A Fall Detection Approach Based on ACF for WI-FI CSI Against Co-Channel Interference

Fall detection is an important research for wireless sensing to recognize human activities. WiFi CSI has emerged as an effective solution to the fall detection due to its advantages of being non-invasive, privacy-preserving. However, commercial WiFi devices are often deployed in complex electromagnetic environments, where interference caused by channel overlap among WiFi devices is especially pronounced. This interference may significantly impact the accuracy of fall detection.. In order to solve the above problems, the paper discusses the effect of cochannel interference on the actual CSI. Then the CSI data were utilized to obtain an improved signal-to-noise ratio (SNR). This approach enables us to establish a mapping relationship between the ratio of dynamic power associated with falls and noise power in the presence of co-channel interference (CCI), as well as the characteristics of human fall movements. According to the experiments, the approach was able to achieve an accuracy of $\mathbf{9 7 \%}$ for fall detection under CCI.