A Lightweight Cyber-Security Defense Framework for Smart Homes

Abstract

Inevitably, the explosion of the Internet of Things (IoT) has alerted the modern human life. Apart from the benefits that this new technology offers to the users of the IoT devices, there are also dangers related to Cyber Security. Traditional methodologies that support and strengthen the Cyber Security can not be applied to the low consumption IoT devices. Hence, many attack detection methodologies and systems that respect the constraints of the IoT have been presented recently. In this paper, an anomaly detection mechanism is proposed that focuses on the threat detection, by combining statistical and machine learning methodologies to detect abnormalities in time-series of the network traffic. Moreover, this new framework is a lightweight Cyber Security solution, since its computational logic is included in the edge layer. The results of the experiments that were conducted in a test bed, indicated the high performance of the methodology in terms of Accuracy, Precision, Recall and F-measure.