Channel State Information-based Device-Free stationary Human Detection with estimating respiratory frequency

In recent years, device-free passive detection becomes important and popular increasingly in a wide range of application. The physical layer information of the Wi-Fi signal can be easily measured, Channel State Information (CSI) is applied widely in many applications. And compared to Received Signal Strength (RSS), this fine-grained information can offer frequency diversity information. So we propose a system to detect static human through estimating the breathing frequency by exploring phase information of CSI. We get more robust data by fusing subcarriers and filter out environmental noise by adopting Butterworth filter and using hampel filter before and during wavelet denoising. For estimating the frequency, we introduce Fast Fourier Transformation (FFT), Estimating signal parameter via rotational invariance techniques (ESPRIT) and Multiple Signal Classification (MUSIC). The results show that detecting accuracy can achieve higher than 95% and averaged evaluating accuracy can reach 89.8% with the novel system.