Abstract
In this work, we consider coordinated beamforming design in a multi-user multiple input single output (MISO) interference channel where receivers are capable of energy harvesting (EH) in a power splitting manner, assuming only channel distribution information (CDI) is known to transmitter. Under a given requirement on the rate outage probability and the average EH power for receivers, we aim to maximize the system utility where the beamforming vectors and the power splitting factor are jointly optimized. This joint optimization problem involves complicated and non-convex constraints. We propose a centralized successive convex approximation (SCA) algorithm via solving a series of approximate convex problems iteratively. We further propose two decentralized implementations of the SCA algorithm: one serial implementation (DSCA-Serial) and the other parallel implementation (DSCA-Parallel). For a typical EH beamforming design, a tradeoff between the harvested energy and the system utility is often expected. Surprisingly, our simulations show that only less than 1% of the system utility is sacrificed while 10% of the transmitted power is harvested at receivers. All three proposed algorithms are shown to achieve near-optimal performance when compared with the exhaustive search method. Significant reduction in computational time is also observed for the proposed DSCA-Parallel when compared with DSCA-Serial, indicating a promising potential for a parallel implementation of decentralized SCA algorithm.