FPGA-Based Lightweight Ascon Encryption for Edge-AI: A Systematic Review

Bringing AI to edge computing is one of the promising approaches for improving responsiveness compared to cloud-based processing. “Artificial Intelligence at the Edge” has been selected as one of the top 12 research and technology trends in this decade by IEEE. However, privacy issues should be considered more carefully when edge devices perform the inference phase to protect data. Although offering many advantages, edge platforms suffer from low computational power and processing performance. Hence, a lightweight and efficient implementation of data-authenticated encryption is essential. In 2023, NIST selected the Ascon algorithm as a new lightweight cryptography standard. Consequently, many studies have proposed and implemented the algorithm for edge-AI platforms due to its effectiveness. This paper systematically reviews the FPGA-based implementation of Ascon for Edge-AI computing platforms. We compare the area, throughput, and energy consumption of these proposals. Finally, we introduce open security issues for FPGA-based Edge-AI platforms.