A Review on Detection of DDoS Attacks using Supervised Learning Techniques

Abstract

Digitalization has paved a way that threatens the users from using various network services because of the cyber-attacks. One of the major cyber-attacks in the network security world is the Distributed Denial of Service attacks. They target several systems in a distributed manner to deny the services on the end user machines or the targeted resources. This type of distributed attacks has extended over a period of time, and there are many types of DDoS attacks that causes major damage to the users’ systems. Detecting and mitigating these attacks take longer time than usual. In this paper, in order to detect these attacks, supervised learning algorithms are applied to the collected dataset directly. The dataset is mined using the Weka tool to predict which model produces accurate results. By applying the machine learning algorithms, each type of DDoS attack produces results based on various parameters. The parameters mainly focused in this paper are Accuracy, precision, recall and F-measure. These parameters help us to predict the detection rate quickly. With the results obtained from training the model, a thorough analysis is done with the algorithms. They are then tabulated for an analysis. Furthermore, ROC area computes the performance of the classifiers. Amongst the different types of classification algorithms used to predict, oneR algorithm produces more than 96% accurate results when compared with all the parameters.  For a more accurate technique to detect, this algorithm can be used in improving the detection rate of the DDoS attacks from flooding or crashing the user’ services.