k-NN (k-Nearest Neighbors) is a non-parametric supervised classification algorithm that assigns a label to an unlabeled sample based on the majority class among its k closest training samples in a chosen feature space, typically using distance metrics such as Euclidean distance. In CSI-based Wi-Fi sensing, k-NN serves as a standard baseline classifier applied to extracted features — such as calibrated CSI amplitude or phase statistics — for tasks like gesture recognition and activity detection, where its simplicity and interpretability make it useful for benchmarking more complex models. Key variants include choices of k value, distance metric (e.g., Euclidean, Manhattan, cosine), and weighting schemes (uniform versus distance-weighted voting), all of which can meaningfully affect classification accuracy depending on the structure and dimensionality of the CSI feature space.
Source Papers
- A survey on CSI-based Wi-Fi sensing datasets and models with a focus on reproducibility ↗ — A survey on CSI-based Wi-Fi sensing datasets and models with
- Device-Free Wireless Sensing for Gesture Recognition Based on Complementary CSI Amplitude and Phase ↗ — Device-Free Wireless Sensing for Gesture Recognition Based o
- Time matters: Empirical insights into the limits and challenges of temporal generalization in CSI-based Wi-Fi sensing ↗ — Time matters: Empirical insights into the limits and challen