Attack detection in industrial IoT using novel ensemble techniques
Kushagra Keserwani, Apoorva Aggarwal, Anamika Chauhan · 2023
The Internet of Things (IoT) is a rapidly growing industry that is expected to connect 22 billion smart devices by 2025 in a market worth $1567 billion. However, Securing IoT devices is challenging due to their integrated and resource-constrained nature. The traditional security measures are relatively ineffective against these continuously changing attacks. Deep learning (DL) and machine learning (ML) offer potential solutions to incorporate intelligence into systems, enabling improved detection of security breaches and attacks. These techniques can also help us suggest suitable mitigation techniques for specific attack types. The paper focuses on suggesting a solution for the detection and classification of attacks on IoT networks using novel ensemble techniques, CatBoost, and XGBoost and training them on realistic Edge-IIoTset Dataset. The performance of the solution is tested by comparing it with traditional ensemble models trained on the same data. The evaluation criteria employed are the accuracy, recall, precision and F1 score.