The paper presents a novel deep learning algorithm that can be used to cancel the interference caused by multi-input multiple-output system called MIMO-NOMA. The proposed method is designed to address the challenges of the traditional SIC schemes, such as the high computational complexity and error propagation. It utilizes a deep neural network to directly decodes the signals from each user to the corresponding data. The proposed scheme is performed through a simulation to evaluate its performance. It shows that it can perform better than the traditional approach when it comes to signal detection and channel estimation. The proposed algorithm performs better than the traditional SIC schemes in terms of its bit error rate. It also maintains low computational complexity.

Deep learning, MIMO-NOMA, Successive interference cancellation, Bit error rate, Softmax.