Hybrid Technique for DDOS Attack Detection Using Machine Learning

Abstract

The demand for the Internet is increasing every day, raising concerns about network security. Distributed Denial of Services (DDoS) attacks have harmed network availability for the decades and there is still no technique available to protect completely against these attacks. DDoS attacks are one of the biggest and fastest-growing cyber threats to network security. A DDoS attacks are an attempt to make a service resource unavailable, making it unusable for some time. Therefore, recognizing different forms of DDoS attacks with improved algorithms and higher accuracy while keeping the computing time less under control has become the most challenging factor. Firewalls and antivirus software are no longer sufficient in today's world to keep a company safe from the variety of assaults it faces. Traditional intrusion detection systems and firewalls can detect attacks based on signature patterns. Existing developments are insufficient to detect unknown threats. In order to identify and classify different types of anonymous attacks, it is necessary to apply intelligent technologies. Machine Learning (ML) has advanced significantly in terms of technology, bringing up lots of new research opportunities for addressing current and future network security challenges. In this research, machine learning methods and the significance of security in the context of various types of DDoS attacks and various ML classification algorithms such as Random Forest (RF), Naive Bayes (NB), K-Nearest Neighbour (KNN), Support Vector Machine (SVM), Adaptive Boosting (AdaBoost), eXtreme Gradient Boosting (XGBoost), etc. have been compared in terms of attack detection. In addition, a faster and more accurate machine learning-based attack detection system has been proposed for accurate and fast DDoS attack detection.