An Edge-Based Quantum-Secure Fully Homomorphic Encryption Scheme

As artificial intelligence becomes increasingly integrated into data-driven decision-making and edge applications, concerns around cybersecurity and privacy preservation have intensified. Fully homomorphic (FH) encryption addresses some challenges by providing arbitrary secure artificial intelligence computations directly on encrypted data. However, current FH solutions are often too complex, insufficiently efficient, or vulnerable to future quantum threats. This article presents a new edge-based, quantum-resistant, FH encryption scheme tailored to the constraints of low-end and IoT devices. The core idea of the proposed scheme is built on a regulator-based design that employs an encrypted set of random values r1,r2,r3,rn satisfying 1nri1n, which plays a central role both during the encryption process and in enabling efficient homomorphic multiplication. The mathematical analysis proves that the designed scheme achieves robust homomorphic properties and computational agility while maintaining strong post-quantum security guarantees. Experimental evaluations demonstrate the feasibility of the scheme for real-world IoT scenarios, offering parametrizable encryption settings that adapt to varying security performance requirements. The proposed scheme achieved a lower complexity compared to classical and post-quantum schemes O(nl2) vs. O(n3),O(nlogn); and a high entropy of 7.27 vs. 7.13, 3.32.

• fully homomorphic encryption
• post-quantum cryptography
• edge computing
• Internet of Things security
• privacy-preserving computation
• homomorphic multiplication
• lightweight encryption
• AI security

• 9 — Industrie, innovation et infrastructure

• Cryptography and Data Security
• Quantum Information and Cryptography
• Cryptographic Implementations and Security
• fully homomorphic encryption
• post-quantum cryptography
• edge computing
• Internet of Things security
• privacy-preserving computation
• homomorphic multiplication
• lightweight encryption
• AI security
• Artificial Intelligence
• Computer Science