Secrecy rate optimization for intelligent reflecting surface aided multi-input-single-output terahertz communication

Intelligent reflecting surface (IRS) is an emerging paradigm to enhance terahertz (THz) communications with low hardware costs and reduced power consumptions. In this work, we study an IRS-aided THz multi-input-single-output secure communication system, where a base station (BS) transmits confidential signals to a single-antenna user with the presence of a single-antenna eavesdropper. We seek to maximize the secrecy rate by designing the discrete phase-shifts at the IRS and the precoder at the BS, which is an intractable nonconvex optimization problem. To solve it, we propose to iteratively design the above two matrix variables. Specifically, with a given phase-shift matrix, we can obtain the optimal precoder by the Rayleigh-Ritz theorem. And with a given precoder matrix, we propose a cross-entropy-based algorithm to obtain the optimal phase-shift matrix. Numerical results demonstrate that our proposed solution is able to improve the secrecy rate than that of the case without IRS.