Abstract:
Optical networks play a crucial role in modern communications, providing high-capacity and high-speed data transmission. However, these networks are susceptible to various failures, particularly soft failures caused by gradual degradation of amplifiers and other components. These soft failures can lead to a decline in service quality over time, affecting the network’s performance and reliability. To address this challenge, this project develops a machine learning-based fault predictor using two distinct approaches. The predictor leverages comprehensive datasets to train models that can detect and predict soft failures, enabling proactive network management. The project’s results provide a strong foundation for mitigating service disruptions and ensuring consistent network quality.
