Artificial-Noise-Aided Secure Transmission for User-Centric Cell-Free IoT Network

Abstract

This article investigates the physical-layer security for user-centric cell-free massive multiple-input-multiple-output (UC-CF-mMIMO)-enabled Internet of Things (IoT) network with multiple active eavesdroppers (Eves). We assume that each Eve intends to decode the information of the user (UE) closest to it and is intelligent enough to eliminate interference of other UEs perfectly. Two eavesdropping modes of noncolluding and colluding Eves are considered. Additionally, access points (APs) inject artificial noise (AN) into confidential data signal to hamper the eavesdropping of Eves. To assess the secrecy performance of UC-CF-mMIMO-enabled IoT network, we derive the worst case secrecy rate expression. Then, a max-min secrecy rate (MMSR) problem is formulated via joint optimization of AP clustering, AN selection, and power allocation, which aims to maximize the minimal secrecy rate among attacked UEs while adhering to the Quality-of-Service requirement of UEs, the limitation on the number of UEs and AN sent by each AP, and maximum transmit power limitation of APs. Due to the mixed-integer and nonconvex nature, the formulated problem is tackled via employing the ${\ell }_{1}$ -norm relaxation and successive convex approximation approach. Finally, we obtain a near-optimal solution to the MMSR problem by iteratively solving a sequence of second-order cone programmings. Simulation results demonstrate the superiority of proposed approach by the comparison with some existing schemes.