Modelling and simulation of DDOS Attack using SimEvents

Abstract

At the beginning of the 21st century, a new form of denial of service (DoS) attack emerged which is the Distributed DoS (DDoS). This new form of attack was launched on huge number of prominent websites such as yahoo, e-bay, Amazon, and buy.com, resulting in tremendous financial loses. DDoS attack has continued to increase over the years due to the rapid increase in internet users. Moreover, even more alarming is the fact that developers of DDoS tools have assumed unprecedented sophistication in their design methods, thus making their attacks highly destructive and undetectable. As a result of these reason and many others, researchers have focused their attention on the study of this new method of attack, they are particularly interested in studying its evolution, and with this knowledge they are being able to design anti-DDoS tools in order to prevent networks from falling into the clutches of DDoS attack. In this research work, a DDoS attack is simulated using MATLAB’s SimEvents, with the aim of finding the quantitative measure of its effect on the victim, experiments conducted in this study show that the server is scarcely utilized in its normal working condi-tions thus having high availability and low average utilization since it accepts requests only from legitimate clients. However, as the attacker launches an attack on the server, its utilization increases sharply and thus resulting in decrease in availability, this is because the server is flooded with illegal requests from the attacker as well as zombies from within the network domain. Additional study reveals that when a warm-up phase is added to the simulation of the server failure, the utilization suddenly increases due to the fact that the attacker seizes the opportunity of the slow recovery of the server to further overwhelm it and eventually push it into saturation.