Improved DetNet Algorithm Based on GRU for Massive MIMO Systems.

Massive Multiple-Input Multiple-Output (MIMO) technology is widely used to achieve high system capacity and spectral efficiency in wireless communication systems. The application of deep learning algorithms in massive MIMO signal detection has attracted great attention with the development of artificial intelligence in recent years. DetNet is a deep detection network for massive MIMO systems. By introducing the reset gate and update gate mechanism of the gate recurrent unit, this paper proposes an improved massive MIMO detection algorithm GRU-DetNet. Then, a hybrid neural network (Hybrid-DetNet) model with a parallel structure is further proposed. Simulation results show that the proposed GRU-DetNet and the Hybrid-DetNet achieve about 0.5 dB and 1 dB performance gain respectively over the existing DetNet scheme. The running time is reduced by about 50% under the same conditions. In addition, the proposed method has good universality since it can handle various modulation modes by training only a single network.