Wavelet against random forest for anomaly mitigation in software-defined networking

摘要: Abstract Security and availability of computer networks remain critical issues even with the constant evolution communication technologies. In this core, traffic anomaly detection mechanisms need to be flexible detect growing spectrum anomalies that may hinder proper network operation. paper, we argue Software-defined Networking (SDN) provides a suitable environment for design implementation more robust comprehensive approaches. Aiming towards automated management prevent potential problems, present an identification mechanism based on Discrete Wavelet Transform (DWT) compare it another model Random Forest. These methods generate normal profile, which is compared actual real recognize abnormal events. After threat detected, mitigation measures are activated so harmful effects malicious event contained. We assess effectiveness proposed schemes using Distributed Denial Service (DDoS) port scan attacks. Our results confirm both as well routines. particular, correspondence between rates confirms enhance anomalous behavior by maintaining satisfactory false-alarm rate.