DDOS Attack Identification using Machine Learning Techniques
Subhashini Peneti, E Hemalatha · 2021
One of the major problems that the world faces today is cyber attacks. Denial of Service attacks have been one of most frequent attacks. Some of the mitigating techniques are whitelisting / blacklisting IP addresses, rate limiting etc. The major goal of any DoS attack is to bring down the reputation of the victim organisation. So, instead of facing the issues after the attack, it is always advisable to develop a smart detection system to detect and prevent DoS attacks. While there are many ways to detect DoS attacks, applying machine learning techniques to detect and prevent the attacks turns out to be a promising one. Since there is a lot of data available about DoS attacks, machine learning algorithms can detect patterns of these DoS attacks and thus apply these patterns to new requests and classify them as malicious or benign requests. We consider the CICIDS2017 dataset. The dataset has data related to requests of 7 days of a week. Among those, Wednesday's dataset contains records related to types of DoS attacks. Even though techniques like AdaBoost, XGBoost and neural networks can be applied, random forest gives great performance.