Principal Component Analysis (PCA) is a dimensionality reduction technique that transforms high-dimensional data into a lower-dimensional representation by identifying orthogonal axes, called principal components, that capture the maximum variance in the data. In the context of Wi-Fi and CSI sensing, PCA is applied to raw CSI measurements to suppress noise, remove static multipath components, and extract the most informative signal features before feeding them into classification or modeling pipelines, thereby improving both computational efficiency and sensing accuracy. Common variants and related approaches include Kernel PCA for nonlinear feature extraction and Incremental PCA for processing streaming or large-scale data, both of which are relevant when handling the high-volume, temporally varying CSI observations that arise in real-world ambient Wi-Fi sensing scenarios.
Dictionary term