DocumentCode
851861
Title
A phase determination method for nonminimum phase ARMA systems by a single cumulant sample
Author
Chi, Chong-Yung ; Kung, Jung-Yuan
Author_Institution
Dept. of Electr. Eng., Nat. Tsing Hua Univ., Hsinchu, Taiwan
Volume
41
Issue
2
fYear
1993
fDate
2/1/1993 12:00:00 AM
Firstpage
981
Lastpage
985
Abstract
The authors present a new phase determination method that begins with the same preprocessing by an inverse filter as G.M. Giannakis and J.M. Mendel´s (1986) MP-AP (minimum phase-all pass) decomposition-based methods and then identifies the all-pass filter by maximizing the objective function J given by J.M. Mendel (see Proc. IEEE, vol.79, no.3, p.278, 1991), which only involves a single M th-order (M ⩾3) cumulant of the output signal of the all-pass system considered. Two phase searching algorithms, PSA1 and PSA2, are presented for finding the maximum of J . The former uses exhaustive search and the latter iterative search. PSA2 is preferred to PSA1 since it not only has a much simpler parallel processing structure but is also faster than PSA1. Some simulation results for the case of M =3 with noise-free synthetic data are provided to support the validity of the method
Keywords
all-pass filters; filtering and prediction theory; parallel algorithms; signal processing; all-pass filter; exhaustive search; identification; inverse filter; iterative search; liner time invariant systems; minimum phase-all-pass decomposition; noise-free synthetic data; nonminimum phase ARMA systems; objective function; output signal; parallel processing structure; phase determination method; phase searching algorithms; preprocessing; signal processing; simulation results; single cumulant sample; Area measurement; Biomedical measurements; Biomedical signal processing; Extraterrestrial measurements; Filters; Higher order statistics; Iterative algorithms; Parallel processing; Phase measurement; Radar signal processing; Search methods; Seismic measurements;
fLanguage
English
Journal_Title
Signal Processing, IEEE Transactions on
Publisher
ieee
ISSN
1053-587X
Type
jour
DOI
10.1109/78.193239
Filename
193239
Link To Document