Multicomponent signals interference detection exploiting HP-splines frequency parameter
Several real-world phenomena can be studied through the analysis of non-stationary signals. Application fields are biology, audio processing, seismology, air traffic control and security. Multicomponent signals are well represented in the time-frequency plane where they are mainly characterized by special curves, called ridges, which carry information about the instantaneous frequency (IF) of each signal component. However, ridges identification usually is a difficult task for signals having interfering components and requires the automatic localization of time-frequency interference regions. Unfortunately, this analysis might suffer from the presence of more than one component in the signal, reducing its accuracy in regions of the time-frequency plane where two or more components occur. As a result, correct IF estimation and components counting can be heavily compromised in the interference regions.\N\NIn this paper the authors investigate a simple, low-cost and data-driven method in order to allow the identification of the interference regions through spectrogram regression. This is done with respect to time and frequency separately, by a spline model with segments defined in a parametric space, the so-called Hyperbolic-Polynomial splines (HP-splines). On one hand, HP-splines depend on a frequency parameter that determines the exponential behavior of the basis functions. On the other hand, the frequency parameter can be estimated from the data without approximating their trend. Therefore, as the signal TF distribution changes in the interference region, the properties of HP-spline bases are expected to be different inside and outside this region, both in terms of parameters and approximation error. In particular, the frequency parameter of the HP-spline basis, which best approximates the data, should be distinctive for the interference regions. To this aim, the interference has been studied in terms of perturbation of an ideal profile and the amplification error has been expressed as a function of the frequency parameter. The latter also offers a great computational advantage, since it can be estimated by a fast and direct method involving only the data. Numerical results show that, regardless of the signal class, the proposed method correctly identifies a rectangular window delimiting specific interference regions, in agreement with the theoretical model. This makes the proposed approach usable in IF estimation methods both for an efficient preprocessing and for a direct integration into estimation procedures. The mathematical modeling is accompanied by graphical illustrations and numerical experiments.
- A linear algebra approach to HP-splines frequency parameter selection
- Accuracy and Stability of Numerical Algorithms
- An automatic and parameter-free information-based method for sparse representation in wavelet bases
- An iterative approach for spectrogram reassignment of frequency modulated multicomponent signals
- Analysis of reassignment operators used in synchrosqueezing transforms: with an application to instantaneous frequency estimation
- Computational error bounds for Laplace transform inversion based on smoothing splines
- Flexible smoothing with B-splines and penalties. With comments and a rejoinder by the authors
- High-Order Synchrosqueezing Transform for Multicomponent Signals Analysis—With an Application to Gravitational-Wave Signal
- scientific article; zbMATH DE number 1465030 (Why is no real title available?)
- Penalized hyperbolic-polynomial splines
- Regularization tools version 4.0 for matlab 7.3
- Reproduction capabilities of penalized hyperbolic-polynomial splines
- Smoothing exponential-polynomial splines for multiexponential decay data
This page was built for publication: Multicomponent signals interference detection exploiting HP-splines frequency parameter
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7027099)
