Machine Learning Assisted PHYSEC Attacks and SLP Countermeasures for Multi-Antenna Downlink Systems

Most physical-layer security (PLS) works employ information theoretic metrics for performance analysis. In this paper, however, we investigate PLS from a signal processing point of view, where we rely on bit-error rate (BER) at the eavesdropper (Eve) as a metric for information leakage. Recently, symbol-level precoding (SLP) has been shown to enhance PLS in the presence of an Eve. In this work, nonetheless, we introduce a machine learning (ML) based attack to which even SLP schemes can be vulnerable. Namely, this attack manifests when an Eve utilizes ML in order to learn the precoding pattern when precoded pilots are sent. With this ability, an Eve can decode data with favorable accuracy. As a countermeasure to this attack, we propose a novel security enhanced precoding technique. The proposed countermeasure yields high BER at the Eve, which makes symbol detection practically infeasible for the latter, thus providing physical-layer security between the base station (BS) and the users. In the numerical results, we validate both the attack and the countermeasure, and show that this gain in security can be achieved at the expense of only a small additional power consumption at the transmitter.