Hybrid IDS architecture for IoT security enhancing threat detection with CPBNN and CNN models

The rising frequency of cyberattacks, especially those targeting critical infrastructure, highlights the urgent need for robust network intrusion detection systems (IDS) specifically designed for the internet of things (IoT). Security issues in IoT networks are particularly complex due to the large number of connected devices and the emergence of new, sophisticated threats. To address these challenges, this research proposes a hybrid IDS architecture that combines machine learning (ML) and neural network (NN) approaches, leveraging cascade backpropagation neural networks (CPBNN) and convolutional neural networks (CNN) to enhance IoT security. The proposed system is designed to identify vulnerabilities in IoT systems, including distributed denial of service (DDoS) attacks, while addressing specific challenges related to scalability and the inherent complexity of IoT networks. The methodology employs a dual-layered approach: CPBNN is used to detect anomalous traffic, focusing on identifying abnormal behaviors, while CNN distinguishes between different types of traffic to determine the nature of the identified anomalies. The proposed hybrid IDS is evaluated using the KDDTest-21 dataset and assessed based on performance metrics including accuracy, precision, recall, and F1-score. Experimental results demonstrate that the hybrid IDS achieves an accuracy of 90% with the CNN model and 82% with the CPBNN model, confirming its effectiveness in detecting and mitigating IoT-specific security threats. These findings highlight the importance of integrating advanced ML techniques to safeguard IoT networks against evolving threats.