LMIPv6ATK:A Labeled Dataset Containing Multiple ICMPV6-DDOS Attacks

Distributed denial-of-service attacks using Internet Control Message Protocol version 6 messages can cause serious damage to IPV6 networks, resulting in huge economic losses. In this paper, we propose a packet feature-based dataset that can be used as a standard dataset for IPV6 attack detection systems and we named as LMIPv6ATK. The dataset is generated from a virtual network and contains both normal and multiple ICMPV6 DDOS attack traffic. Three classifiers: Naive Bayesian, decision tree, and random forest are used to test the dataset for the sake of tesifing the accuracy of our dataset. The accuracy of all three classifiers with multiple training is greater than 95%. The average accuracy was 96.73%. Among the results,Random forest takes the longest time, with an average accuracy of 98.77%. Thus,we can draw an conclusion that LMIPv6ATK dataset can be proved to be accurately applied to the attack detection system with high detection accuracy and low false alarm rate.