Fully homomorphic encryption (FHE) is a promising new technology that enables an untrusted party to efficiently compute directly on ciphertexts. For instance, with FHE a cloud server without access to the user's encrypted content can still provide text search services. An efficient FHE scheme would significantly improve the security of sensitive user data stored and processed on cloud servers. Significant progress has been made in bringing FHE proposals closer to practice. The PIs conduct a multi-pronged investigation of next generation homomorphic encryption schemes that do not require heavy computation or storage.
The investigation is structured in two research modules. The first module focuses on the theoretical foundation of FF-Encrypt - a new scheme by the PIs - including a thorough security analysis, selection of parameters for secure and efficient construction, and investigation of noise mitigation techniques. The second module compares the new scheme to existing ones with similar properties to establish its security and efficiency. Further techniques are studied to improve the scalability of the scheme. The performance of the developed theoretical homomorphic encryption schemes is scrutinized under the lens of optimized software and hardware implementations.
The investigation develops strategies for creating and optimizing next generation FHE schemes. In particular, the proposed new FF-Encrypt scheme has the potential to reduce parameter sizes while also eliminating evaluation keys, pushing FHE a step closer to the realm of real-world applications. The challenges in this project include the analysis of a new homomorphic scheme based on a new assumed computationally hard problem.
