Super-resolved parameterization of dispersive scattering mechanisms in the time-frequency plane

We implement a super-resolution algorithm that allows the full parameterization of waveguide-type dispersion mechanisms. In our algorithm, ESPRIT is used as the processing engine. Since super-resolution methods in their original form are not equipped to handle dispersive behavior, additional processing is required to parameterize the data. We use ESPRIT both in the time and frequency domain to extract the cutoff frequencies and the delays associated with each mode. A nonlinear sampling scheme is then used to come up with an improved estimate of the cutoff frequencies. Finally, a polynomial fit for the amplitude of each mode is performed to fully parameterize the data. The resulting curves can be displayed in the time-frequency plane with very high resolution.