Massive MIMO Data Detection Using 1-dimensional Convolutional Neural Network
Isayiyas Nigatu Tiba, Ben Baraka Kulimushia, Chrianus Kyaruzi Kajuna · 2020
In this work, we explore the use of an adaptive one-dimensional convolutional neural network (1d-CNN) for the massive multiple-input multiple-output (MIMO) data detection. We construct datasets corresponding to 100's of base station antennas serving 10's of transmit antennas and train an efficient detector by employing the feature extraction ability of CNNs. To improve the performance and be able to detect under the randomly varying channel scenario, we employ a data augmentation approach based on the existing computationally cheaper detectors. Our method is simple, and a non-iterative which works by unfolding the potential of deep network layers. We will show through simulation that the proposed method can significantly improve the learning under the randomly varying channel condition, and able to achieve competitive performances compared to the traditional detectors.