A Defensive Mechanism based on PCA to Defend Denial-of-Service Attack
P. Rajesh Kanna, K. Sindhanaiselvan, M. K. Vijaymeena · International Journal of Security and Its Applications · 2017
The Network security is the implementation of policies to prevent unauthorized access and to detect attacks in network traffic.The main aim is to preserve the system with respect to confidentiality, availability and integrity.Various network security threats are affecting the Internet and the most important one is Denial of Service (DoS) attacks, which are most difficult to address as they are very effortless to launch, difficult to track.Multivariate Correlation Analysis (MCA) is an existing method is used to detect both unknown and known attacks.It is done by extracting the geometrical correlations between network traffic features.The unknown and known DoS attacks will be differentiated by the system from legitimate network traffic.But it cannot be possible to detect the Land, Teardrop and Neptune attacks.This in-turn increases the system complexity and it could be lowered by using Principal Component Analysis (PCA).PCA performs dimensionality reduction in order to reduce the cost of computation.Histogram based images are compared in order to detect all known and unknown DoS attacks efficiently.The detection of DoS attacks will be improved and it can be measured using parameters such as Accuracy, Detection rate, True negative rate and the False positive rate.Anomaly based classification could be used for better outsourcing performance in detection of unknown and known DoS attack.