Large Intelligent Surface Aided Physical Layer Security Transmission

Abstract

In this paper, we investigate a large intelligent surface-enhanced\n(LIS-enhanced) system, where a LIS is deployed to assist secure transmission.\nOur design aims to maximize the achievable secrecy rates in different channel\nmodels, i.e., Rician fading and (or) independent and identically distributed\nGaussian fading for the legitimate and eavesdropper channels. In addition, we\ntake into consideration an artificial noise-aided transmission structure for\nfurther improving system performance. The difficulties of tackling the\naforementioned problems are the structure of the expected secrecy rate\nexpressions and the non-convex phase shift constraint. To facilitate the\ndesign, we propose two frameworks, namely the sample average approximation\nbased (SAA-based) algorithm and the hybrid stochastic projected\ngradient-convergent policy (hybrid SPG-CP) algorithm, to calculate the\nexpectation terms in the secrecy rate expressions. Meanwhile, majorization\nminimization (MM) is adopted to address the non-convexity of the phase shift\nconstraint. In addition, we give some analyses on two special scenarios by\nmaking full use of the expectation terms. Simulation results show that the\nproposed algorithms effectively optimize the secrecy communication rate for the\nconsidered setup, and the LIS-enhanced system greatly improves secrecy\nperformance compared to conventional architectures without LIS.\n