An Efficient Linear Precoding Algorithm for Energy Harvesting Transmitter with Finite-Alphabet Inputs and Instantaneous CSI

This paper investigates the precoder design in multiple-input multiple-output (MIMO) systems with energy harvester. The objective is to maximize the sum of mutual information (MI) over multiple time slots with causal knowledge of energy arrivals. Different from existing works, the system is considered with finite-alphabet inputs, and the transmitter is assumed to know not only the statistical channel state information (CSI) but also the instantaneous CSI of the current time slot at the beginning of each time slot. The optimal precoder design leads to an intractable multi- dimensional matrix optimization problem. We show that this difficult optimization problem can be transformed into an equivalent scalar optimization problem with respect to the transmit power. By doing this, the computational complexity of the precoder design is significantly reduced without any performance loss. Furthermore, we provide an efficient algorithm to solve the equivalent problem based on the dynamic programming. Simulation results verify the effectiveness of the proposed algorithm.