Robust beamforming and power allocation for secrecy in multi-pair decode-and-forward relay networks

Abstract

We address a robust beamforming design and power allocation problem for a multi-pair one-way relay network, where the single-antenna users of each pair implement communication via a decode-and-forward multi-antenna relay. In this network, a secure user and other unclassified users coexist with a multi-antenna eavesdropper, and only the imperfect eavesdropper's channel state information (ECSI) is available. We aim to maximize the worst-case secrecy rate of the secure user while guaranteeing the quality of service of the unclassified users with a total power constraint. Motivated by this challenge, we explore the beamforming design and power allocation among the sources and the relay with the channel matrix error. Unfortunately, the proposed design constitutes a non-convex problem, which involves an infinite number of constraints because of the imperfect ECSI. To this end, we propose an efficient algorithm by using the techniques of semidefinite relaxation, successive convex approximation and S-procedure techniques. Following a similar optimization framework, we further investigate the scheme with the channel covariance matrix error. Simulation results validate the effectiveness of the proposed algorithm.