AN ENSEMBLE LEARNING BASED INTRUSION DETECTION MODEL FOR INDUSTRIAL IOT SECURITY

Abstract

The "Industrial Internet of Things (IIoT)" has its own security issues that need specific methods for finding intrusions. A lot of new feature engineering and machine learning techniques are being used in this project, like "Isolation Forest (IF)", Pearson's Correlation Coefficient (PCC), and Random Forest (RF) classifier, to improve Intrusion Detection Systems (IDSs) in IIoT settings. Using datasets like BoT-IoT, UNSW-NB15, and NFUNSW-NB15-v2” for testing shows that the results are very accurate and the predictions are made quickly. We look at more than just the base study. We also look at ensemble methods like Voting Classifier and Stacking Classifier, which are 100% accurate. For testing in the real world, a user-authenticated Flask-based front end is also made. This study makes a big step forward in IIoT security by providing a strong attack detection model that quickly finds and stops threats, making industrial networks more resilient overall