Y. Avargel and I. Cohen (Israel)
System identiﬁcation, modeling, estimation, wavelet trans
In this paper, we introduce an explicit representation of
linear time-invariant system in the discrete-time wavelet
transform (DTWT) domain. It is shown that crossband ﬁl
ters between subbands are required for perfect represen
tation of the system. These ﬁlters depend on the DTWT
parameters and on the system impulse response, and are
shown to be time-varying. An approximate representation
based on band-to-band ﬁlters without crossband ﬁlters is
employed for system identiﬁcation in the wavelet domain.
We show that for longer and stronger input signals, longer
band-to-band ﬁlters may be estimated. Experimental re
sults validate the theoretical analysis and demonstrate the
proposed system identiﬁcation approach.