A comparative analysis of machine learning techniques for botnet detection

Abstract

Day by day more and more devices are getting connected to the Internet and with the advent of the Internet of Things, this rate has had an exponential growth. The lack of security in devices connected to the IoT is making them hot targets for cyber-criminals and strength of botnet attacks have increased drastically. Botnets are the technological backbones of multitudinous attacks including Distributed Denial of Service (DDoS), SPAM, identity theft and organizational spying. The 2016 Dyn cyber attack involved multiple DDoS attacks with an estimated throughput of 1.2 terabits per second; the attack is the largest DDoS attack on record. In this paper, we compare three different techniques for botnet detection with each having its unique use cases. The results of the detection methods were verified using ISCX Intrusion Detection Dataset and the CTU-13 Dataset.