作者: Juan Caballero , Dawn Song , Avrim Blum , Shobha Venkataraman , Jennifer Yates
DOI:
关键词:
摘要: Automatic identification of anomalies on network data is a problem fundamental interest to ISPs diagnose incipient problems in their networks. gather diverse sources from the for monitoring, diagnostics or provisioning tasks. Finding this huge challenge due volume collected, number and diversity be detected. In paper we introduce framework anomaly detection that allows construction black box detector. This detector can used automatically finding with minimal human intervention. Our also us deal different types collected network. We have developed prototype framework, TrafficComber, are process evaluating it using warehouse tier-1 ISP.