Low rate cloud DDoS attack defense method based on power spectral density analysis

The major threat to the availability of cloud computing resources and services is Distributed Denial-of-Service (DDoS) attack. DDoS is a multifaceted attack, and the detection of Low-rate DDoS (LDDoS) attack is a challenging task due to its stealthy and low-rate attack traffic behavior. The objective of the letter is to propose an approach which detects and mitigates the LDDoS attack in the frequency-domain. A power spectral density (PSD) based approach is proposed which monitors and analyzes real-time aggregate traffic for the attack detection. It mainly consists of five phases; the first four phases are in the time-domain while the last phase is in the frequency-domain. The approach is implemented on the OpenStack-based closed setup of a real cloud environment. The experimental results show that the approach is adaptive, and provides 3.7% false positive rate (FPR) and 4.9% false negative rate (FNR) which are comparable.