Physics-Informed Wi-Fi Channel State Information Sensing for Device-Free Human Activity Recognition: A Survey

Abstract

Wi-Fichannel state information (CSI) enables non-intrusive device-free human activity recognition (DFHAR) on commodity wireless infrastructure, yet the electromagnetic sensitivity that resolves sub-millimeter motion makes trained models fragile to environment, hardware, and user change. Prior surveys treat cross-domain adaptation, physics-informed modeling, self-supervised learning, and edge deployment largely in isolation and seldom explain how measurement inconsistency, information loss, and distributional shift interact as coupled failure modes. This survey synthesizes the Wi-Fi CSI DFHAR corpus through three lenses—physics-constrained learning, information-theoretic compression, and cross-domain adaptation—and interprets reported results under matched protocol attributes. Physics-informed residuals in the reviewed literature primarily act as training-time regularizers: they help mainly under scarce labels and a residual commensurate with the CSI grid, and become redundant or harmful when labels already fit the task or the operator is not identifiable from commodity measurements. A structured extraction schema and a dry-run/pilot/sustain commissioning playbook are offered as reporting templates for auditable wireless sensing deployments.