AI-Driven Data Mining for Proactive Network Intrusion Detection in IoT Environments
N. Nithyalakshmi, A. Muthukumaravel · 2025
The growing use of IoT devices has also posed new and unique security threats and the need for efficient intrusion detection systems in constrained settings. It proposes a low overhead and low complexity anomaly-based intrusion detection system using DNN tailored for IoT networks. The proposed system includes the feature extraction and normalization of network traffic data using a preprocessing pipeline along with DNN-based detection deployed close to the network. When examined through the experimental metrics on the NSL-KDD dataset, the framework’s performance reaches first-rate levels: Accuracy of $98.75 \%$, precision: of $97.80 \%$, and a time to detect intrusion of 12 ms. From the previous methods used, it is evidenced that there is a tremendous enhancement exhibited in identification efficacy and throughput. The results demonstrate the utility of the proposed framework for risk prevention and low computational costs. It offers a strong solution for IoT network protection and opens up opportunities to continue developing edge-based cybersecurity solutions.