Regularized Zero Forcing Beamforming for Serving More Users in Energy-Harvesting Enabled Networks

This paper considers energy-harvesting (EH) enabled networks, which serve more users by a lesser number of transmit antennas. By transmitting energy and information over different fractions of a time-slot, it employs conjugate beamforming (match filtering) to amplify the power of received signals for EH and regularized zero-forcing (RZF) beamforming to suppress the multi-user interference for information decoding (ID), maintaining the computational efficiency in the beamforming design. The problem of max-min users' information throughput optimization subject to EH constraints is addressed by a path-following algorithm, which solves a simple convex quadratic problem at each iteration to generate a better feasible point. A similar optimization problem for the simultaneous wireless information and power transfer (SWIPT) is also addressed. The provided simulations show the advantage of the former over the latter.

I. Introduction

Next-generation wireless communication networks are envisioned to integrate data and energy services, for which information and energy are delivered over wireless mediums [1]. Such integration poses a serious challenge in managing interferences, which resist throughput in information transfer but provide an asset to energy transfer. Simultaneous wireless information and power transfer (SWIPT) over a time slot must rely on a single beamformer and thus hardly addresses such challenge [2]. For computational tractability of SWIPT, the work [3] has to use the conjugate beamforming (CB), which indeed is capable of amplifying the signal power for EH but is not best for ID because it cannot effectively suppress the interference as zero-forcing (ZF) or regularized ZF (RZF) beamforming does [4]. But by suppressing the interference and thus weakening the signal power, the latter is certainly not good for EH and thus is not suitable for SWIPT. The recent studies [5]-[7] have shown that the transmit time-switching (transmit-TS) of transmitting power within a fraction of the time slot and transmitting information within the remaining fraction essentially outperforms SWIPT.¹ In contrast to SWIPT, transmit-TS [5]-[7] does not rely on a single beamformer for both EH and ID but on an energy beamformer, which is capable of amplifying the signal power for EH, and an another information beamformer, which is capable of suppressing the interference for ID. The design of energy and information beamformers in [5]-[7] is mathematically formulated as a large dimensional nonconvex problem, which is not practically computed even for moderate numbers of transmit antennas. This paper proposes to rely CB for EH and RZF beamforming [9] for ID in considering the problem of max-min users' information throughput optimization subject to EH constraints. Its contributions are following: •

Instead of designing of energy and information beamformers as in [5]-[7], the paper aims to design the power allocation for CB and RZF, which is formulated as a moderate dimensional nonconvex problem, paving a way for practical computation;

•

A new path-following algorithm, which solves a simple convex quadratic problem at each iteration to generate a better feasible point is developed for computation of this nonconvex problem;

•

A new path-following algorithm is also developed for computation of a similar optimization problem for the conventional PS relying on CB to implement SWIPT. The development of computational solution for both optimization problems helps to show the superiority of the transmitTS relying on CB for EH and RZF for ID over the SWIPT relying on CB, which could not previously obtained from [5]-[7].