Underwater Wireless Optical Communication Utilizing a Semi-Supervised Twin Neural Network-Based Post-Equalizer with Interleaved Consistency Regularization
Zihao Du,Tianhao Zhang,Haipeng Wang,Chengye Cai,Xiaoxu Ma,Yuxin He,Zhongjun Ding,Dongfang Ma,Jing Xu
DOI: https://doi.org/10.1109/jlt.2024.3379217
IF: 4.7
2024-01-01
Journal of Lightwave Technology
Abstract:In this work, a novel semi-supervised twin neural network (semiTNN)-based post-equalizer (PE) is proposed and developed for high-speed 4-level pulse amplitude modulation (PAM4) underwater wireless optical communication (UWOC). Furthermore, semiTNN is enabled by a dual-drop strategy and interleaved consistency regularization (ICR), thereby only requiring a small amount of labeled PAM4 symbols as references to be well-trained. The experimental results over a 14-m/4-Gbps underwater link reveal that the proposed semiTNN PE paired with ICR, with only 5% labeled data, can achieve nearly the same bit error rate (BER) performance as the one trained by a fully-supervised learning process, resulting in 0.8-dB and 4.9-dB sensitivity improvements comparing with the traditional Volterra nonlinear equalizer (VNE) and linear feedforward equalizer (FFE) at a specific BER of 1×10 -3 , respectively. When the transmission distance is further extended to 56 m, the semiTNN PE with 10% labeled data can reach a data rate up to 4.26 Gbps under the hard-decision forward error correction (HD-FEC) threshold, which is 90 Mbps and 150 Mbps larger than that of VNE and FFE, respectively. Moreover, even facing bias-current fluctuations, turbulence, and bubbles, the UWOC systems supported by the semiTNN can still maintain a satisfying BER performance to some extent. This is the first time to utilize the proposed semiTNN PE combined with an ICR-based loss function for high-speed UWOC, and it can be highly beneficial to the practical deployment of NN-based PEs in the future.