REST-API based DDoS Detection Using Random Forest Classifier in a Platform as a Service Cloud Environment

Abstract

Cloud services are often delivered through HTTP protocol for ease and reduced cost for both service providers and users. The only drawback is that they are more prone to DDoS attacks. There is a need for a detection setup on this architecture that is lightweight, robust, and easily deployable with improved efficiency. We thus propose a novel DDoS detection setup using the REST-API (Application Programming Interface). The cloud platform we are using is Heroku, one of the first developed platforms as a service cloud platform by salesforce. The GitHub repository holds the pre-trained machine learning model and the cloud hosts the machine learning model through the GitHub repository. The HTTP request and response are sent to a cloud architecture through the Postman API client where the actual attack prediction is done. This makes our cloud burden-free as the detection is done outside its domain. The classifier is the Random Forest classifier, which is trained on the pre-processed KDD Cup 99 dataset. It shows the best accuracy results among most of the machine learning classifiers on this dataset with Information Gain as the feature selection technique.