An optimal federated learning-based intrusion detection for IoT environment.

Area of Science:

• Cybersecurity
• Artificial Intelligence
• Internet of Things (IoT)

Background:

• Traditional machine learning-based intrusion detection systems (IDS) struggle with evolving attack patterns in IoT networks due to reliance on specific training data.
• The complexity of analyzing diverse attack vectors challenges the effectiveness of conventional IDS.
• Federated Learning (FL) offers a decentralized approach to model training, enabling adaptation to new threats without centralizing sensitive data.

Purpose of the Study:

• To propose a novel federated learning-based intrusion detection approach for enhanced cybersecurity in IoT environments.
• To leverage deep learning classifiers within a federated learning framework for improved attack detection.
• To utilize an optimization algorithm for effective feature selection in the proposed IDS.

Main Methods:

• Implementation of a federated learning framework to train deep learning classifiers across distributed IoT systems.
• Integration of the Chimp optimization algorithm for selecting optimal features to enhance detection accuracy.
• Validation of the proposed approach using the benchmark MQTT dataset.

Main Results:

• The federated learning-based IDS achieved a maximum detection accuracy of 95.59%.
• The proposed method demonstrated superior performance compared to traditional machine learning algorithms in detecting various attacks.
• Effective feature selection using the Chimp optimization algorithm contributed to improved intrusion detection capabilities.

Conclusions:

• Federated learning provides a robust framework for developing adaptive and accurate intrusion detection systems in IoT networks.
• The proposed FL-based IDS effectively addresses the challenge of detecting novel and diverse cyber threats.
• This research highlights the potential of combining federated learning, deep learning, and optimization algorithms for advanced IoT security.