CSI Dataset for WiFi Based Human Detection

This dataset presents time-series Channel State Information (CSI) data collected using a Raspberry Pi 4 Model B equipped with the Broadcom BCM43455c0 WiFi chipset and the Nexmon CSI extraction toolchain. Operating on the 2.4 GHz frequency band with a 20 MHz channel bandwidth, the system passively captures raw CSI packets from a predefined access point within a controlled indoor environment. The primary objective of this dataset is to enable real-time indoor occupancy detection through passive WiFi sensing, leveraging the pervasive nature of wireless signals and cost-effective embedded hardware.The dataset includes a comprehensive range of occupancy scenarios, such as an empty room, single-person presence (both static and dynamic), and multiple-person occupancy, reflecting realistic environmental conditions. Each CSI frame contains high-resolution amplitude and phase data across 64 subcarriers. Ground truth labels specifying occupancy states (Occupied \/ Not Occupied) are provided to facilitate supervised learning and time-series analysis. Alongside raw CSI data in .pcap format, the dataset includes pre-processed .csv files, Python scripts for real-time packet capture, wavelet-based denoising using Discrete Wavelet Transform (DWT), and Doppler-based motion feature extraction to enhance signal interpretation.This publicly available dataset is a valuable resource for researchers and practitioners working in the fields of wireless sensing, signal processing, and machine learning. It enables development and benchmarking of models for occupancy detection, activity recognition, device-free localization, and smart building automation. Designed with a focus on real-time performance and privacy-preserving sensing, the dataset stands out for its accessibility, reproducibility, and relevance in advancing intelligent indoor environment applications using open-source tools and low-cost hardware platforms.