Attacker Detection in Massive MIMO Systems Over Spatially Uncorrelated Rician Fading Channels

Physical layer security is a promising research direction for the fifth-generation and beyond networks. This paper investigates the physical layer security of massive multiple-input multiple-output (MIMO) systems over spatially-uncorrelated Rician fading channels in time-division duplex mode. In such systems, uplink training stage is required for the base station to estimate channel state information (CSI) for design downlink precoding matrices and for uplink data detection. Illegitimate users, or attackers, could intentionally send jamming signals during the training stage to degrade the quality of the CSI obtained at the base station, thus affecting the performance. In this paper, we propose a method for detecting the presence of an attacker. Based on the fundamental properties of Massive MIMO communication, the base station can treat the jamming signals as additive white Gaussian noise. A threshold to detect the existence of the attacker is, therefore, computed in closed-form expression with a sufficiently large number of antennas at the base station. The key merit of our proposed method is that it only requires statistical channel state information and two training time slots to detect the jamming activity. Numerical results show that our proposed attacker-detecting method is effective over various system parameter settings. Furthermore, the benefits of the dominant line-of-sight (LoS) components have been testified. In particular, the detection probability is improved by about 1.5 times with the presence of the LoS components, while the false-alarm probability gets improved by more than ten folds.