Capabilities and limitations of local strategies in dynamic networks
This sub-project started in 2011 with the objective to explore the capabilities and limits of local methods for control and optimization of big dynamic networks. Our focus lies on overlay networks, which allow the interaction between actors of the OTF market (the clients) and service providers to support services and provide infrastructure. "Local" in this context means that the control and optimization is not performed by a central instance but distributed by the actors, based on their local information. In the first funding period, we focused our research on developing and analyzing algorithms, which e.g. allow the efficient search for services, the distributed organization of actors in groups, or to adapt the positioning of resources in an overlay to better serve the requirements of the clients.
In the second phase, we expand the focus of our research in two directions: Firstly, we deal with the dynamics of applications. For example, these dynamics manifests themselves as the fact that cooperation partner of an actor changes or the provided resources change. Supporting such dynamic interaction requirements needs a continuous adaptation of the overlay. Our focus lies on how we can answer changes of requirements with small modifications of the overlay. Secondly, we deal with external dynamics. Such dynamics are beyond the control of our algorithms. This could be faulty conditions or network load caused by foreign applications.
Specifically, the subproject is organized in the following working areas.