Prediction: Edge Industrial IoT (EIIoT) Dataset Analysis and Determination of threats by using DNN Technique

Abstract

Usages of IoT devices are increased day by day in many applications, IoT devices are very weak and can be vulnerable easily, IoT devices are connected to different types of network and these devices are target of criminal to force cyber attacks. Providing security to the network environment that is connected with IoT devices are one of the crucial task. These factors created a scope of providing efficient security mechanisms for IOT computing environment. The various classes IOT cyber attacks are determine by using machine learning (ML) and Deep Learning (DL) models. To analyze cyber attacks we proposed a new method with an objective to combine ML with deep learning models. In this research we used ML techniques like Decision Tree (DT),Support Vector Machine (SVM), Random Forest (RF), KNN model Enhance Tree Classifier (ETC), and in addition to that we used deep neural network (DNN) to analyze various types of IOT attacks. The steps followed in our research are first gathered edge industrial IoT (EIIoT) dataset to conduct experiments, dataset are preprocessed using PCA (Dimensionality Reduction Technique) to discover features, imbalance in the classes are identified using over sampling method and basic scaling method. Experiments are conducted were ML and DL methods managing, detecting, and predicting various types of cyber attacks. Performance metrics are evaluated, binary classifier shows 99% accuracy, six class methods shows 97% accuracy, and fifteen class methods shows 94.5% of accuracy.

Topics & Concepts