CSI-based versus RSS-based Secret-Key Generation under Correlated Eavesdropping

TL;DR

This paper compares CSI-based and RSS-based secret-key generation methods in wireless systems, analyzing their capacities under correlated eavesdropping and high SNR, revealing significant performance differences.

Contribution

It provides a comparative analysis of CSI and RSS-based secret-key capacities considering correlated eavesdropper observations and high SNR conditions, with analytical expressions.

Findings

RSS-based key generation is heavily penalized compared to CSI-based methods.

At high SNR, the capacity penalty is a halved pre-log factor and about 0.69 bits.

The penalty diminishes as the eavesdropper's channel becomes highly correlated.

Abstract

Physical-layer security (PLS) has the potential to strongly enhance the overall system security as an alternative to or in combination with conventional cryptographic primitives usually implemented at higher network layers. Secret-key generation relying on wireless channel reciprocity is an interesting solution as it can be efficiently implemented at the physical layer of emerging wireless communication networks, while providing information-theoretic security guarantees. In this paper, we investigate and compare the secret-key capacity based on the sampling of the entire complex channel state information (CSI) or only its envelope, the received signal strength (RSS). Moreover, as opposed to previous works, we take into account the fact that the eavesdropper's observations might be correlated and we consider the high signal-to-noise ratio (SNR) regime where we can find simple analytical expressions for the secret-key capacity. As already found in previous works, we find that RSS-based secret-key generation is heavily penalized as compared to CSI-based systems. At high SNR, we are able to precisely and simply quantify this penalty: a halved pre-log factor and a constant penalty of about 0.69 bit, which disappears as Eve's channel gets highly correlated.