2. ALGORITHM FOR VIBRATION ESTIMATION
SAR receives a series of returned pulses that are reflected off
ground targets. Upon in-phase and quadrature demodulation
of these pulses (to remove the chirped carrier) and so-called
range compression (to retrieve the range position) [8], the returned
signal from a vibrating target (with reflectivity of σi)
located at position (xi(τ ), yi(τ )) is
si(τ ) ≈ σi exp
−j2ωcxi(τ )/c+jkyy¯iτ
, |τ | ≤ τ0/2, (1)
where y¯i is the average azimuth position and τ0 is the total
observation time. Because the carrier frequency, ωc, of SAR
is large (more than 10 GHz), the phase modulation caused
by small vibrations (amplitude of several millimeters) is detectable.
By taking the Taylor-series expansion of the vibration
motion and applying a second-order approximation for
a chirp signal, it concentrates a chirp signal into an impulse
in the angle-frequency plane, and the location of the impulse
is determined from the center frequency and chirp rate of the
signal. The concentration is analogous to what the discrete
Fourier transform (DFT) does to a pure sinusoid. Particularly,
the DFRFT is the same as the DFT when α = π/2. The details
on how the calculate the DFRFT is given in [12]. Here
we provide an example with a signal x[n] that has two components:
a pure 150 Hz sinusoid and a chirp signal with a center
frequency of 225 Hz and chirp rate of 400 Hz/s. The DFRFT
of x[n] is shown in Figure 1. The chirp component of x[n] is
transformed into an impulse at a certain angle α.