Advanced quantisation schemes for physical layer security in vehicle communication

Security is a mandatory requirement in wireless communications since their broadcasting nature make them vulnerable to passive eavesdropping, as well as more advanced active threats, such as jamming or man-in-the-middle attacks. Physical layer security protocols have recently been deployed, derived from the characteristics of the medium for randomness extraction and further key generation. This has a two-fold advantage, as it allows for low computational complexity and high energy efficiency while potentially achieving (unconditionally) secure communications. The core part of physical layer security is channel quantisation, where estimates are harvested and converted into candidate keys. However, existing quantisation approaches do not thoroughly consider the trade-off among relevant metrics, such as the bit generation rate, the bit mismatch rate and key entropy. Furthermore, they fail to consider and exploit the peculiar propagation characteristics of vehicular environments, mainly due to their high mobility. Consequently, the current research aims to design a novel quantisation protocol to strike an optimal balance among the aforementioned metrics, using vehicle-to-vehicle communication as the primary use case. The first step consisted of a deep analysis of existing quantisation schemes through a comprehensive critical narrative of both existing and future developments in the field. This ushered in the novel algorithm’s design with holistically optimised metrics. Quantisation thresholds are analytically derived from the probability density function, stemming from channel modelling, which was meticulously tailored for the chosen propagation environment. Experiments showed that the extracted streams have a superior entropy compared to the state-of-the-art quantisation approach. In another contribution, the algorithm was extended to cope with a continuously changing channel reciprocity, a common phenomenon in vehicle communications. A new feedback block was designed to adapt the thresholds to the varying channel noise, attempting to optimise a newly defined metric. Since the latter includes both quantisation and information reconciliation, experiments confirmed that it is able to reach higher performances than fixed approaches, intrinsically exploiting the capability of existing and future error-correcting techniques. Finally, the algorithm was applied to multi-carrier wireless communication, as depicted by the current standards. In doing so, this study considers the estimation of time-varying coherence regions arising from the non-WSSUS behaviour of vehicle communications for the first time in the literature. Such a novelty was used to adapt the probing rate, avoiding sampling the channel for useless correlated estimates and improving the scheme’s efficiency and statistical robustness.