Enhancing Signature-Based Intrusive Detection System (IDS) for IoT Networks Using Machine Learning Algorithm
Khandakar Rabbi Ahmed, Md Anisur Rahman Chowdhury, Md. Razaul Karim, Md. Sayham Khan, Md Afjal Hosien, Shah Tawkir Nesar, Ronny Bazan-Antequera · 2025
Internet of Things (IoT) security challenges grow stronger as the network expands and more devices join the network, even though many devices contain built-in security flaws. Traditional intrusion detection systems (IDS) based on signature detection struggle to identify new cyber threats, requiring advanced solutions. The proposed research develops a Deep Neural Network (DNN)-based IDS, which addresses IoT security through deep learning methods that enhance anomaly detection. This model harnesses network traffic data examination to detect observed and unidentified attack series because of its robust feature exploration functionality. The benchmark IoT dataset evaluation shows that the model attains superior accuracy rates for different attack types, and the model demonstrates its greatest strength in detecting wormhole attacks because it reaches a detection with an accuracy of 95.3 %, precision of 94.7 %, and recall of 95.1 %, leading to a strong F1-score of 94.9 %. The research findings confirm that deep learning technology works effectively for IoT security, yet additional issues originate from restricted computational capacity in hardware and IoT environment variability. The research enhances intrusion detection technology development by demonstrating the potential of DNN-based IDS for dealing with new security threats in IoT environments.