Artificial Noise Aided Hybrid Precoding Design for Secure mmWave MIMO System

Abstract

This paper exploits the potential of millimeter wave (mmWave) system, where large-scale antenna arrays are allowed to implement in small physical dimension. We investigate a novel hybrid beamforming design for joint data and artificial noise (AN) precoding and power fraction selection in massive multi-input multi-output (MIMO) system. We aim at the secrecy rate maximization problem with respect to hybrid precoders design. The challenge of this problem lies in its non-convexity. To address this issue, we decouple the design for analog and digital precoders. We conduct analog precoder to maximize corresponding channel gain. For digital data precoder design, we first remove the non-convex codebook constraint and propose an iterative algorithm for optimal equivalent digital precoder design. Then, reconsidering the constraint, we conduct the digital data precoder to approach to the optimal design. Next, aiming to maximize AN power aligned at the eavesdropper, AN precoder design is optimally derived in closed form. Finally, we get power fraction by one-dimensional (1-D) search. Simulation results indicate that our proposed AN- aided hybrid precoding scheme achieves better secrecy performance compared with existing hybrid precoding schemes.