Research on an atmospheric turbulent channel equalization algorithm using the spatiotemporal feature fusion method

Atmospheric turbulence causes optical intensity scintillation and multipath interference on the transmitted signals in wireless optical communication systems, leading to an increase in the bit error rate (BER) at the receiving end. In this paper, an atmospheric turbulence channel model reflecting light intensity scintillation and multipath effects is proposed. The atmospheric turbulence channel with different distributions at different distances is constructed by applying the measured light intensity data. A new deep learning channel equalization algorithm based on spatiotemporal fusion is proposed for the constructed atmospheric turbulence channel, which solves the performance bottleneck of the traditional channel equalization, and effectively eliminates the fading effect of the atmospheric turbulence channel. Simulation results demonstrate that under 16QAM modulation, the spatiotemporal feature fusion algorithm significantly reduces BER compared to convolutional neural network methods. For instance, under log-normal distribution (0.42 km), the BER decreases from 10^-2 to 10^-5, while under Gamma-Gamma (10.3 km) and exponential Weibull (10.3 km) distributions, the BER improves from 10^-2 to 10^-6 and 10^-5. Similar performance enhancements are observed with DCO-OFDM 16QAM modulation. Compared with traditional methods, the new algorithm reduces the bit error rate to 10^-5-10^-6 under low signal-to-noise ratio conditions, fully demonstrating its efficiency and application potential.