Hardware efficient arithmetic reconfigurable fully homomorphic encryption (ARFHE) accelerator of low power IoT based RISC-V processor

Abstract

Homomorphic encryption has emerged as an essential cryptographic approach for protecting data privacy in cloud computing and IoT applications. This research describes a high-performance BGV-FHE accelerator combined with a RISC-V processor to improve efficiency, security, and flexibility over current implementations. The proposed approach uses a Reconfigurable Booth Polynomial Multiplier to improve polynomial operations and reduce computational complexity. The Artix-7 FPGA-based accelerator improves encryption and decryption times by 12.5% compared to previous designs, with times of 1.05 µs and 1.01 µs, respectively. The proposed design provides a throughput of 68.12 MB/s, exceeding traditional homomorphic encryption accelerators. Furthermore, it provides optimal FPGA resource utilization by requiring only 8915 LUTs, 4120 FFs, and 4 DSPs, making it ideal for low-power applications. Compared to previous studies, the proposed accelerator provides improved processing efficiency (229.4 MB/s per W) while maintaining a strong 128-bit security level, ensuring resistance to quantum attacks. The flexibility of the design allows for easy scalability across different FPGA architectures. These enhancements establish the proposed work as the best option for real-time, secure computations in cloud-based encryption and IoT security frameworks.