Visible-Light Channel-Equalization Algorithm Based on the Fusion of a Neural Network and a Temporal-Feature Memory Structure

Light communication channels sometimes have changing characteristics that are difficult to predict. This article proposes a structure based on a convolutional neural network (CNN) structure, with a balanced algorithm for a residual structure. The optical-channel balance-coefficient vector is a time sequence that reflects the channel noise characteristics and has memory characteristics. The balancer algorithm can accurately learn the complex channel characteristics and calculate the compensation coefficient, based on the channel characteristics, to restore the original transmission signal. In addition, this article discusses a method of compensating for the receiving terminal when in the mobile state. Long short-term memory (LSTM) parameters are used to represent the sequence causal relationship in the memory channel, and the CNN and residual structure are used to refine the results to improve the reconstruction accuracy. The simulation results show that this algorithm can effectively eliminate the declining characteristics of the optical channel, improve the performance of the system-transmission error rate, solve the overall channel-damage problem, and accurately restore the original transmission signal at a faster convergence speed. In addition, compared with traditional competition-balance methods, this method achieves a better balance between performance and complexity, proving its potential and effectiveness.