Impact of Environmental Conditions on WiFi Sensing

WiFi sensing has emerged as a promising alternative to traditional vision-based systems, offering advantages such as privacy preservation and robustness in dark and occluded conditions. This study investigates the impact of environmental conditions on WiFi sensing performance, exploring both indoor and outdoor settings. By collecting and analyzing Channel State Information (CSI) data across different environments, the research examines how factors such as room size, inclusion of furniture, and the presence of obstacles affect sensing performance. A machine learning-based approach is utilized, incorporating a Faster R-CNN architecture with a modality translation network to transform CSI data into an image-like format for human detection. The study evaluates model performance using Average Precision (AP) metrics and explores variations in detection performance across different environmental configurations. Results indicate that indoor environments generally yield higher performance than outdoor settings, with small, enclosed spaces enhancing signal multipath effects and improving detection. However, disruptive materials and objects, such as whiteboards, negatively impact performance by distorting signal propagation. The findings highlight the sensitivity of WiFi sensing to environmental changes and underscore the importance of exploring the corresponding impact that various environments can have to ensure reliable performance. In particular, some findings suggest that walls may act as a filtering mechanism, allowing strong signals to pass through while reducing disruptive interference, potentially improving detection performance in certain scenarios. Future work should prioritize expanding environmental diversity in both indoor and outdoor settings, examining the effects of various physical configurations and disruptive materials on WiFi sensing performance. Furthermore, exploring how the positioning of transceivers impacts signal propagation and accuracy across various scenarios could further boost WiFi sensing deployment.