Ensemble Regression for 1-Bit Channel Estimation

Abstract

Employing 1-bit analog-to-digital converters (ADCs) is necessary for large-bandwidth massive multiple-antenna systems to maintain reasonable power consumption. However, conducting channel estimation with such 1-bit ADCs and with low complexity is a challenging task. In this paper, we propose to employ an Ensemble Regression (ER) model to conduct low-complexity and high-quality channel estimation. The amount of proposed computations are less than 3% of that proposed by similar deep learning (DL) methods, and in turn requires approximately 4% of the power consumed in computations while maintaining the same level of performance.