In the digital era, computer networks form the backbone of modern organizations, ranging from enterprise data centers to IoT-driven infrastructures. As networks scale, maintaining their performance and preventing service interruptions becomes increasingly challenging. Traditional threshold-based monitoring systems are reactive in nature and often fail to predict faults before they occur. This paper proposes an AI-based Network Fault Detection and Management System that leverages Machine Learning (ML) to automatically predict, detect, and classify network faults by analyzing parameters such as latency, jitter, packet loss, bandwidth, and CPU utilization. A Random Forest Classifier is trained on a synthetically generated dataset that simulates real-world network conditions following ITU-T QoS standards. The model achieved an accuracy of approximately 93%, effectively distinguishing between normal and faulty states. In addition, the system integrates a Graphical User Interface (GUI) developed using Tkinter, enabling real-time testing and visualization of predictions. The combination of predictive analytics and an intuitive interface offers a robust and scalable solution for proactive fault management in computer networks.
Brindha et al. (Wed,) studied this question.