Advanced Feature Extraction and Selection Approach Using Deep Learning and Aquila Optimizer for IoT Intrusion...

Area of Science:

• Computer Science
• Cybersecurity
• Artificial Intelligence

Background:

• Cybersecurity is crucial for the Internet of Things (IoT), demanding robust solutions for intrusion detection and malicious identification.
• Machine learning (ML) techniques are vital for enhancing IoT cybersecurity.
• Sustainable computing for IoT requires advanced security measures.

Purpose of the Study:

• To develop new feature extraction and selection methods for an Intrusion Detection System (IDS) tailored for IoT cybersecurity.
• To leverage Swarm Intelligence (SI) algorithms for advanced feature engineering in IDS.
• To improve the accuracy and efficiency of identifying malicious activities in IoT networks.

Main Methods:

• A feature extraction mechanism was designed using Conventional Neural Networks (CNN).
• An alternative feature selection (FS) approach was implemented using the Aquila Optimizer (AQU), a Swarm Intelligence (SI) algorithm.
• The developed IDS approach was evaluated on four public datasets: CIC2017, NSL-KDD, BoT-IoT, and KDD99.

Main Results:

• The developed IDS approach demonstrated high performance across various evaluation metrics.
• Extensive comparisons confirmed the competitive performance of the proposed method against other optimization techniques.
• The integration of CNN for feature extraction and AQU for feature selection proved effective for IoT intrusion detection.

Conclusions:

• The proposed IDS, utilizing CNN and the Aquila Optimizer, offers a highly effective solution for IoT cybersecurity.
• The developed feature engineering methods significantly enhance the capability to detect intrusions and malicious activities in IoT environments.
• This research contributes a novel and efficient approach to securing the Internet of Things through advanced machine learning techniques.