Negative Feedback Noise Suppression and Equalization Algorithm Based on DNN

Channel equalization is one of the important methods for addressing inter-symbol interference (ISI) in wireless multi-path channel. In fast-changing channel, nonlinear equalization algorithms with feedback are commonly used to equalize received signals. However, due to noise and other factors, the convergence speed of the traditional algorithms are slow. In order to improve convergence speed and reduce BER, a deep neural network (DNN) is proposed. The algorithm uses neural networks to estimate the noise, which is fed back into the equalization algorithm to correct the convergence speed, correct the filter weights, and realize the optimization of the equalization algorithm. Simulation results show that, the algorithm converges faster and the fluctuation amplitude is reduced.