Table of contents
Definition of Signal Transformation
The signal transform is applied to the time series of CSI
时频分析
Note that the signal transform output represents the frequency of the CSI variation pattern, not the carrier frequency
Fast Fourier Transform (FFT)
FFT 广泛用于寻找不同的主频率, and can be combined with LPF to remove high-frequency noise
, and can also obtain a specific frequency target signal through a band-pass filter (BPF). For example, time series of CSI have different dominant frequencies when nearby people are stationary or moving
FFT and BPF can also be used for human motion detection and respiration estimation
Short Time Fourier Transform (STFT)
The short-time Fourier transform (STFT) divides the input into segments of equal length and computes the FFT coefficients on each segment separately
STFT 可以将CSI时域和频域上同时表示
Discrete Hilbert Transform (DHT)
Adding π 2 \frac {\pi}{2} to the negative frequency components of the FFT2p, which converts a time series of real-valued data into its analytical representation, the complex spiral sequence
DHT 可用于分析 CSI时间序列的瞬时属性
Discrete Wavelet Transform (DWT)
STFT cannot guarantee good frequency resolution and time resolution at the same time. Long windows provide good frequency resolution but poor time resolution, making it easy to identify frequency components but unable to locate the timing of frequency changes. However, a shorter window length allows detecting when the signal changes, but cannot precisely identify the frequency of the incoming signal.
小波变换为低频信号提供良好的频率分辨率，为高频信号提供良好的时间分辨率. The output of the DWT can be fed to a wavelet filter to remove noise. DWT preserves mobility information in different scenarios and is more robust than Doppler phase shift
references
WiFi Sensing with Channel State Information: A Survey