Detecting malicious components in large-scale internet-of-things systems and architectures

Abstract

Current large-scale Internet-of-Things systems and architectures incorporate many components, such as devices or services, geographic and conceptually very sparse. Thus, for final applications, it is very complicated to deeply know, manage or control the underlying components, which, at the end, generate and process the data they employ. Therefore, new tools to avoid or remove malicious components based only on the available information at high level are required. In this paper we describe a statistical framework for knowledge discovery in order to estimate the uncertainty level associated with the received data by a certain application. Moreover, these results are used as input in a reputation model focused on locating the malicious components. Finally, an experimental validation is provided in order to evaluate the performance of the proposed solution.