Designing RF Sensing Transmission Rates to Counter Nonuniformity in Wi-Fi Packet Propagation

The rapid proliferation of Radio Frequency (RF) devices and advancements in communication standards such as 6G and 802.11bf has facilitated the emergence of new sensing paradigms. These standards not only boost communications throughput, but also enable RF signals to be processed using advanced Machine Learning (ML) pipelines to uncover information about the physical environment. This has received tremendous attention, particularly in the Wi-Fi domain, where commodity devices have been used to monitor the environment and human activities using Channel State Information (CSI) data harvested from Wi-Fi packets. As is the case with any sensor system, Wi-Fi sensing requires sufficiently frequent CSI measurements. This has been a challenge in prior work, since the half-duplex nature of Wi-Fi and channel access protocols has meant that transmission systems can be undermined by third parties. This paper is hence motivated to investigate transmission characteristics in the context of Wi-Fi sensing to understand it’s robustness and reliability. By developing and testing a Wi-Fi based movement sensing system, we provide nontrivial insights into the non-uniformity of CSI data. We further investigate this by comprehensively analyzing the propagation characteristics of Wi-Fi packets. Through probability distribution fitting and statistical analysis, we demonstrate that the non-uniformity of CSI is worse at lower Wi-Fi packet rates. To compensate for this, we develop a regression model that facilitates the design of transmission rates in Wi-Fi sensing systems to maintain the CSI uniformity within user defined tolerances. These advancements bridge the gap towards realising robust Wi-Fi sensing deployments in practical environments.