Beam-Domain Channel Estimation for FDD Massive MIMO Systems With Optimal Thresholds

Abstract

Massive multiple-input multiple-output (MIMO) systems are expected to operate in the frequency-division duplex (FDD) mode, which is feasible in the channel environment with limited scattering. Since accurate channel estimation is critical for gaining unprecedented capacity, we investigate beam-domain channel estimation and feedback for FDD massive MIMO systems. In particular, we focus on the threshold-based method for channel estimation, where an enhanced estimator is proposed to exploit the common support among all beam-domain channels. For threshold-based estimation, we derive its closed-form mean-squared error (MSE) expression, and obtain an optimal threshold as a function of sparsity, noise variance, and channel variance, and a simplified threshold, which is a function of noise variance only. For the enhanced estimator, we present a threshold to identify the common support, with which an algorithm is designed to improve the estimation accuracy. As for channel feedback, we suggest to feed back only significant elements (above the given threshold) in the beam domain. Numerical results validate our derived MSE expression and demonstrate the superior performance of proposed threshold-based estimators.