Estimating the model parameters of deep-level transient spectroscopy data using a combined wavelet/singular value decomposition Prony method

In this article, a combined wavelet/singular value decomposition-Prony meth od to estimate the time constants associated with deep-level transient spec troscopy data is presented. A filtering scheme based on wavelet denoising i s used to provide a preprocessing technique that allows the singular value decomposition-Prony method to be applied to transient capacitance data to a ccurately estimate the associated time constants. Results for both simulate d multiple exponential model data with additive white-Gaussian noise and re al transient spectroscopy data are presented to illustrate the applicabilit y of the presented technique. Furthermore, the concept of detecting multipl e time constants is investigated and a statistical analysis is performed to address the constraints associated with the presented technique to achieve effective detection and estimation. (C) 2001 American Institute of Physics .