研究目的
To propose and demonstrate a simple and versatile nonlinear equalizer based on functional-link neural network for mitigating nonlinear effects in optical coherent transmissions.
研究成果
The FLNN-based nonlinear equalizer demonstrates comparable performance to DNN with lower computational complexity, effectively reducing BER by half in the tested system. It offers a promising approach for nonlinear impairment mitigation in optical communications.
研究不足
The study is limited to a specific transmission system (128-Gb/s single carrier dual-polarization 16-QAM over 600 km) and may not generalize to other systems without further investigation. The optimal number of mapped nodes in FLNN requires careful selection to avoid redundancy and noise.
1：Experimental Design and Method Selection:
The study employs a functional-link neural network (FLNN) as a nonlinear equalizer in a 128-Gb/s single carrier dual-polarization 16-QAM signal transmission system. The FLNN is compared with a deep neural network (DNN) for performance evaluation.
2：Sample Selection and Data Sources:
The experiment uses a pseudo-random binary sequence (PRBS) mapped to 16-QAM for transmission over 600 km of single-mode fiber (SMF).
3：List of Experimental Equipment and Materials:
Includes an AWG for signal generation, an IQ modulator, polarization beam splitter (PBS) and combiner (PBC), EDFA for amplification, and a coherent receiver with a digitizing oscilloscope.
4：Experimental Procedures and Operational Workflow:
The signal is transmitted over 6×100-km spans of SMF, with nonlinear equalization applied post-detection. The FLNN and DNN equalizers are compared based on BER performance.
5：Data Analysis Methods:
BER performance is analyzed versus optical launch power, and the computational complexity of FLNN and DNN is compared.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
