Abstract

The rapid evolution of telecommunications networks, driven by the growth of 5G, IoT, and edge computing, has introduced unprecedented complexity and dynamic challenges. Traditional network management approaches, reliant on static rule-based systems, are insufficient to address the real-time demands of modern networks. This study explores the integration of machine learning (ML) into dynamic network management, focusing on traffic prediction, resource allocation, and fault detection. Advanced ML models, including LSTMs, reinforcement learning, and autoencoders, were implemented and evaluated for their performance in enhancing network efficiency and reliability. Results demonstrated significant improvements in traffic prediction accuracy, bandwidth utilization, and fault detection rates, with ML models consistently outperforming traditional methods. Real-time testing in hybrid edge-cloud systems confirmed low latency and scalability across varying network scenarios. Despite challenges in data quality and computational requirements, the findings highlight the transformative potential of ML in telecommunications, offering a pathway to intelligent, adaptive, and proactive network optimization