WiStride: CSI-Based Gait Dataset from Indoor Walking Subjects

This dataset, titled WiStride, contains high-resolution Channel State Information (CSI) data captured from individuals walking in an indoor environment, designed to support research in gait recognition, human activity analysis, and wireless sensing. The data was collected using commodity WiFi devices operating under the IEEE 802.11n standard in a static indoor setting. A total of 25 participants walked back and forth across a monitored space, and their unique walking patterns were recorded using CSI extracted across 64 subcarriers from multiple antennas.Each participant completed several walking sessions to capture a wide range of motion dynamics, including variations in gait speed, stride length, and body posture. The resulting data is represented as a multi-dimensional time series containing both amplitude and phase information. The dataset is carefully organized and labeled, making it suitable for developing and evaluating both supervised and unsupervised machine learning algorithms.WiStride is particularly valuable for studying the robustness of gait recognition models in realistic, multipath-rich indoor environments. It supports diverse research applications including passive human identification, privacy-preserving biometrics, domain adaptation, multi-person tracking, and CSI signal enhancement. By providing access to real-world, reproducible CSI data, this dataset contributes to advancing the field of device-free, non-intrusive wireless sensing for biometric and activity recognition applications