Sign-Regressor Wilcoxon and Sign-Sign Wilcoxon

Author

Sahoo, Upendra Kumar ; Panda, Ganapati ; Mulgrew, Bernard

Author_Institution

Electron. & Commun. Eng. Dept., Nat. Inst. of Technol., Rourkela, India

fYear

2010

fDate

16-17 Oct. 2010

Firstpage

35

Lastpage

39

Abstract

It is known that sign sign LMS and sign regressor LMS are faster than LMS. Inspiring from this idea we have proposed sign regressor Wilcoxon and sign-sign wilcoxon which are robust against the outlier present in the desired data and also faster than Wilcoxon and sign Wilcoxon norm. It had applied to varities of linear and nonlinear system identification problems with Gaussian noise and impulse noise present in the desired. The simulation results are compared among Wilcoxon,sign Wilcoxon and proposed sign-sign Wilcoxon and sign-regressor Wilcoxon. From simulation results it has proved that the proposed techniques are robust against outlier in the desired data and convergence speed are faster compared to other two norms.

Keywords

Gaussian noise; impulse noise; regression analysis; Gaussian noise; impulse noise; linear system identification problem; nonlinear system identification problem; sign regressor LMS; sign sign LMS; sign-regressor Wilcoxon; sign-sign Wilcoxon; Adaptive systems; Convergence; Equations; Least squares approximation; Noise; Robustness; System identification; Sign Wilcoxon; Sign-regressor Wilcoxon; Wilcoxon; sign-sign Wilcoxon;

fLanguage

English

Publisher

ieee

Conference_Titel

Advances in Recent Technologies in Communication and Computing (ARTCom), 2010 International Conference on

Conference_Location

Kottayam

Print_ISBN

978-1-4244-8093-7

Electronic_ISBN

978-0-7695-4201-0

Type

conf

DOI

10.1109/ARTCom.2010.37

Filename

5656867