AI BOX: Artificial intelligence-based autonomous abnormal network traffic response mechanism

Abstract

This paper presents an artificial intelligence-based approach to detecting intrusions. The artificial intelligence analyses intercepted packets and detect the presence of network attacks or abnormal network traffic based on artificial intelligence algorithms. If the network traffic behavior is unnatural or disruptive, it decides whether to interrupt, modify or isolate it immediately. Artificial intelligence models are used to detect anomalies. It enables or restricts data transmission to secure information technology (IT) and operational (OT) devices against attacks from anomalous network traffic, system outages, betrayal, and other threats. It can also move freely between heterogeneous networks to support data processing and transformation. We have developed machine learning models, packet tracking capabilities, and AI BOX environment options. We trained the model using two public datasets and obtained a 99.9% prediction accuracy. The AI BOX device successfully tested the effect of artificial intelligence models.