A Study on Performance Comparison of Algorithms for Detecting the Flooding DDoS Attack

A DDoS Attack is a cybercrime in which the attacker floods a server with internet traffic. This prevents authorized users from accessing the offered services. DDos attack can be a Flood attack, Application Level attack or Protocol Level attack. The motivation behind the attack can be fun or even business rivalry. In February 2020 Amazon Web Services (AWS) was attacked with DDoS. Before that GitHub was the target of the attackers. There are many different approaches proposed by researchers to detect DDoS attacks. The detection of DDoS attacks still remains a challenge. In this research paper, we compare the performance of three different techniques namely the Entropy-based algorithm, Support Vector Machine (SVM), and Multilayer Perceptron (MP)algorithm for detecting the DDoS attack accurately. Database containing 5000 normal records and 5000 DDoS attack records were given as input to the above-mentioned algorithms using the same simulation environment. Analysis has been carried out by identifying precision, recall, and accuracy. Among the three compared algorithms, SVM has shown the highest precision of 98.8 percent, recall of 98.56, and the accuracy of 99.37 percent. This concludes that SVM performs better as compared to Entropy-based algorithms and MP.