Self-Adjusting Networks

Projektbeschreibung

Grundlagen selbstregulierender Netzwerke

Netzwerke bilden die kritische und kostenintensive Infrastruktur unserer digitalen Gesellschaft. Neue rekonfigurierbare optische Technologien ermöglichen selbstregulierende Netzwerke: Diese Netzwerke können ihre Topologie dynamisch an die ihnen abgeforderte tatsächliche Arbeitslast anpassen und auf diese Weise die Leistung und Effizienz des Netzes verbessern. Das EU-finanzierte Projekt AdjustNet soll die Grundlagen für selbstregulierende Netzwerke legen, indem Modelle, Metriken und Algorithmen für die Anpassung der Online-Topologie entworfen sowie anhand von Fallstudien (z. B. Netzwerken für Rechenzentren) validiert werden. Die Motivation für die neuartige Methodik bildet die faszinierende Verbindung von selbstregulierenden Netzwerken mit Datenstrukturen und Informationstheorie.

Ziel

Communication networks have become a critical infrastructure of our digital society. However, with the explosive growth of data-centric applications and the resulting increasing workloads headed for the world’s datacenter networks, today’s static and demand-oblivious network architectures are reaching their capacity limits.
The AdjustNet project proposes a radically different perspective, envisioning demand-aware networks which can dynamically adapt their topology to the workload they currently serve. Such self-adjusting networks hence allow to exploit structure in the demand, and thereby reach higher levels of efficiency and performance. The vision of AdjustNet is timely and enabled by recent innovations in optical technologies which allow to flexibly reconfigure the physical network topology.
The goal of AdjustNet is to lay the theoretical foundations for self-adjusting networks. We will identify metrics that serve as yardstick of what can and cannot be achieved in a self-adjusting network for a given demand, devise algorithms for online adaption, and validate our framework through case studies. Our novel methodology is motivated by an intriguing connection of self-adjusting networks to known datastructures and to information theory.
AdjustNet comes with significant challenges since, similar to self-driving cars, self-adjusting networks require human network operators to give away control, and since more autonomous network operations may lead to instabilities. AdjustNet will overcome these risks and achieve its objectives by pursuing a rigorous approach, devising a theoretical well-founded framework for self-adjusting networks which come with provable guarantees and incorporate self–protection mechanisms.
The PI is well-equipped for this project and recently obtained first promising results. As the community is currently re-architecting communication networks, there is a unique opportunity to bridge the gap between theory and practice, and have impact.