SA Dots: A Sensitive and Self-adaptable Framework for Cross-Region DDoS Processing

This paper proposed a SA DOTS architecture and system with attack types awareness and alarm threshold self-adaptable mechanism to improve the capability of DDos mitigation which across multi-networks. Firstly, we designed an expanded Dots protocol that includes attack type extensions which enables accurate sensing of attack types. Then, the feedback receiving module are extended in Dots framework to realize the alert threshold self-learning ability which based on time slice. It can effectively establish dynamic threshold curve and intelligently match real attack distribution. Verified by actual network, SA DOTS mechanism can keep high correct rate of warning while increase the ability to detect attacks. While using SA DOTS architecture, the experiment analysis shows that the accuracy increased by 70.22% and attack detection probability increased by 19.5%.