Toward AI-Enhanced VLC Systems for Industrial Applications

In this paper, artificial intelligence tools are implemented in order to predict trajectory positions, as well as channel performance of an optical wireless communications link. Case studies for industrial scenarios are considered to this aim. In a first stage, system parameters are optimized using a hybrid multi-objective optimization (HMO) procedure based on the grey wolf optimizer and the non-sorting genetic algorithm III with the goal of simultaneously maximizing power and spectral efficiency. In a second stage, we demonstrate that a long short-term memory neural network (LSTM) is able to predict positions, as well as channel gain. In this way, the VLC links can be configured with the optimal parameters provided by the HMO. The success of the proposed LSTM architectures was validated by training and test root-mean square error evaluations below 1%.