Abstract: Homomorphic encryption (HE) enables secure computations on encrypted data without decryption, offering a transformative solution for privacy-preserving computation. This review presents a ten-year retrospective (2014–2024) on HE’s evolution since Gentry’s 2009 fully homomorphic encryption (FHE) scheme, which introduced the concept of performing arbitrary computations on ciphertexts. Early schemes were hindered by inefficiencies like computational overhead and noise accumulation. Over the past decade, significant advancements have addressed these barriers. Schemes such as BGV, BFV, and CKKS have been developed for efficient integer and approximate real-number computations. Algorithmic innovations like optimized bootstrapping and improved noise management have reduced complexity. Hardware acceleration using GPUs and FPGAs has enhanced performance, while integration with secure multi-party computation and zero-knowledge proofs has broadened HE’s applicability. Applications now span privacy-preserving machine learning, genomic data analysis, and financial analytics. Toolkits such as SEAL, HElib, and PALISADE have improved accessibility for developers and researchers. Despite progress, challenges remain, including balancing efficiency and security, and improving usability for non-experts. The article also explores HE’s reliance on lattice-based problems like Learning With Errors (LWE) and Ring-LWE, which provide quantum resistance. As hybrid cryptographic models emerge, HE is increasingly recognized as a key component in securing sensitive data in the postquantum era. This review highlights HE’s maturation from a theoretical concept to a practical solution, demonstrating its potential as a cornerstone for secure, privacy-preserving computing across industries.
Deshmukh et al. (Fri,) studied this question.