Security-Constrained Power Allocation in MU-Massive-MIMO With Distributed Antennas.

We consider the physical layer security in a multi-user massive multiple-input multiple-output system, where an eavesdropper (Eve) can intercept the information intended for any of the legitimate users. Several remote radio heads (RRHs), each equipped with massive antennas, are deployed to create the distributed antenna system (DAS). With the cooperative maximum-ratio transmission, we propose three secure transmit strategies, i.e., the centralized and autonomous full (CF & AF) transmissions with explicit artificial noise (AN), and the autonomous DAS (AD) scheme using implicit AN. We first derive closed-form deterministic equivalents of the signal to interference-plus-noise ratios (SINRs) of users and Eve for all the schemes. Then, we formulate two security-constrained power allocation problems aiming at: 1) maximizing the minimum user SINR and 2) minimizing the sum transmit power. Except for the optimization in the CF scheme, which yields linear programming, all others result in complementary geometric programming (GP), and are solved by successive approximations into ordinary GP. Numerical results demonstrate: 1) the significance of deploying the RRHs and autonomous power allocation to increase the user SINR; 2) the notable power-saving in the AF scheme; and 3) the improved power reduction in the AD scheme with a limited number of RRHs and antennas per RRH.