AI-Enabled Modulation Format Identification in Fiber-Optic Networks
Abstract
This report presents a comprehensive documentation of the final year project titled, “AI-Enabled Modulation Format Identification in Fiber-Optic Networks”
Fiber optic communication systems have transformed modern telecommunications by enabling high-speed data transmission over long distances. Accurate identification of modulation formats used in these systems is crucial for their optimal performance. This project focuses on utilizing machine learning, specifically deep learning algorithms, to address the challenge of Modulation Format Identification (MFI) in Elastic Optical Networks (EONs). EONs offer a flexible and efficient infrastructure for high-capacity communication services, with dynamic optical spectrum allocation based on actual demand. The project involved simulations of optical networks with various modulation formats under diverse link conditions. Three deep learning models - Long Short-Term Memory (LSTM), Artificial Neural Networks (ANN), and Convolutional Neural Networks (CNN) - were trained and evaluated using different data representations, i.e., IQ Data, Amplitude histogram data, and Eye diagrams respectively. CNN was chosen as the best-performing model that achieves highest accuracy of 100% as compared to the other models. Successful implementation of this approach can significantly enhance the performance and reliability of Fiber optic communication systems, thereby advancing the telecommunications industry.