Evaluation of PCA Variants for Intrusion Detection in IoT Networks

The proliferation of the Internet of Things (IoT) has led to a surge in connected devices, bringing about significant benefits as well as heightened security challenges. Intrusion Detection Systems (IDS) are critical for protecting IoT networks from malicious activities. However, the high-dimensional nature of IoT data necessitates effective dimensionality reduction techniques to enhance detection performance and computational efficiency. In this study, we systematically evaluate three Principal Component Analysis (PCA) variants-traditional PCA, Incre-mental PCA, and Randomized PCA-in combination with a K-Nearest Neighbors (KNN) classifier for IDS in IoT networks. We conduct extensive experiments on two prominent IoT datasets, IoT23 and CICIoT2023, to assess the accuracy and computational efficiency of each method. Our findings reveal that Incremental PCA and Randomized PCA offer significant improvements in computational efficiency without compromising detection accu-racy, making them viable alternatives to traditional PCA for real-time IDS applications in large-scale IoT networks.