Challenges when Partially Training a Machine Learning-Based Optical Communication System in Variable Experimental Conditions

Abstract

We present challenges when training a machine learning-based underwater wireless optical communication system in selected experimental scenarios. The system is tested under different conditions, that include minor beam misalignment and varying optical turbulence.