In general, parameterization of EMP transient response can be accomplished by using the singularity expansion method SEM), where transient data are represented as a linear combination of complex exponentials. Techniques are presented to obtain the SEM parameters from measured data, particularly data with a low signal-to-noise ratio.
The measured data are smoothed by using a digital low-pass filter where the cutoff frequency is controlled by a circular serial correlation test. The filter provides a preprocessing technique which allows both the Prony and the singular value decomposition methods to be applied to the data. It also reduces the random noise contribution to the data and estimates the noise level. Results for simulated data with added white-Gaussian noise and for real measured data are obtained.
Two notable applications of the developed technique are presented. First, it is shown that the limited data record of fast digitizers can be extended to improve spectral resolution. Second, it is shown that spectral estimation can be accomplished.