Skip to main content
Log in

Finite word-length effects in recursive least squares algorithms with application to adaptive equalization

Effets dus a la précision finie dans les algorithmes des moindres carrés récursifs. Application a l’égalisation adaptative

  • Stabilité Numérique de L’Algorithme des Moindres Carrés Récursifs
  • Published:
Annales des Télécommunications Aims and scope Submit manuscript

Abstract

In this paper we provide a summary of recent and new results on finite word length effects in recursive least squares adaptive algorithms. We define the numerical accuracy and numerical stability of adaptive recursive least squares algorithms and show that these two properties are related to each other, but are not equivalent. The numerical stability of adaptive recursive least squares algorithms is analyzed theoretically and the numerical accuracy with finite word length is investigated by computer simulation. It is shown that the conventional recursive least squares algorithm gives poor numerical accuracy when a short word length is used. A new form of a recursive least squares lattice algorithm is presented which is more robust to round-off errors compared to the conventional form. Optimum scaling of recursive least squares algorithms for fixedpoint implementation is also considered.

Analyse

On présente un résumé des nouveaux résultats concernant les effets de la précision finie dans les algorithmes adaptatifs des moindres carrés récursifs. Pour ces algorithmes, on définit la piécision numérique, la stabilité numérique et l’on montre que ces deux propriétés sont liées sans être équivalentes. La stabilité numérique est analysée théoriquement et la précision numérique sur des mots de longueur finie est examinée par simulation sur ordinateur. On montre que l’algorithme récursif classique des moindres carrés atteint une faible précision numérique lorsque des mots de longueur faible sont utilisés. On présente une nouvelle forme d’algorithme des moindres carrés en treillis qui se montre plus résistant aux erreurs d’arrondi que l’algorithme classique. On considère aussi une normalisation optimale des algorithmes des moindres carrés récursifs pour une implantation en virgule fixe.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Similar content being viewed by others

References

  1. Householder (A. S.). The theory of matrices in numerical analysis.Blaisdell Pub. Co., NY (1964).

    MATH  Google Scholar 

  2. Cybenko (G.). The numerical stability of the Levinson-Durbin algorithm for Toeplitz systems of equations.SIAM J. Sci. Statist. Comput., USA (sep. 1980),1.

  3. Bierman (G. J.). Factorization methods for discrete sequential estimation.Academic Press, New York (1971).

    Google Scholar 

  4. Leeet al. Recursive least squares ladder estimation algorithms.IEEE Trans. ASSP, USA (June 1981),29, pp. 627–641.

  5. Samson (C.),Reddy (V. U.). Fixed-point error analysis of the normalized ladder algorithm.IEEE Trans. ASSP, USA (Oct. 1983),31, n∘ 5, pp. 1177–1191.

    Article  MATH  Google Scholar 

  6. Proakis (J. G.). Digital communications.McGraw-Hill, New York (1983).

    Google Scholar 

  7. Godard (D.). Channel equalization using a Kaiman filter for fast data transmission.IBM J. Res. Develop., USA (May 1974), pp. 267–273.

  8. Ljung (L.). Fast algorithms for integral equations and least-squares identification problems. Ph. D. Dissertation,Linkoping Univ., Linkoping, Sweden (1983).

    Google Scholar 

  9. Hsu (F. M.). Square-root Kaiman filtering for highspeed data received over fading dispersiveHf channels.IEEE Trans. IT, USA (Sep. 1982),28, n∘ 5, pp. 753–763.

    Article  MATH  Google Scholar 

  10. Cioffi (J. M.). Fast transversal filters for communications applications. Ph. D. Dissertation,Stanford Univ. (1984).

  11. Ling (F.). Rapidly convergent adaptive filtering algorithms with applications to adaptive equalization and channel estimation. Ph. D. Thesis.Northeastern Univ., Boston, Mass. (Sep. 1984).

    Google Scholar 

  12. Ling (F.), Proakis (J. G.). A recursive modified Gram-Schmidt algorithm with applications to least squares estimation and adaptive filtering.Proc. 1984IEEE International Symposium on Circuits and Systems, Quebec, Canada (May 1984).

  13. Widrow (B.)et al. Stationary and nonstationary learning characteristics of theLms adaptive filter.Proc. IEEE, USA (Aug. 1976),64, pp. 1156–1162.

    Article  MathSciNet  Google Scholar 

  14. Markel (J. D.),Gray (A. H.) Jr. Round-off noise characteristics of a class of orthogonal polynomial structures.IEEE Trans. ASSP, USA (Oct. 1975),23, pp. 473–486.

    Article  Google Scholar 

  15. Lin (D. W.). On digital implementation of the fast Kalman algorithm.IEEE Trans. ASSP, USA (Oct. 1984),32, pp. 998–1005.

    Article  Google Scholar 

  16. Ling (F.), Manolakis (D.), Proakis (J. G.). Least squares lattice algorithms with direct updating of the reflection coefficients.Proc. 1985GRETSI Internat. Signal Proc. Applicat., Nice, France (May 20–24, 1985).

  17. Ling (F.), Proakis (J. G.). Numerical accuracy and stability. Two problems of adaptive algorithms caused by round-off error.Proc. 1984IEEE Internat. Conf. ASSP, San Diego, CA (Mar. 1984).

Download references

Author information

Authors and Affiliations

Authors

Rights and permissions

Reprints and permissions

About this article

Cite this article

Ling, F., Manolakis, D. & Proakis, J.G. Finite word-length effects in recursive least squares algorithms with application to adaptive equalization. Ann. Telecommun. 41, 328–336 (1986). https://doi.org/10.1007/BF02998639

Download citation

  • Received:

  • Accepted:

  • Issue Date:

  • DOI: https://doi.org/10.1007/BF02998639

Key words

Mots clés

Navigation