Automated Network Management
The management of software-defined networks constitutes a highly complex task. The integration of in-band monitroing, data-plane reconfiguration at runtime in addition to simultaneously supporting heterogeneous network applications is difficult and cumbersome. In this research area we focus on methods to allow adaptive application-tailored dataplane functions at line rate.
Model- and Data-driven Performance Evaluation of Networked Systems
We focus on developing analytical models that can be enriched through collected data and statistical techniques to understand and design complex networked systems. In this research area we consider different communication paradigms such as parallel computing in large cloud systems, unified caching and forwarding networks as well as time-sensitive networks. Our analytical models enable understanding the interactions in such systems as well as optimizing the ressource use therein
Multi-interface / Multipath Communication
Wired and wireless connectivity has long evolved beyond a single interface as devices become capable of utilizing mutliple communication links at the same time. A prominent example is utilizing WiFi and cellular communications simultaneously or spraying data packets over mutliple links in data center topologies. However, Multi-interface / Multipath Communication can be very difficult to use and if misconfigured may lead to performance drops. In this reasearch area we focus on (i) enabling mutlipath connectivity for different protocols such as Google's QUIC, (ii) enabling an application defined utilization of mutli-connectivity and finally (iii) evaluating the performance of multipath communication.
Immersive Multimedia Streaming
360° video as well as gaming streaming as well as 6 Degrees of Freedom (hologram) streaming face challenges with respect to interactivity as well as minimally required quality of service. In this research area we develop methods for scalable network-assisted interactive and immersive multimedia streaming that utilize in-network processing together with adaptive protocols on end-user devices.