Underground Channel Model for Visible Light Wireless Communication Based on Neural Networks

An accurate channel model is both essential and challenging when designing a reliable wireless Visible Light Communication (VLC) system. Modeling the channel in a restricted and harsh environment such as an underground mine, is even more difficult. The objective in our work is to investigate a suitable design of a VLC link for an underground environment, while providing a reliable connectivity between the optical transmitter and optical receiver. In this paper, we present an experimental study that aims to determine a general model of an underground mine optical channel. Considering that the underground optical channel has a nonlinear dynamic behavior described by a nonlinear autoregressive exogenous (NARX) mathematical model, the system’s order is established with a feedforward neural network (NN), with a parallel architecture. In this work, in order to define a proper Underground Mine Optical Channel Model (UMOCM) for reliable VLC, we present an innovative approach based on a suitable mathematical model and neural networks (NNs).