Research on Optimizing Digital Back-Propagation using Coefficient-Optimized Fourth-Order Runge-Kutta in the Interaction Picture

The digital back-propagation (DBP) is a method that can eliminate the effects of dispersion and nonlinearities on the signal during the fiber transmission. The main work of this paper is to replace the split-step Fourier method (SSFM) in DBP with the Fourth-Order Runge-Kutta in interaction picture (RK4IP) for reverse transmission. The article employs Bayesian optimization algorithm (BOA) to optimize the weights of various slopes in RK4IP, making it more suitable for the DBP algorithm. This effectively improves the Q-factor of the received signal at the same complexity level. In the experiment of transmitting the 20 GBaud 16QAM signal over the standard single-mode fiber (SSMF) spanning 9×100 km, the RK4IP-based DBP algorithm achieved the maximum improvement of 0.89 dB in the Q-factor compared to the traditional SSFM-based DBP algorithm at equivalent complexity.