Deep Unfolding Based Simplified Information Geometry Approach for Massive MIMO-OFDM Channel Estimation

In this paper, we investigate the channel estimation in massive multiple-input multiple-output orthogonal frequency division multiplexing (MIMO-OFDM) systems. The recently proposed simplified information geometry (SIG) algorithm offers a promising solution for channel estimation with relatively low computational complexity. However, the damping factors used in the SIG algorithm are based on a heuristic strategy, which results in inconsistent performance. To address this issue, we propose a deep unfolding based SIG (DU-SIG) approach in this paper. Specifically, each iteration of the SIG algorithm is unfolded into a layer-wise structure resembling a neural network and the damping factors are optimized based on deep unfolding technique. Notably, the optimized damping factors can be directly integrated into the original SIG algorithm to improve the performance of channel estimation without increasing the computational complexity. Simulation results validate the effectiveness and superiority of our proposed algorithm.