Supervised Learning-Based Noisy Optical Signal Estimation for Underwater Optical Wireless Communications

In this paper, we propose a novel artificial neural network (ANN)-based supervised learning algorithm for the classification of the received optical signal in underwater optical wireless communication systems. This work is regarded as the first of its kind in terms of both novel activation function for underwater optical systems and developed supervised learning algorithm. The proposed activation function is smooth and differentiable at zero. It is found that the proposed supervised learning algorithm for optical signal estimation performs well without any knowledge of the channel state information (CSI) of the underlying optical wireless channel. Furthermore, the bit-error-rate (BER) performance of the proposed ANN-based algorithm is independent of the learning rate.