Efficiency Enhancement of Intrusion Detection in Iot Based on Machine Learning Through Bioinspire

The Internet of Things of the future will also have a profound technological, business, and social impact on society. In IoT environments, networking devices are typically resource-constrained, which enable them, draw cyber threats targets. In this context, comprehensive efforts have been made, mainly through conventional cryptographic methods, to resolve privacy and security issues in IoT environments. The specific features of IoT devices, however, make current techniques inadequate to protect the IoT networks' extensive privacy spectrum. This is, at only in part, due to resource constraints, heterogeneity, vast IoT device-generated real-time data, and that networks extremely complex conduct. The present study aims to identify network behavior and detect anomalies on the IoT network. The proposed experimental setup improves the performance of IoT based Intrusion Detection System for feature selection and dimensionality reduction. For feature selection, Chi-Square and ExtraTree classifier with SVM and dimensionality reduction methods Principal Component Analysis (PCA) and Firefly Algorithm (FA) used. The complete experiment carried on the BOT-IOT dataset which contains attack traces of the IoT network. From our study, it's found SVM+FA gives high accuracy 99.38%, Recall 99.3 %, precision 100 %, and F1-Score 99.78 % which is better than other methods used with SVM.