A Semi-Supervised Learning Approach for Network Anomaly Detection in Fog Computing

Abstract

Machine learning plays a vital role in the detection of network anomalies. In this paper, we first briefly examine the different categories of machine learning models, regarding to the acquisition of data label. With the support of fog computing, we then propose data-driven network intelligence for anomaly detection. The proposed framework includes fog enabled infrastructure and fog assisted artificial intelligence (AI) engine. Fog enabled infrastructure provides efficient computing resources for the selection of optimal learning model and sampling ratio. Fog assisted AI engine trains effective and robust semi-supervised learning models for detecting anomalies. We demonstrate that the optimal learning model achieves high detection accuracy and effective computational performance, with the close cooperation between infrastructure and AI engine in a fog computing environment.