arXiv (Cornell University) · 2016 · 49 citations · 38 references
ComputingData ScienceNetwork SecurityNetwork VisualizationQueuing TheoryNetwork ReliabilityNetwork Traffic AnalysisRoutingSystem MonitoringAnomaly Detection In RoutingAnomaly Detection In TracesDisruptionsNetwork Change-point DetectionNetwork Failure DetectionNetwork Fault DiagnosisNetwork MonitoringPacket Loss Analysis
Understanding network health is essential to improve Internet reliability.\nFor instance, detecting disruptions in peer and provider networks facilitates\nthe identification of connectivity problems. Currently this task is time\nconsuming for network operators. It involves a fair amount of manual\nobservation because operators have little visibility into other networks. In\nthis paper we leverage the RIPE Atlas measurement platform to monitor and\nanalyze network conditions. We propose a set of complementary methods to detect\nnetwork disruptions from traceroute measurements. A novel method of detecting\nchanges in delays is used to identify congested links, and a packet forwarding\nmodel is employed to predict traffic paths and to identify faulty routers in\ncase of packet loss. In addition, aggregating results from each method allows\nus to easily monitor a network and identify coordinated reports manifesting\nsignificant network disruptions, reducing uninteresting alarms. Our\ncontributions consist of a statistical approach providing robust estimation for\nInternet delays and the study of hundreds of thousands link delays. We present\nthree cases demonstrating that the proposed methods detect real disruptions and\nprovide valuable insights, as well as surprising findings, on the location and\nimpact of identified events.\n
38
Allan L. Coates, Alfred O. Hero, Robert Nowak et al. · IEEE Signal Processing Magazine · 2002 · 462 citations