The project explores the principles, feasibility and quantitative properties of applying emergent behavior to provide QoS. As a basis for this, efficient resource utilization in distributed communication systems is required. (Emergent behavior, in this context, is global behavior emerging from a large number of simple, loosely coordinated agent behaviors/network elements/etc.) This may be regarded as a distributed stochastic optimization problem. From the rare event theory based on the cross entropy optimization principle, simple agent behavior is derived, which finds good solutions to NP-hard problems, and with added heuristics, this deals well with QoS management issues. An important aspect of future service provision is to provide availability, continuity, etc. of services, tailored to requirements (i.e. dependability differentiation). These aspects are addressed in this project primarily with emergent techniques.
Project coordinator: Professor Bjarne E. Helvik