Intel Wi-Fi CSI Respiratory Sensing Dataset
Abstract
This dataset contains Wi-Fi Channel State Information (CSI) measurements collected for developing and validating real-time human presence detection algorithms based on respiratory pattern analysis. The dataset was collected as part of the European Horizon MultiX project (Multi-band Integrated Sensing and Communications) to advance privacy-preserving sensing technologies for intelligent personal computing devices.
Full text
METHODOLOGY (Summary) 1. Experimental Setup - Hardware: Intel 802.11ax chipsets specifications - Software: Data collection framework - Environments: Office, laboratory, home office descriptions - Equipment positioning and calibration 2. Participant Protocol - Recruitment criteria - Informed consent process - Session structure and duration - Activity instructions 3. Data Collection Procedure - Three experimental scenarios - Distance variations (20-60 cm) - Orientation variations - Environmental conditions monitoring 4. Signal Processing Methods - Phase correction algorithm (median-MAD) - Temporal phase unwrapping - Subcarrier selection criteria - Feature extraction techniques 5. Quality Control - Data validation procedures - Outlier detection and handling - Missing data treatment - Consistency checks 6. Machine Learning Pipeline - Training/validation split (5-fold cross-validation) - Feature engineering - Classifier selection (SVM, Random Forest, Logistic Regression) - Hyperparameter optimization - Performance evaluation metrics