Deep Learning-Aided Iterative Detector for Massive Overloaded MIMO Channels

The paper presents a deep learning-aided iterative detection algorithm for massive overloaded MIMO channels. The proposed algorithm is based on the iterative soft thresholding algorithm for sparse signal recovery. The notable feature of the proposed scheme is that the detector has a reasonably low computational cost and contains trainable parameters which can be optimized with standard deep learning techniques. The number of trainable parameters is constant to the channel size, which promotes fast and stable training processes for the detector. The numerical simulations show that the proposed detector achieves a comparable detection performance to the state-of-the-art IW-SOAV detector for massive overloaded MIMO channels.