Back to Search Start Over

Characterization of the Undesirable Global Minima of the Godard Cost Function: Case of Noncircular Symmetric Signals.

Authors :
Houcke, Sébastien
Chevreuil, Antoine
Source :
IEEE Transactions on Signal Processing; May2006, Vol. 54 Issue 5, p1917-1922, 6p, 3 Charts, 1 Graph
Publication Year :
2006

Abstract

The deconvolution of a filtered version of a zero-mean normalized independent and identically distributed (i.i.d.) signal (s<subscript>n</subscript>)<subscript>n∊Z</subscript> having a strictly negative Kurtosis γ<subscript>2</subscript> = E[∣s<subscript>n</subscript>∣<superscript>4</superscript>] - 2(E[∣s<subscript>n</subscript>∣²])² - ∣E[s<subscript>n</subscript>²∣²] is addressed. This correspondence focuses on the global minimizers of the Godard function. A well-known result states that these minimizers achieve deconvolution at least if the input signal shows the symmetry E[s²] = 0. When this constraint is relaxed, (s<subscript>n</subscript>)<subscript>n∊Z</subscript> is said to be noncircular symmetric: It is shown that the minimizers achieve deconvolution if and only if 2∣E[s<subscript>n</subscript>²]∣² < -γ<subscript>2</subscript> (s). If this condition is not met, the global minimizers are found to be finite-impulse-response filters with two taps. [ABSTRACT FROM AUTHOR]

Details

Language :
English
ISSN :
1053587X
Volume :
54
Issue :
5
Database :
Complementary Index
Journal :
IEEE Transactions on Signal Processing
Publication Type :
Academic Journal
Accession number :
20780333
Full Text :
https://doi.org/10.1109/TSP.2006.872584