Homomorphic encryption and computation on encrypted data

• Grant number: 23/12755-8
• Support Opportunities: Research Grants - Young Investigators Grants
• Start date: February 01, 2025
• End date: January 31, 2030
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Theory of Computation
• Principal Investigator: Hilder Vitor Lima Pereira
• Grantee: Hilder Vitor Lima Pereira
• Host Institution: Instituto de Computação (IC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Associated scholarship(s): 25/03888-0 - Transciphering: block ciphers compatible with fully homomorphic encryption, BP.MS; 24/23608-9 - Private information retrieval protocols and applications, BP.MS

Abstract

Several recent technologies, such as cloud computing, internet of things (IoT) and smart cities, rely on large-scale data collection and data analysis. Encrypting the data protects people's privacy, but many of the functionalities obtained through these technologies become impossible when the data is encrypted. Fully homomorphic encryption (FHE) combines usefulness with privacy by allowing us to perform computation on encrypted data, thus ensuring that only the holder of the data can access them, but that external agents, such as a server in the cloud, can still use them. Despite the revolutionary potential of FHE, it is still difficult to use it in practice for several reasons: existing homomorphic ciphers are inefficient both in terms of execution time and memory; the libraries that implement these ciphers require a lot of technical knowledge about FHE details; and the security model that FHE assumes is not suitable for some applications. Hence, this project aims to study FHE and its applications, bringing solutions to make this cryptographic primitive more practical. For this, we will analyze FHE as a whole, from theoretical aspects, such as asymptotic cost of existing homomorphic ciphers, trying to improve them or even to create new ciphers with less costly operations, to practical aspects, such as optimized implementations for specific platforms such as FPGA. Improvements on these primitives will yield more efficient applications. Then we will use FHE to project privacy-preserving versions of advanced algorithms and primitives, such as machine learning algorithms and private information retrieval (PIR) protocols. (AU)