Efficient Transceiver Design for Large-Scale SWIPT System with Time-Switching and Power-Splitting Receivers

Pham-Viet TUAN  Insoo KOO  

IEICE TRANSACTIONS on Communications   Vol.E101-B   No.7   pp.1744-1751
Publication Date: 2018/07/01
Online ISSN: 1745-1345
DOI: 10.1587/transcom.2017EBP3237
Type of Manuscript: PAPER
Category: Terrestrial Wireless Communication/Broadcasting Technologies
simultaneous wireless information and power transfer (SWIPT),  large-scale antenna arrays,  semidefinite relaxation (SDR),  sequential parametric convex approximation (SPCA),  alternating direction method of multipliers (ADMM),  

Full Text: PDF(912.7KB)
>>Buy this Article

The combination of large-scale antenna arrays and simultaneous wireless information and power transfer (SWIPT), which can provide enormous increase of throughput and energy efficiency is a promising key in next generation wireless system (5G). This paper investigates efficient transceiver design to minimize transmit power, subject to users' required data rates and energy harvesting, in large-scale SWIPT system where the base station utilizes a very large number of antennas for transmitting both data and energy to multiple users equipped with time-switching (TS) or power-splitting (PS) receive structures. We first propose the well-known semidefinite relaxation (SDR) and Gaussian randomization techniques to solve the minimum transmit power problems. However, for these large-scale SWIPT problems, the proposed scheme, which is based on conventional SDR method, is not suitable due to its excessive computation costs, and a consensus alternating direction method of multipliers (ADMM) cannot be directly applied to the case that TS or PS ratios are involved in the optimization problem. Therefore, in the second solution, our first step is to optimize the variables of TS or PS ratios, and to achieve simplified problems. After then, we propose fast algorithms for solving these problems, where the outer loop of sequential parametric convex approximation (SPCA) is combined with the inner loop of ADMM. Numerical simulations show the fast convergence and superiority of the proposed solutions.