IPv6 flood attack detection based on epsilon greedy optimized Q learning in single board computer

Abstract

Internet of things is a technology that allows communication between devices within a network. Since this technology depends on a network to communicate, the vulnerability of the exposed devices increased significantly. Furthermore, the use of internet protocol version 6 (IPv6) as the successor to internet protocol version 4 (IPv4) as a communication protocol constituted a significant problem for the network. Hence, this protocol was exploitable for flooding attacks in the IPv6 network. As a countermeasure against the flood, this study designed an IPv6 flood attack detection by using epsilon greedy optimized Q learning algorithm. According to the evaluation, the agent with epsilon 0.1 could reach 98% of accuracy and 11,550 rewards compared to the other agents. When compared to control models, the agent is also the most accurate compared to other algorithms followed by neural network (NN), K-nearest neighbors (KNN), decision tree (DT), naive Bayes (NB), and support vector machine (SVM). Besides that, the agent used more than 99% of a single central processing unit (CPU). Hence, the agent will not hinder internet of things (IoT) devices with multiple processors. Thus, we concluded that the proposed agent has high accuracy and feasibility in a single board computer (SBC).