Harnessing feature pruning with optimal deep learning based DDoS cyberattack detection on IoT environment.

Area of Science:

• Cybersecurity
• Artificial Intelligence
• Internet of Things

Background:

• The proliferation of Internet of Things (IoT) devices necessitates robust cybersecurity solutions.
• Distributed Denial of Service (DDoS) attacks pose a significant threat to IoT networks and urban systems.
• Traditional machine learning struggles with the complexities of real-world IoT traffic for effective DDoS detection.

Purpose of the Study:

• To propose an advanced Artificial Intelligence (AI)-based technique for detecting IoT-based DDoS threats.
• To enhance the accuracy and efficiency of DDoS attack detection in IoT environments.
• To address the limitations of classical machine learning in identifying sophisticated cyber threats.

Main Methods:

• Feature Pruning with Optimal Deep Learning-based DDoS Attack Detection (FPODL-DDoSAD) technique.
• Data scaling using min-max scalar and feature selection via Improved Pelican Optimization Algorithm (IPOA).
• DDoS attack recognition using Sparse Denoising Autoencoder (SDAE) optimized by Fish Migration Optimizer (FMO).

Main Results:

• The FPODL-DDoSAD technique demonstrated superior performance on the BoT-IoT dataset.
• Achieved an exceptional accuracy of 99.80% in detecting IoT-based DDoS attacks.
• Outperformed existing DDoS detection methods in comparative analysis.

Conclusions:

• The proposed FPODL-DDoSAD technique offers a highly effective solution for IoT cybersecurity.
• AI-based methods, particularly deep learning with optimization, are superior to classical ML for DDoS detection.
• The study highlights the potential of advanced AI for securing the rapidly expanding IoT ecosystem.