Anomaly Detection for Cyber Internet of Things Attacks: A Systematic Review

Abstract

ABSTRACT From past few years, the Internet of things (IoT) is an emerging and encouraging technology that has gained prominence in the industries. Due to its increasing usages, a huge amount of data are exchanged within IoT architecture using the internet, which is why privacy and cyber-security are major issues. The heterogeneous nature of various technologies that are combined using IoT makes it problematic to provide security using prescriptive networking. The future of secure IoT depends on privacy issues. The research intends to improve security mechanisms based on intrusion and anomaly detection for IoT using deep learning. In this context, a systematic literature review (SLR) is conducted to identify ‘How to perform data transformation analysis of IoT dataset to detect anomaly detection for cyber IoT attacks? The SLR result found 24 datasets used for IoT analysis, 35 performance metrics to evaluate IoT problems, 6–42 features identified for detection, 42 preprocessing techniques have been used for transforming data, and 26 different methods and models were used to process the given problem. The SLR highlights further enhancement for the issue and identification of cyber-security in IoT. Anomaly detection can be done based on reinforcement deep learning after a thorough analysis of SLR.