Improving an Ultra Lightweight AES for IoT Applications

The universe of the Internet of Things has been expanding every year, supporting several applications. However, this growth generates a massive amount of data that flows through the network daily, so the concern with the security of this information is essential. Cryptography is a good strategy for data protection, however, the classic algorithms are very costly in terms of processing and storage, which is challenging for IoT applications, where they are usually implemented in constrained resources devices. This paper proposes optimizations in the AES algorithm, specifically in the SubBytes and MixColumns stages, with the aim of improving secure IoT applications on devices with low computational power, making it a lightweight algorithm. The proposed optimizations show significant results in terms of performance without impacting the security of the algorithm. In terms of performance, the proposed version obtained an average reduction of 86.71% in execution time. Additionally, reductions of 31.82% and 89.04% were achieved in program and dynamic memory allocation, respectively, compared to the original AES. In terms of security, a 50.41% avalanche effect was obtained, along with a better distribution of O's and 1's in ciphertext compared to the original version. Moreover, the proposed version has been approved in NIST tests, making it more attractive for IoT applications.