An Efficient CNN-Based Intrusion Detection System for IoT: Use Case Towards Cybersecurity

Today's environment demands that cybersecurity be given top priority because of the increase in cyberattacks and the development of quantum computing capabilities. Traditional security measures have relied on cryptographic techniques to safeguard information systems and networks. However, with the adaptation of artificial intelligence (AI), there is an opportunity to enhance cybersecurity through learning-based methods. IoT environments, in particular, work with lightweight systems that cannot handle the large data communications typically required by traditional intrusion detection systems (IDSs) to find anomalous patterns, making it a challenging problem. A deep learning-based framework is proposed in this study with various optimizations for automatically detecting and classifying cyberattacks. These optimizations involve dimensionality reduction, hyperparameter tuning, and feature engineering. Additionally, the framework utilizes an enhanced Convolutional Neural Network (CNN) variant called Intelligent Intrusion Detection Network (IIDNet) to detect and classify attacks efficiently. Layer optimization at the architectural level is used to improve detection performance in IIDNet using a Learning-Based Intelligent Intrusion Detection (LBIID) algorithm. The experimental study conducted in this paper uses a benchmark dataset known as UNSW-NB15 and demonstrated that IIDNet achieves an outstanding accuracy of 95.47% while significantly reducing training time and excellent scalability, outperforming many existing intrusion detection models.