On pre-processing of generalized gamma fading channel samples for secret key generation

With the advent of innovative wireless technologies and the rapid growth of diverse connected devices, ensuring communication security has become a critical challenge for the future massively connected beyond 5th generation (B5G) wireless networks. The physical layer security (PLS) approach leverages the inherent physical medium characteristics of communication networks to provide secure communications. In particular, the wireless channel-based secret key generation (SKG) is a PLS approach that creates symmetric keys for message encryption/decryption by exploiting the randomness inherent in wireless channels while relying on the channel to be reciprocal between the nodes. This work proposes a novel SKG algorithm that emphasizes the pre-processing and quantization stages with the aim to increase the key generation rate (KGR) and the key agreement rate (KAR) between the legitimate nodes. The algorithm employs biasing of selective samples and sliding-window weighted averaging of the Generalized Gamma fading channel samples based on the channel's average contiguous duration (ACD) metric, a second-order fading statistic, to enhance the channel reciprocity. The proposed scheme may also be used to improve the key randomness rate (KRR) by using a non-contiguous multi-level quantization strategy that evenly sets the quantization intervals to accommodate a similar number of channel samples across these intervals. Thus, a desired trade-off between the KGR, KAR, and KRR to meet operational requirements can be achieved with the proposed scheme. Finally, a comprehensive analysis of the performance gains of the proposed scheme in relation to other notable works is conducted by investigating the impact of several channel and algorithmic parameters on the KGR, KAR, and KRR of these schemes.