Candidate
Kaveh Fazli
Title
Tools and Methodologies for FHE-Based Privacy-Preserving Image Processing Cloud Application Development
Supervisor
Guang Gong
Abstract
Using cloud services has become popular for both individuals and businesses in recent years. The popularity is because the cloud computing can drastically reduce the installation, operation and maintenance cost of IT infrastructures. However, using remote servers and cloud computing raises many security and privacy concerns. To address the privacy and security issues, the users encrypt their data before sending to the cloud servers. With the traditional encryption methods, the server cannot do any operation on the data. As a result, with the traditional encryption methods, when security and privacy are the main concerns, cloud servers are only used as storage.
Homomorphic encryption, on the other hand, allows operations on encrypted data without decrypting it. The major breakthrough in this subject was the introduction of the first fully homomorphic encryption (FHE) by Gentry in 2009. Fully homomorphic encryption is a scheme that supports any mathematical and logical operation on encrypted data. The inventor of an FHE scheme usually proves that the basic logic operations can be evaluated and concludes that any complex logical function can be decomposed into these basic operations and hence can be evaluated. The difficulty of the decomposition of a complex logical function into the basic operations supported by an FHE scheme is another factor that limits the FHE usage in real applications.
In this research we have developed tools and methodologies that allow FHE users overcome some of the FHE problems and be able to develop privacy-preserving cloud applications. Although most of our tools and methodologies can be used for any application types, our emphasis is on image processing applications.