Machine Learning-Based Recommendation Framework for Post-Quantum Cryptographic Selection in IoT Environment

초록

Implementing post-quantum cryptography (PQC) in the IoT environment presents challenges because of the limited computing capabilities of IoT devices. Lattice-based cryptosystems, e.g., CRYSTALS-Kyber and SABER, provide high resistance to potential quantum attacks. However, selecting optimal PQC configuration schemes for specific IoT characteristics remains challenging, as parameters such as battery capacity, processing capability, and network latency are not accounted for in static approaches. To address the problem of dynamic selection of cryptographic algorithms for IoT applications, this paper proposes a machine learning-based recommendation system. In the proposed solution, a client-server architecture is implemented in which a server-based machine learning model processes synthetically generated operational parameters representing IoT conditions, including battery capacity, Central Processing Unit (CPU) usage, memory usage, and network latency, to recommend an appropriate PQC algorithm configuration. We use a synthetic dataset reflecting several IoT use cases to train and evaluate the model. Simulation-based evaluation results show that the proposed technique achieves high classification accuracy and outperforms static cryptography selection approaches.

키워드