Secure Constructive Interference Precoding for IRS-Aided NOMA Networks

In non-orthogonal multiple access (NOMA) networks, intelligent reflecting surface (IRS) and artificial noise (AN) can provide a double guarantee to achieve the secure transmission, especially essential for the far user with poor channel. However, AN is often eliminated via successive interference cancellation (SIC), which severely mitigates the secrecy energy efficiency. To tackle this issue, we propose a constructive interference precoding (CIP) enabled secure IRS-NOMA scheme, leveraging both the inter-user interference and AN to boost the legitimate transmission of far user, while inducing the eavesdropper to decode the deceptive information. In the CIP-NOMA scheme, we minimize the transmit power under the perfect channel state information (CSI), subject to the CIP constraint for the far user and eavesdropper, while guaranteeing the quality of service and SIC for the near user and the IRS unit modulus constraint. To handle this non-convex problem, we propose an alternating optimization algorithm. Specifically, by alternately optimizing the precoding vectors at the base station and the IRS reflecting matrix via the successive convex approximation and the penalty-based algorithm, respectively, a reliable solution can be obtained. Furthermore, to ensure the robustness, we also extend the scheme to a more practical case of imperfect CSI, where we utilize the S-procedure to deal with the channel uncertainty. Simulation results demonstrate that the proposed scheme can achieve better security performance with less energy consumption compared to the conventional NOMA in both cases.