Machine Learning–Enhanced UWOC System for Robust Underwater Communication with FEC Integration

Abstract

Underwater wireless optical communication (UWOC) often suffers from channel-induced bit errors that degrade adaptive threshold detection, particularly due to dynamic turbidity conditions. Standard approaches of applying mitigation techniques may improve the link performance to an extent. In this study, we have incorporated an integrated approach based on machine learning (ML) into the UWOC system, along with forward error correction (FEC), and tested it on various turbidity levels. This has significantly reduced bit errors and improved link performance. To train the ML models, experimental data were collected using an underwater testbed, with various turbidity levels up to seven nephelometric turbidity units (NTU) to simulate different underwater conditions. The results revealed that across different water models, fluctuations due to angle of arrival (AoA) changes cause a one-order change in magnitude for clear water. In contrast, for harbor water, changes occur only at short ranges. Similarly, the signal-to-noise ratio changes sharply in harbor water compared to clear water as the communication range increases. When we studied the effect of ML models and FEC, we found that an additional 5 m of error-free range can be achieved in underwater environments for 100 Mbps UWOC links using a 2-watt green laser. In different scenarios, ML models alone were able to provide an error reduction from 10⁻¹ to 10⁻² across all water conditions.

Received: 6 November 2025 | Revised: 28 January 2026 | Accepted: 30 March 2026

Conflicts of Interest

The authors declare that they have no conflicts of interest to this work.

Data Availability Statement

Data are available from the corresponding author upon reasonable request.

Author Contribution Statement

Shreyas Jain: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Writing – original draft, Writing – review & editing, Visualization. Harshit Kumar Sharma: Software, Validation, Investigation. Appala Venkata Ramana Murthy: Conceptualization, Resources, Data curation, Writing – original draft, Writing – review & editing, Visualization, Supervision, Project administration.