Fast Convergence by Machine Learning Optimizer for Adaptive MIMO Equalizer Used in SDM Transmission over Coupled-Core 4-Core Fiber and 4-Core EDFA

摘要

We applied several optimizers from the machine learning field to an adaptive MIMO equalizer for SDM transmission. We experimentally compared their convergence properties in a SDM transmission over a 52-km coupled-core 4-core fiber and 4-core EDFA and showed over 22% faster convergence with Adam.