Linear Precoder Design for SWIPT With Finite-Alphabet Inputs and Statistical CSI

This paper considers the transmit linear precoding problem for a multiple-input-multiple-output system with simultaneous wireless information and power transfer (SWIPT), where we assume that the transmitter only has statistical channel state information. The optimal precoder is designed based on maximizing the lower bound of average mutual information under the constraints of transmitting power and harvested energy level. Different from previous works on SWIPT, this paper formulates the design from the standpoint of realistic communication systems with finite-alphabet input signals instead of Gaussian input signals. The formulated problem is NP-hard, so a globally optimal solution cannot be found with polynomial-time complexity. However, by exploiting the structure of the problem, we develop two algorithms to obtain a near optimal solution, among which one is based on a semidefinite relaxation (SDR) technique and the other is based on power allocation. The SDR-based algorithm has more extensive applicability, while the power-allocation-based algorithm offers higher efficiency. The simulation results show the efficacy of the proposed algorithms.