of outsourced data while ensuring functionality.

This dissertation introduces the algorithm designs and corresponding implementations for privacy-preserving search over different types of outsourced data. First, we propose a dynamic searchable encryption scheme over keyword-file indexes, which allows the cloud to search encrypted data without learning access-pattern information and update the index obliviously. The cloud can search and update databases within only a single communication round. Second, we design an Online Ride-Hailing scheme that enables the cloud to help a rider find the nearest driver over road networks without learning location information. The cloud server does not need to communicate with other servers while searching. Third, we introduce a privacy-preserving multi-range query scheme over numeral data. It does not need multiple communication rounds so that the storage server can directly search multiple range-matched results over encrypted data. Moreover, the ordering information among different values can be protected even after range comparisons. This line of work addresses different challenges when dealing with different types of data, and the comprehensive experimental results demonstrate the security and efficiency of our scheme. The proposed designs promote the development of privacy-preserving search techniques in outsourcing services and develop a sense of security for outsourcing service users.