Autonomous management for service specific overlay networks.

摘要

Overlay networks emerging as a main player in content delivery because they provide effective and reliable services that are not otherwise available. Extensive research has recently focused on the design of Service Specific Overlay Networks (SSON) to deliver media in a heterogeneous environment. This dissertation investigates the problem of SSON's management, and proposes an autonomous SSON management framework. The framework consists of a policy layer that in turn constitutes a set of Overlay Policy Enforcement Points (OPEP) and Overlay Policy Decision Points (OPDP). An OPEP is where policy decisions are actually enforced---policy decisions are made primarily at the OPDP. The research plan presented in this dissertation addresses the functionalities of these components.;The increased complexity and heterogeneity of SSONs led to the proposal of autonomic overlays management architecture. Overlays are viewed as a dynamic organization for self-management in which self-interested nodes can join or leave according to their specific goals. It dynamically adapts the behavior of the overlay network to the preferences of the user, network, and service providers.;To capture the overlay nodes autonomic behavior, a new approach for SSONs self-organized composition is proposed. Using a self-organizing approach, autonomic entities are dynamically and seamlessly composed into SSONs to achieve system-wide goals. The algorithm that encompasses that approach is powered by learning rules induced from biological systems, and endowed with filtering rules to achieve the highest possible performance.;Experimental studies are presented to demonstrate the performance of the proposed schemes.;To realize dynamic SSONs construction, a novel, fault-resilient semantic overlay for MediaPorts resource discovery is proposed. It allows services to be efficiently and accurately located, and is based on a widely studied family of chordal rings called the optimal chordal ring. In addition to the semantics of the services offered, our solution is based on the geographical locations of the nodes.