On the inference and prediction of DDoS campaigns

摘要: This work proposes a distributed denial-of-service DDoS inference and forecasting model that aims at providing insights to organizations, security operators, emergency response teams during after attack. Specifically, our strives predict, within minutes, the attacks' features, namely intensity/rate packets/second size estimated number of used compromised machines/bots. The goal is understand future short-term trend ongoing attack in terms those features thus provide capability recognize current as well similar situations hence appropriately respond threat. Further, investigating campaigns by proposing clustering approach infer various victims targeted same campaign predicting related features. Our analysis employs real darknet data explore feasibility applying models on attacks evaluate accuracy predictions. To achieve goal, proposed leverages time series fluctuation techniques, statistical methods, approaches. extracted inferences from case studies exhibit promising reaching some points less than 1% error rate. could lead better understanding scale, speed, generates be adopted for immediate mitigation. Moreover, accumulated purpose long-term large-scale analysis. Copyright © 2014 John Wiley & Sons, Ltd.