Implementation of Machine Learning Based DDOS Attack Detection System

The DoS/DDoS attacks are intended to bring the availability of the online services down to authentic users. The attacker seeds malware into the computers over the internet without the knowledge of the computer user/owner when they visit malicious websites, opening attachments from the unknown senders. DDoS attacks disrupt the availability of Web services for an indefinite amount of time by flooding the company’s servers with false requests and refusing genuine ones, resulting in economic losses due to unavailability delivered services. The goal of this paper is to demonstrate how to detect prototype DDoS attacks using a supervised learning model based on Support Vector Machines, which captures network traffic, filters HTTP headers, normalises the data based on operational variables such as rate of false positives, rate of false negatives, and rate of classification, and then sends the information to the appropriate training and testing sets. Various machine learning models, such as Navies Bayes, SVM, and suggested approaches based on Navies Bayes, were developed using the given characteristics. Bayes are a type of algorithm used to identify DDoS attacks. The results of our experiments suggest that Fuzzy c-means clustering is more accurate in recognising attacks.