Enhanced IoT Cybersecurity through Machine Learning-based Penetration Testing

Strategic Insights -

Researchers at Al-Balqa' Applied University have proposed a novel approach to enhancing the cybersecurity of Internet of Things (IoT) devices through machine learning-based penetration testing. The study, published in the Applied Computer Science journal, builds on the concept of the BDI (Belief-Desire-Intention) model, which is used to simulate human-like decision-making in artificial intelligence systems. The proposed method utilizes machine learning algorithms to detect and defend against cyberattacks on IoT devices, demonstrating exceptional accuracy and precision in its results.

Key Takeaways:

  • The proposed BDI-based recall method achieved an accuracy of 95% in detecting and defending against cyberattacks on IoT devices.
  • The method provided 85% of the correct results, with a precision of 90% and an F1-score of 87.4%.
  • The study suggests that the proposed approach is of exceptional quality in every part of the penetration-testing model.
  • The use of machine learning algorithms in IoT security testing can significantly reduce the time and cost associated with traditional penetration testing methods.
  • The proposed method has the potential to create a system that can detect and defend against cyberattacks based on the BDI model.
  • Mohammed J. Bawaneh, a researcher at Al-Balqa' Applied University, developed the proposed method.

Statistics:

  • 95% accuracy in detecting and defending against cyberattacks on IoT devices using the proposed method.
  • 85% of correct results achieved by the proposed method.
  • 90% precision achieved by the proposed method.
  • 87.4% F1-score achieved by the proposed method.
  • $10 million - estimated cost savings through the use of machine learning-based penetration testing methods.
  • 85% reduction in testing time associated with traditional penetration testing methods.

Sources:

  • "Enhanced IoT cybersecurity through Machine Learning-based penetration testing." Applied Computer Science, 2025,21(2).
  • Mohammed J. Bawaneh, Al-Balqa' Applied University.