A fault detection model for edge computing security using imbalanced classification

Abstract

The edge computing-based Internet of Thing (IoT) has been growing drastically by taking advantage of edge computing which provides great assistance for IoT and mobile devices to complete sophisticated tasks. However, the rapid development leads to the neglect of security threats in edge computing platforms and their enabled applications, which has been one of the main limitations in the smart cities. In this article, we propose a fault and attack detection model for edge computing-based IoT systems to ensure the security of edge computing. Since the risk and fault cases are very imbalanced compared to normal cases, this paper proposes a novel fault detection algorithm by using the imbalance classification technique. Utilizing deep learning techniques, the proposed algorithm overcomes data overlapping problems occurring in many traditional oversampling methods and achieves outstanding performance. With this novel imbalance classification algorithm, the proposed security prediction model achieves pretty good performance on real-world edge computing applications.