Edge Intelligence for Intrusion Detection
Yi Qian, Rose Qingyang Hu, Shengjie Xu · 2022
Abstract This chapter presents three key modules of edge intelligence and its use in intrusion detection. The entire architecture includes edge cyberinfrastructure, edge AI engine, and threat intelligence. Edge cyberinfrastructure acts as the supporting foundation in edge intelligence. It lies in the bottom layer, and it serves for the exchange of network traffic data seamlessly between itself and the edge AI engine. Edge AI engine is essential to data pre-processing, model learning, and predictive analytics. It lies in the middle layer. Threat intelligence lies in the upper layer. It integrates a delivery model of many applications for intrusion detection, including real-time traffic monitoring system and emergency alarming system. A preliminary study is conducted using two popular intrusion detection datasets. The study is carried out by applying logistic regression, random forest, decision tree, and ensemble model. A fivefold cross-validation is performed on both datasets to obtain the prediction results. Confusion matrices are then formed given the values of prediction results. Based on the values, the mean error rate (MER) is calculated. The proposed edge intelligence for intrusion detection is evaluated in terms of detection accuracy and computational efficiency. The preliminary results demonstrate the feasibility of edge intelligence for intrusion detection.