Privacy preserving search and data retrieval from data clouds

Recently, cloud computing systems are considered as one of the most important advances in the field of Information Technology (IT) applications. However, security and privacy still form the main concern that slower the widespread use of these systems in the sensitive applications that need high level of privacy and security on their data. Outsourcing the data to unknown locations in the cloud needs securing these data by encrypting them. On the other hand, retrieving these data (or part of them) may need revealing sensitive data to unauthorized parties. Therefore, many techniques are proposed to handle this problem, which is known as Privacy-Preserving Data Retrieval (PPDR). These techniques tried to minimize the sensitive data that need to be revealed, which negatively affects the security and privacy of the data and/or the quality (or accuracy) of data retrieval. In this thesis, 9 security requirements are defined to satisfy a high level of security and privacy in a PPDR system based on the reported techniques in the literature. Together with these 9 security requirements, a retrieval efficiency requirement is defined to keep the retrieval efficiency high. Therefore, a new technique is proposed to satisfy these 9 security and 1 efficiency requirements (which are noted as 9+1 requirements in this thesis). The proposed technique utilizes Query Anonymous Authentication -which is derived from another newly proposed Radio Frequency Identification (RFID) anonymous authentication technique called HEADA- together with a multi-server setting to satisfy these requirements. The technique provides an efficient ranking-based data retrieval by using the cosine similarity of the TF-IDF vectors. The accuracy and ranking of the data retrieval is tested on three different datasets and shown to be high compared to other techniques as well as to the un-encrypted indexes and data. The analysis of the proposed technique also shows that the technique is able to satisfy all the 9 security requirements, which are unsatisfied completely in the techniques reported in the literature.