A heuristic intrusion detection system for Internet-of-Things (IoT)

Today, digitally connected devices are involved in every aspect of life due to the advancements in Internet-of-Things (IoT) paradigm. Recently, it has been a driving force for a major technological revolution towards the development of advanced modern computer networks connecting physical objects around us. The emergence of IPv6 and installation of open access public networks is attracting cyber-criminals to compromise the user specific security information. This is why the security breaches in IoT devices are dominating the headlines lately. In this research we have developed a random neural network based heuristic intrusion detection system (RNN-IDS) for IoTs. Upon feature selection, the neurons are trained and further tested at different learning rates with NSL-KDD dataset. Two methods are adopted to analyse the proposed scheme where the accuracy of RNN-IDS increased from 85.5% to 95.25%. Results also suggest that upon comparison with other machine learning algorithms, the proposed intelligent intrusion detection has higher accuracy in recognition of anomalous traffic from normal patterns.