Joint and simultaneous design of artificial noise and secure precoding

Abstract

In this thesis, a novel approach is introduced for enhancing communication security in wireless net- works. The proposed approach focuses on the combined optimization of secure precoding and artificial noise (AN), which is useful in downlink scenarios involving multiple-input and multiple-output wire- tap channels with several eavesdroppers. The core challenge in this domain stems from the intricate and irregular nature of the secrecy rate’s function, which is not straightforward to optimize. Traditional methods often use an alternating framework to determine the configurations of beamforming vectors and the AN covariance matrix, but these methods fall short in effectively maximizing the secrecy rate. Addressing the aforementioned issues, the research presents a groundbreaking algorithm for secure pre- coding that concurrently optimizes both the beams and AN covariance matrix. This is applicable for scenarios where transmitters have complete or limited information about eavesdroppers’ channels. The study establishes a refined version of the secrecy rate as a smooth function and formulates the first- order conditions for optimality as a nonlinear eigenvalue problem (NEP). A novel algorithm is proposed for efficiently identifying the primary eigenvector of the NEP, serving as a practical solution for secure precoding. Simulation results reveal that my proposed approach markedly enhances the secrecy rate, outperforming existing techniques.