Artificial Neural Network Based Framework for Predicting the Performance of Underwater Optical Wireless Link

Due to high speed and low latency, underwater optical wireless links have emerged as a promising candidate for underwater applications. In contrast to traditional acoustic links, underwater optical wireless links can contribute significantly in removing the bandwidth crunch of underwater communication networks. However, the underwater optical wireless links suffer from adverse channel conditions, and various phenomena, like scattering and absorption, degrade the performance of the underwater optical link by deteriorating the signal-to-noise ratio (SNR) of the received signal. Furthermore, the underwater channel conditions are highly dynamic in nature, and it becomes difficult for conventional techniques to model the non-linear behavior of the underwater link and capture the complex relation between various link parameters and the performance of the link. To address these challenges, an artificial neural network (ANN) based framework has been proposed in this work to predict the SNR for a high-speed underwater optical wireless link. An extensive dataset has been developed by considering key input features like the type of water channel, laser power, beam divergence, and link distance. The proposed ANN network has demonstrated accurate prediction of SNR of the underwater optical wireless link by delivering a coefficient of determination (R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>) with a value of 0.99704. In addition to this, the proposed underwater link uses 16-QAM modulation for transmitting data at a rate of 20 Gbps and can deliver an underwater optical wireless link of 10.55 m, while ensuring forward error correction (FEC) compatible bit error rate of 3.8 × 10<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">−3</sup>.