High-Speed Coherent 400 Gbps/λ DP-16QAM Optical Transmission Enabled by TCN Based Nonlinear Equalizer

In this work, we propose a temporal convolutional network (TCN)-based nonlinear equalizer to compensate Kerr effect-induced nonlinear distortions in high-speed long-haul coherent transmission. By leveraging the multi-layer dilated convolutions, TCN can exponentially expand its receptive field without increasing model depth, which could effectively capture the long-range dependencies in data sequences. For 2×400 Gbps dual-polarization (DP)-16QAM signal, two TCN configurations with different residual blocks are systematically evaluated. The Q-factor under TCN equalization exhibits at most 2.75 dB improvement compared to Volterra-series equalizer which exceeds the hard-decision forward error correction (HD-FEC) limit at a transmission distance of 80 km with 10 dBm optical input power. To the best of our knowledge, this is the first time to employ TCN equalizer in high-speed coherent optical communication systems.