Convergence analysis of state-space recursive least-squares

Author

Malik, Mohammad Bilal ; Mohammad, Ejaz ; Maud, Mohammad Ali

Author_Institution

Coll. of Electr. & Mech. Eng., Nat. Univ. of Sci. & Technol., Pakistan

fYear

2004

fDate

11-13 June 2004

Firstpage

175

Lastpage

178

Abstract

State-space recursive least-squares (SSRLS) is a new addition to the family of RLS adaptive filters. Beginning with a review of SSRLS, we show that this time-varying filter converges to an LTI (linear time invariant) filter. With observation noise as the input, BIBO (bounded input, bounded output) stability of the estimator is discussed next. We carry out the convergence analysis of SSRLS and its steady-state counterpart. Our discussion includes convergence in mean, mean-square error, mean-square deviation and learning curves. This development is imperative for a complete understanding of SSRLS to aid a designer to make the best use of the filter in advanced applications and analysis.

Keywords

adaptive filters; least squares approximations; mean square error methods; numerical stability; parameter estimation; state-space methods; statistical analysis; time-varying filters; BIBO stability; LTI filter; RLS adaptive filters; bounded input bounded output stability; convergence analysis; learning curves; linear time invariant filter; mean-square deviation; mean-square error; state-space recursive least-squares; statistical analysis; time-varying filter; Convergence; Educational institutions; Filters; Mechanical engineering; Resonance light scattering; Riccati equations; Stability; State estimation; Steady-state; Technology management;

fLanguage

English

Publisher

ieee

Conference_Titel

Networking and Communication Conference, 2004. INCC 2004. International

Print_ISBN

0-7803-8325-7

Type

conf

DOI

10.1109/INCC.2004.1366600

Filename

1366600