Scalable QoS Degradation Locating from End-to-End Quality of Flows on Various Routes

Methods to infer the locations of QoS degradation from end-to-end quality of flows have been proposed. These methods find the minimum set of links that covers all the bad quality flows and infer the links as the locations of QoS degradation. Since the computational complexity to find the minimum set cover is high, these methods have a difficulty in the scalability for the large-scale networks. In this paper, we propose a scalable locating method in which a network is (logically) divided into subnetworks. Bad quality flows going across the sub-networks create the dependency among sub-networks in inferring. Resolving this dependency, our proposed method enables each sub-network to run inferring algorithm independently in parallel. Simulation results show that the proposed method can reduce the inferring time significantly while the accuracy of inferring is not degraded