Talk Title: Advances in Time Series Forecasting and Its Applications in Service Provider Networks

Modern service provider networks are rapidly evolving in scale and complexity, driven by 5G, cloud-native architectures, and increasingly dynamic user behavior. This talk explores recent advances in time series forecasting and their practical impact on network performance, capacity planning, and operational intelligence. Drawing from real-world experience in large-scale telecom environments, the session highlights how combining traditional statistical approaches with machine learning and AI techniques can significantly improve forecasting accuracy and decision-making. The presentation will showcase a novel residual modeling framework developed for 5G traffic forecasting, integrating network performance metrics with application-level insights to reduce prediction error and better capture traffic dynamics. It will also discuss multi-group sector classification strategies that enable targeted modeling for high-utilization network segments, leading to improved resource allocation and cost efficiency. In addition, the talk will cover advanced anomaly detection pipelines that support proactive identification of network issues, enhancing service reliability and operational visibility. The talk will conclude with key challenges and future directions, including scalability, data latency, and the role of AI-driven autonomous networks. Attendees will gain both theoretical insights and practical perspectives on applying advanced time series techniques to solve real-world problems in modern service provider networks.

Biography:

Mohammad Hossein Yaghmaee Moghaddam is a Full Professor in the Department of Computer Engineering at Ferdowsi University of Mashhad (FUM), where he has been a faculty member since 1999. He has taught a wide range of courses in computer engineering, led numerous industrial projects, and published extensively in his areas of expertise. Since June 2024, he has been affiliated with the Electrical and Computer Engineering (ECE) Department at the University of Toronto as Visiting Professor. He has also been actively involved in several data science and AI/ML projects in collaboration with telecom providers such as TELUS and Rogers Communications. His research interests include computer networking, Internet of Things (IoT), time series forecasting, anomaly detection, optical transport service provisioning, and the application of AI/ML techniques in telecommunications.