Network and Service Management
Driven by the variety of applications, there is a plenty of different network architectures including related network optimization and modeling approaches. One of the classic areas of research at the Chair is traffic management for centralized and decentralized networks, but also new topics such as Blockchain for Networking or Internet of Things represents a core competence of the Chair.
The subject of cloud networks has recently been studied by the Chair in the field of edge and fog computing. Here, placement algorithms were tested and designed. The basis for this is always a distributed architecture, in which micro data centers in addition to conventional data centers provide computing power to operate services close to the user. Thematic overlap with research directions such as quality of experience and service as well as cloud monitoring helps to generate new ideas in this field.
Due to the complexity and heterogeneity, management approaches to communication networks have become indispensable in order to adapt the networks to a wide range of applications. The Chair is engaged in research with autonomous, intelligent network management, which offers possibilities to cope with today's emerging situations such as encrypted data traffic and high data volume in the networks.
The integrity and reliable operation of modern communication networks are constantly threatened by performance degradation, security breaches, and unforeseen faults. Research in this area focuses on developing sophisticated models and techniques to automatically identify subtle or blatant deviations from normal network behavior.
We investigate novel approaches based on time-series analysis and unsupervised machine learning (e.g., clustering and deep learning autoencoders) to detect anomalies in various network datasets, including traffic flows, resource utilization metrics, and Quality of Experience (QoE) indicators. A core challenge is handling the massive volume and high dimensionality of network telemetry data, particularly in high-speed, dynamic environments like 5G/6G and Edge Clouds. Our work aims to improve the real-time diagnosis of network failures and security incidents, ultimately contributing to more robust and self-healing network management systems.
The immense complexity and scale of emerging communication systems, such as 6G and Massive IoT, demand a paradigm shift from reactive manual configuration to fully autonomous operation. This research area centralizes the application of Artificial Intelligence (AI) and Machine Learning (ML) methodologies to design, optimize, and control future networks.
We focus on end-to-end automation, ranging from Intent-Based Networking (IBN), where AI translates high-level operational or business goals into executable network configurations, to optimizing resource allocation in virtualized network functions (VNF). Key research streams include:
-
Reinforcement Learning for dynamic routing and congestion control.
-
Federated Learning for distributed network intelligence across Edge Computing nodes.
-
The use of Digital Twin technology to simulate and predict the effects of management actions before deployment, ensuring proactive and efficient service delivery across heterogeneous network infrastructures.
