Coordination in constrained and natural distributed systems

Projektbeschreibung

Für perfekte Berechnungen in verteilten Systemen

Informatik und Biologie nähern sich einander immer weiter an. In diesem Zusammenhang scheint das verteilte Rechnen den besten Ansatz für die Modellierung natürlicher Prozesse zu bieten. Diese Methode orientiert sich an den hochrangigen Gestaltungsprinzipien biologischer Systeme. Für die Untersuchung komplexer selbstorganisierender Prozesse in der Natur wird jedoch dringend ein neuer Rahmen benötigt, da es bisher keine Modelle für verteiltes Rechnen gibt, die alle Aspekte gleichzeitig berücksichtigen. Das EU-finanzierte Projekt CoCoNat wird diese Lücke durch die Theorie der verteilten Synchronisation und Koordinationsaufgaben in eingeschränkten Modellen des verteilten Rechnens schließen, indem es die Linse des verteilten Rechnens zur Modellierung natürlicher Phänomene verwendet. Es wird außerdem verschiedenen Bereichen der Technik und Informatik zugutekommen, indem es verschiedene Synchronisations- und Koordinationsaufgaben löst, wie etwa bei Glühwürmchenpopulationen oder der embryonalen Entwicklung.

Ziel

In recent years, an algorithmic theory of natural and biological systems has been increasingly advocated as providing a much needed framework for investigating complex self-organising processes in nature. This project contributes to this research program by employing the distributed computing lens to model natural phenomena. Biological systems exhibit many properties also studied in distributed computing: they comprise several independently acting entities, tend to operate in noisy and dynamic environments, thus requiring them to be highly-resilient and adaptive, solve intricate coordination tasks, and display sophisticated communication techniques.

This project aims to develop the theory of distributed synchronisation and coordination tasks in restricted models of distributed computing. These tasks are some of the most fundamental problems in distributed computing, as they are essential in computer networks as well as numerous other areas of engineering and computing. In addition, they are ubiquitous in natural and biological systems, ranging from molecular to population-level systems, which are known to solve various synchronisation and coordination tasks: examples include symmetry-breaking during the development of the nervous system, consensus decision making in species communities, and synchronisation in firefly populations and embryonic development.

Unlike computer networks, biological distributed systems have unique features: (1) the agents typically have limited computational abilities, (2) communication is unreliable and restricted, (3) the system has a dynamic spatial structure, and (4) the environment may be noisy. Currently, distributed computing models that consider all aspects simultaneously are lacking. The proposed research approaches this goal from multiple angles by developing new models and analysis methods for determining the limitations of synchronisation and related tasks in both strong and weak models of computing.