Adaptive Post Quantum Cryptography Tuning for Constrained Devices Using Kyber and Saber

The rapid development of quantum computing poses a significant threat to classical cryptographic systems, including RSA and ECC, highlighting the urgency for Post Quantum Cryptography (PQC) solutions. However, the implementation of PQC algorithms, such as Kyber and Saber, in resource constrained environments remains a challenge due to their high computational and memory demands. This research aims to design and evaluate a lightweight, adaptive tuning framework for optimizing the parameters of Kyber and Saber algorithms in runtime systems with limited resources, such as CPUs and RAM. The proposed framework dynamically adjusts algorithm parameters based on the system operational conditions, including full load, multitasking, and idle states. A single device simulation approach, using software based limitations to mimic the conditions of embedded systems, was employed to test the framework. Experimental results show that lightweight configurations, such as Kyber512 and Saber-light, perform optimally under high load conditions, while more resource intensive variants excel in low load scenarios. The adaptive framework demonstrated significant improvements in encryption time, memory usage, and CPU load without compromising the security level required by NIST standards. This research provides a practical solution for implementing PQC in edge and IoT devices, offering a replicable, cost effective method for evaluating cryptographic performance in constrained environments. Future work will explore the integration of machine learning for dynamic parameter tuning and extend the framework to various hardware architectures.