A Novel Approach for Network Vulnerability Analysis in IIoT

Abstract

Industrial Internet of Things (IIoT) offers efficient communication among business partners and customers. With an enlargement of IoT tools connected through the internet, the ability of web traffic gets increased. Due to the raise in the size of network traffic, discovery of attacks in IIoT and malicious traffic in the early stages is a very demanding issues. A novel technique called Maximum Posterior Dichotomous Quadratic Discriminant Jaccardized Rocchio Emphasis Boost Classification (MPDQDJREBC) is introduced for accurate attack detection with minimum time consumption in IIoT. The proposed MPDQDJREBC technique includes feature selection and categorization. First, the network traffic features are collected from the dataset. Then applying the Maximum Posterior Dichotomous Quadratic Discriminant analysis to find the significant features for accurate classification and minimize the time consumption. After the significant features selection, classification is performed using the Jaccardized Rocchio Emphasis Boost technique. Jaccardized Rocchio Emphasis Boost Classification technique combines the weak learner result into strong output. Jaccardized Rocchio classification technique is considered as the weak learners to identify the normal and attack. Thus, proposed MPDQDJREBC technique gives strong classification results through lessening the quadratic error. This assists for proposed MPDQDJREBC technique to get better the accuracy for attack detection with reduced time usage. Experimental assessment is carried out with UNSW_NB15 Dataset using different factors such as accuracy, precision, recall, F-measure and attack detection time. The observed results exhibit the MPDQDJREBC technique provides higher accuracy and lesser time consumption than the conventional techniques.