question about linear prediction of the chirp

Started by Farzane Ahmadi August 30, 2011
Hi

I am trying to measure the resonances of an arbitrary cavity. I use an
ultrasonic signal x to excite the cavity and try to estimate the resonances
from the cavity output (y).

x --> [cavity] -->y

I use a chirp pulse train as input (x) to excite the cavity.
http://www.mathworks.com/help/toolbox/signal/chirp.html

Chirp has a flat spectrum so in frequency domain it has a uniform frequency
response and consequently the output (y) will show the resonances of the
cavity.

Since the cavity is arbitrary, I need an "estimate" of its resonances.
question is, can I use linear prediction to estimate the poles of the
transfer function of the cavity from (y)? Because the exciataion (x- the
chirp) is not a stationary signal and I doubt if linear prediction is still
valid in my problem.

Many thanks
Farzaneh