Symbol-Level Precoding Design in IRS-Aided Secure Wireless Communication Systems

Abstract

This letter investigates an intelligent reflective surface (IRS) aided secure wireless communication system, in which a multi-antenna base station (BS) transmits confidential messages to multiple users via symbol-level precoding (SLP). We consider the two cases where the eavesdropper (Eve) has no channel state information (CSI) and global CSI, respectively, to investigate the possible worst and best eavesdropping performance, and the security improvement brought by the joint SLP-IRS design. For the former, we consider the user-receive signal-to-interference-noise ratio (SINR) balancing problem under the constraints of transmission power and security, and prove that SLP is inherently safe. For the latter, we assume that a smart Eve can intercept signals using brute-force and maximum likelihood detection methods. First, its received SINR is derived under the artificial noise framework, and then a two-step design scheme for SLP and IRS is proposed. In particular, an iterative penalty method is proposed for the IRS design step. Simulation results verify the analysis and show that IRS can significantly enhance the security of system.