Adaptive support detection based channel estimation for mmWave beamspace MIMO

Abstract

In this thesis, we focus our attention on a beamspace MIMO channel at mmWave frequencies, which has considerable advantages in large antenna arrays, and use the sparse characteristics of the mmWave channel to efficiently estimate the channel. Our main contribution is to develop a computationally-efficient support detection algorithm (adaptive support detection (ASD)) for channel vectors, which achieves as good performance as that of exhaustive search algorithm. The main idea of the proposed ASD algorithm is to determine the size of the support depending on the SNR and using the Saleh-Valenzuela channel model. In simulations, we demonstrate that our proposed ASD algorithm outperforms the other support detection (SD) based channel algorithms that estimate a fixed number of nonzero elements. Moreover, our algorithm achieves the estimation error as low as that of the exhaustive search algorithm, which determines the best support size in minimizing estimation error by changing all possible support sizes and comparing their estimation errors, but with a much reduced computational complexity.