Detection of DDoS Attack by Machine Learning Algorithms

Abstract

A distributed denial of service (DDoS) attack is a form of cyber-attack in which the attacker attempts to deny the network/server services by inundating the network/server traffic with superfluous requests that make it incapable of serving legitimate user requests. Softwares that detects and mitigates a DDoS attack are commercially available, but the high cost of these software makes it difficult for small and mid-scale enterprises to afford it. The proposed work aims to fill this void by developing a highly effective, low cost DDoS attack detection model using machine learning  for small to mid-scale businesses that can be used to detect a highly accurate DDoS attack on a network.  In proposed DDoS attack detection model using machine learning techniques initially Data preprocessing prepares raw data and make it relevant for a model of DDoS attack detection by machine learning. Feature selection selects significant features to generate new dataset by using Correlation based feature selection(CFS) for the classification to enhance acceptability of model and accuracy. The new dataset is used to train classifiers. After training the classifier the test dataset is tested to detect whether each packet is a DDoS attack or normal data.