Topic Cluster Chair: Univ.-Prof. Dipl.-Ing. Dr. Siegfried Benkner

Research Agenda

The research cluster Algorithms and Computing deals with questions about networks which can often be modelled as graphs. These networks not only include communication networks which form the backbone of our digital society, but also other types of networks such as social networks. For the use of and communication through such networks, new algorithms are necessary that fulfil high efficiency and scalability requirements: networks are currently growing rapidly in many areas, which results in an increased energy consumption. Due to the immense popularity of data-centric applications (in the areas health, business, social networking, etc.), data traffic to and from data centres virtually explodes and consequently wide-area networks could soon reach their capacity limits. Studies also predict that data centres will account for approximately 5% of global energy consumption by 2025. Energy efficiency and sustainability are, therefore, an important research focus.

Many questions in the area of such large networks require the solution of algorithmic problems on graphs. Efficient graph algorithms are developed, theoretically analysed, and also empirically evaluated. Research activities also include dynamic, distributed and parallel algorithms. Examples of application areas for these algorithms are new types of communication technologies such as Software-Defined Networks, programmable Data Planes, reconfigurable optical networks, or “Self-* Networks” which optimise and repair themselves autonomously, which makes them more efficient, secure and reliable.

Research activities closely related to this, which form strong links to the research cluster Data and Knowledge, deal with algorithms for understanding neural networks and algorithms for gaining knowledge from social networks. Graph-based abstractions also serve as the basis for algorithms for and programming of future computer architectures which are not only highly parallel, but also increasingly heterogenous for reasons of energy efficiency. Task-based runtime systems make it possible to represent complex scalable and adaptive algorithms, the basis for computationally and data intensive applications, as dynamic graphs. They play an important role in the development of a new generation of parallel programming models. Networks are also of central importance for Cloud Data centres and supercomputers, which now include millions of processors within a system.

 Additional links

» Doctoral Programme in Computer Science and Business Informatics

» Doctoral Research Seminars

» Research Groups