Anomaly detection in VoIP traffic with trends
Felipe Mata, Piotr Żuraniewski, Michel Mandjes, Marco Mellia · UvA-DARE (University of Amsterdam) · 2012
Abstract—In this paper we present methodological advances in anomaly detection, which, among other purposes, can be used to discover abnormal traffic patterns under the presence of deterministic trends in data, given that specific assumptions about the traffic type and nature are met. A performance study of the proposed methods, both if these assumptions are fulfilled and violated, shows good results in great generality. Our study features VoIP call counts, but the methodology can be applied to any data following, at least roughly, a non-homogeneous Poisson process (think of highly aggregated traffic flows). I.