Publication [6.11] of Tomás Oliveira e Silva

Reference

Tomás Oliveira e Silva, "On the Application of an Optimal Spline Sampling Theorem to Parametric Modeling of Nonstationary Signals," 1997 International Workshop on Sampling Theory and Applications, pp. 425-430, 1997. [SAMPTA'97]

Abstract

Nonstationary signals are often modeled by expressing each time-varying parameter as a time-invariant linear combination of a predefined set of functions. In this paper, we apply approximate reconstruction formulas, based on B-splines, to the time-varying parameters. Due to the finite support of the B-splines, we show that the computational complexity of the (least-mean-squares) estimation process is quite small, and is almost independent of the number of basis functions used. We also show, in a particular case, that the product sampling period-variance is approximately constant, i.e., one has to trade a good approximation quality for a smaller variance. Two examples illustrate these results, and one of them shows that the use of a large number of basis functions (1000) is feasible.

BibTeX entry

@InProceedings
{
  Oliveira.e.Silva-1997-1-SSTMNS,
  author = {Oliveira e Silva, Tom{\'a}s},
  title = {On the Application of an Optimal Spline Sampling Theorem to
          Parametric Modeling of Nonstationary Signals},
  booktitle = {1997 International Workshop on Sampling Theory and
              Applications},
  year = {1997},
  pages = {425--430},
  address = {Aveiro, Portugal},
  comment = {SAMPTA'97}
}