Location-Aided Deep Learning-Based Channel Estimation for Hybrid Massive MIMO Systems

In hybrid analog-digital (HAD) massive multiple-input multiple-output (MIMO) systems, channel estimation is a critical challenge due to information loss caused by limited radio-frequency chains. The conventional compressive sensing (CS) algorithms suffer from unsatisfactory performance and high complexity. In this paper, we propose a novel location-aided deep learning (DL)-based framework to better exploit the sparsity structure of channels in the angular domain. Specifically, the entire angular space is segmented into many small regions and a dedicated neural network is trained offline for each region. During testing, the most suitable network is selected based on the location information of the user. Inside each network, the region-specific measurement matrix and channel estimator are jointly optimized, which not only improves the signal measurement efficiency, but also enhances the channel estimation capability. Simulation results show that the proposed approach significantly outperforms the state-of-the-art CS algorithm and the conventional measurement matrix.