Parameter extraction for microstrip components via FDTD and superresolution techniques

The finite-difference time-domain (FDTD) method has been extensively used as a flexible, accurate way to predict the electromagnetic behavior of various guided wave and radiation structures. The most important feature of the FDTD method is that broadband frequency information can be provided in a single-pass simulation. Since the results directly produced by a FDTD run are time domain data which may contain multiple frequency and mode contents, the extraction of desired parameters from these data becomes an important issue. Superresolution algorithms such as Prony's method have been used in conjunction with FDTD simulation for parameter extraction previously. We investigate several other applications of superresolution techniques for parameter extraction. The superresolution engine used is the ESPRIT (Estimation of Signal Parameters via Rotational Invariance Technique) algorithm. It is considered to be more robust with respect to noise when compared to Prony's scheme. In the first application, we consider the problem of extracting multimode parameters of a microstrip transmission line from a single-pass FDTD simulation. The modal dispersion characteristics and the broadband modal field templates of a three-line coupled microstrip structure are extracted by applying ESPRIT to the spatial data samples. In the second application, we consider the problem of extracting the oscillation frequencies of an active microstrip antenna containing active nonlinear devices from the FDTD simulation data.