Smart IoT applications of multi attack detection using cluster F1MI approach.

Area of Science:

• Cybersecurity
• Machine Learning
• Internet of Things (IoT)

Background:

• IoT networks face significant security challenges from constant cyber attacks.
• Machine Learning (ML) offers scalable solutions for identifying anomalous activity in large IoT data volumes.

Purpose of the Study:

• To introduce the Smart Secured IoT Framework (SSIF) for enhancing the security of smart IoT networks.
• To utilize the Bot-IoT dataset for framework development and validation.

Main Methods:

• Data preprocessing for high-quality input.
• Cluster F1-MI feature engineering for selecting informative attributes.
• Cluster-based feature validation to remove feature correlation and improve efficiency.
• Utilizing ML classifiers including SVM, Random Forest, Gradient Boosting, XGBoost, and Neural Networks for threat classification.

Main Results:

• The Random Forest classifier achieved an accuracy exceeding 0.97 in classifying attack types.
• The framework effectively identifies anomalies, triggering email notifications and alarms.
• The SSIF demonstrated effectiveness, affordability, and ease of implementation.

Conclusions:

• The proposed SSIF is a robust and efficient solution for securing smart IoT environments against cyber threats like Bot-IoT.
• The framework's feature engineering and ML classification approach significantly improve threat detection capabilities.