Nonlinear mitigation using multilayer photonic reservoir computing for long-haul transmission

Optical communication systems are essential for high-speed data transfer over larger distances, but they face significant challenges from nonlinear impairments such as chromatic dispersion and Kerr effects. These problems affect system performance, especially in long-haul scenarios. Traditional approaches, such as digital backpropagation and Volterra nonlinear equalizers, are computationally costly and not suited for real-time applications. We propose a multilayer photonic reservoir computing (RC) technique for reducing nonlinearities in long-haul optical communication systems. RC is a sort of recurrent neural network. This system directly processes optical signals by avoiding optical-to-electrical conversions and increasing efficiency. A multilayer design improves nonlinear correction by providing deeper signal processing and memory retention. The multilayer RC works significantly better than the single-reservoir system, according to simulation results. Improved signal quality is indicated by the multilayer RC’s quality factor (Q-factor), which reaches 11.5 compared with 9.5 for the single reservoir. In addition, the multireservoir system bit error rate is reduced to 2.9×10−7 compared with 4.5×10−5 for the single reservoir. Performance can be improved by integrating support vector machine regressors with different kernels. These results show multilayer photonic RC’s potential for real-world use in optical communication networks. The method paves the way for improvements in high-speed, long-haul optical transmission systems by offering a scalable and effective nonlinear mitigation solution.