P2IDF: A Privacy-Preserving based Intrusion Detection Framework for Software Defined Internet of Things-Fog (SDIoT-Fog)

Abstract

The Software Defined Internet of Things-Fog (SDIoT-Fog) has provided a new connectivity paradigm for effective service provisioning. SDIoT-Fog uses network resource virtualization to provide services to heterogeneous IoT devices. However, data privacy, and security are the two major challenges that prevents faster realization of SDIoT-based frameworks. Motivated from the aforementioned challenges, we present a Privacy-Preserving based Intrusion Detection Framework (P2IDF) for protecting confidential data and to detect malicious instances in SDIoT-Fog network traffic. This framework has two key engines. Firstly, a Sparse AutoEncoder (SAE)-based privacy-preservation engine is suggested that transforms original data into a new encoded form that avoids inference attacks. Secondly, an intrusion detection engine is suggested that uses Artificial Neural Network (ANN) to train and evaluate the outcomes of the proposed privacy-preservation engine using an IoT-based dataset named ToN-IoT. Finally, experimental results showed that the proposed P2IDF framework outperforms with some recent state-of-the-art frameworks in terms of detection rate, accuracy and precision score.