A Comparison of Six Machine Learning Techniques for Cloud DDoS Attack Detection

Abstract

Cloud computing is a network access approach that allows for convenient, limitless, ondemand network access to a public computer resource pool. The DDoS assault is one of
the most serious risks to cloud users since it jeopardizes cloud providers' services and
renders them inaccessible to legitimate clients. Machine learning approaches are capable
of detecting DDoS assaults as well as preventing them. CICDDoS2019 dataset is used in
this work. This research involves four phases which are pre-processing, feature selection,
classification and parameter evaluation. The six machine machine learning techniques
implemented in this research are Logical Regression, Random Forest, Support Vector
Machines, Decision Tree, Naive Bayes and K-Nearest Algorithms. To evaluate the
effectiveness of each machine learning algorithm, accuracy, precision and recall are the
parameters used. It is found that Random Forest, Support Vector Machines and K-Nearest
Algorithms produce the best results in terms of accuracy, precision and recall.