Intrusion Detection System with an Ensemble Learning and Feature Selection Framework for IoT Networks

Abstract

The Internet of Things (IoT) and its applications are currently the most popular research areas. The properties of IoT are easily adapted to real-life applications but they disclose threats. In computer security, the Intrusion Detection System (IDS) plays an essential role in identifying and repealing malicious deeds in computer networks. The main purpose of this work is motivated by IoT security enhancement for IDS development using ensemble learning and proposing suitable methods for classifier performance. Initially, the preprocessing strategy is used for data cleaning, encoding and normalization, which are conducted in the RPL-NIDDS17 dataset. After that, the Synthetic Minority Oversampling Technique (SMOTE) is used to balance the dataset. Secondly, the Convolution Neural Network (CNN) has been used to extract the features from the dataset. From the extracted features, the optimal features are selected by the proposed Arithmetic Optimization Algorithm (AOA). Finally, it is applied to the proposed weighted majority voting classifier. The AOA with the Butterfly Optimization Algorithm (BOA) is utilized to integrate the predictions of different classifiers to select the most vote class. This enhances the chances of perceived RF, kNN, SVM kernel, Bi-LSTM and GRU classifiers. The proposed method experiment is conducted in the MATLAB platform with the RPL-NIDDS17 dataset. The proposed scheme shows better performances in terms of accuracy, error, sensitivity, specificity, FPR, F1_score, Kappa and MCC, which are compared with the existing methods and algorithms.