Channel Estimation of mmWave Massive MIMO System Based on Manifold Learning

Abstract

In this paper, a channel estimation algorithm for millimeter wave communication system based on Manifold Learning Extreme Learning Machine (ML-ELM) is proposed. Particularly, the proposed algorithm uses the manifold learning to reduce the characteristic dimension of the received signal, such that the dimension of the received signal is reduced. Then, a one-shot training Extreme Learning Machine (ELM) is used to estimate the Channel State Information (CSI). Due to the low-dimensional date, the ELM can be better trained such that the channel is well estimated. Moreover, the computational complexity can also be reduced. Simulation results show that compared with LS, MMSE and Deep Neural Network (DNN) algorithms, the proposed algorithm provide a better performance, which meets the real-time requirements of the communication system.