Fourier models for non-linear signal processing
Introduction
Non-linear signal processing includes a wide range of applications where existing linear processing tools fail to provide appropriate results. The search of a `suitable’ functional extension in terms of the input signal is of paramount importance in the design of a non-linear system (NLS). The non-linear models more widely reported are usually based on polynomial functional extensions, such as the Volterra model or the G-functional Wiener model [11], [23]. More recent functional extensions are those approached in a neural network framework, particularly the so-called radial basis functions successfully applied in modern communication receivers [17], [26]. In addition, the contributions of fuzzy logic to the non-linear field are worth mentioning [5], [6], [12], [27].
As a contribution to all these NLS models based on the mapping of the input data, this work reports the advantages of using a trigonometric functional extension of the NLS input/output relationship [7], [9]. Previous miscellaneous applications of trigonometric functionals can be found for instance in [28] where they are useful for synchronous communications, in [16] to approximate the angular distribution in a nuclear reaction, and in [20] to develop an architecture that follows, as much as possible, the guidelines stated in Kolmogorov's theorem. A recent work on NLS identification and prediction is found in [24] where the trigonometric functionals are normalised by the input PDF in order to become orthogonal.
The major contributions of this work, derived from the use of the Fourier model over other existing alternatives in the fields of non-linear processing, are three:
- •
First, it is proved that, in a minimum mean square error (MMSE) design of the weights of the Fourier model, the uniform PDF plays the same role as the white spectral density in linear processing. Furthermore, it is shown that the data covariance matrix, involved in the MMSE design, is formed by samples of the characteristic function (CF), whence a parallel is drawn between non-linear processing based on the CF and linear processing based on the autocorrelation function.
- •
Second, the constant power of trigonometric kernels together with their high bandwidth offers, in adaptive algorithms, absolute control of misadjustment, fast trace convergence, and direct matrix inversion (DMI) performance from instantaneous gradient (LMS) updates.
- •
Third, the mentioned parallelism is further exploited introducing new PDF estimation procedures, which, acting over the CF, are formulated in the same manner as classical spectral density estimators.
Section snippets
The Fourier functional extension
This section reports the basic formulation of the MMSE design of NLSs based on trigonometric functional extensions. The resulting NLS is referred to as the Fourier model, although this does not imply the orthogonality between the functionals involved in the model [22], nor the direct use of the Fourier integral to compute the corresponding coefficients. For the sake of generality, the modelling of a memoryless NLS is first presented, and then the results are generalised to the case of NLSs with
Adaptive implementation of the Fourier model
As in linear systems, the major trade-off found in the adaptive design of NLSs is the one between complexity and both convergence and misadjustment [15]. Particularly, in gradient-based adaptive algorithms it is crucial to estimate the trace of the data covariance matrix. The reason is twofold. First, the time needed to estimate the trace is a lower bound on the convergence rate of the algorithm; and secondly, the misadjustment control depends on how accurate this estimation is. This is
PDF whitening and PDF estimation
This section exploits the parallelism outlined in Section 2 between the power spectral density and the PDF functions that the Fourier model shows. Particularly, two different applications related to the PDF function are approached: the problem of PDF `whitening’ (that is, to obtain a uniformly distributed random variable (r.v.) from another one), and the derivation of PDF estimates using well-known spectral estimation methods.
Simulation results
This section includes some simulation results that cover the main points described in 2 The Fourier functional extension, 3 Adaptive implementation of the Fourier model, 4 PDF whitening and PDF estimation. It is divided into four subsections. The first two subsections compare the performance of the Fourier and Volterra models in the identification of NLSs with and without memory. This is done focusing on applications of NLSs in non-linear companders and equalisers in communications. In the
Conclusions
It has been shown that the trigonometric functional extension is a valid alternative to polynomial or Volterra functionals and, in general, to any other dependent or independent PDF functional extensions. This statement is supported by three particular features associated with the trigonometric functionals. First of all, decoupled computation of the weights is possible in the MMSE identification of NLSs with uniformly distributed input data. Secondly, the constant power property of the
Acknowledgements
This work has been supported by TIC96-0500-C10-01, TIC98-0412, TIC98-0703 and CIRIT 1998SGR-00081. The authors would like to thank Lluı́s Torres and Josep Sala-Álvarez for the time spent in reviewing the paper.
References (29)
Non-linear black-box models in system identification: mathematical foundations
Automatica
(1995)- T.M. Apostol, Mathematical Analysis, Addison-Wesley, Reading, MA, 1974, pp....
- E. Biglieri, S. Barberis, M. Catena, Analysis and compensation of non-linearities in digital transmission systems, IEEE...
Principles and Practice of Information Theory
(1987)Estimation of multivariate density
Ann. Inst. Statist. Math.
(1966)- J.L. Castro, Fuzzy logic controllers are universal approximation, IEEE Trans. Systems Man Cybernet. 25 (4) (April 1995)...
- J.L. Castro, M. Delgado, Fuzzy systems with defuzzification are universal approximations, IEEE Trans. Systems Man...
- R.M. Gray, J.W. Goodman, Fourier Transforms. An Introduction for Engineers, Kluwer Academic Publishers, Dordrecht,...
- S. Haykin, Adaptive Filter Theory, Prentice-Hall, Englewood Cliffs, NJ, 1991, pp....
- G. Karam, H. Sari, Analysis of predistortion, equalization and ISI cancellation techniques in digital radio systems...