6. Spectral-domain MBE'E 6.1 Antenna Applications he MBPE (model-based parameter estimation) procedure T described above has been tested for a variety of wave-equation phenomena, of which seFeral examples are included here. Plotted in Figure 2 are the input conductance and susceptance of a monopole, as a function of the antenna size in wavelengths [26]. These results were obtained by sampling a generating model (GM) (NEC) at 0.15 intervals in L//z, shown by the open crosses. The crosses 30 ! I I I" + NEC Samples 5 2 5 -----Linear Interpolationv E FM Result w20 -0 . . z 0 0 5 I 1 0 1 2 3 LengthWavelength Figure 2a. The input conductance of a monopole antenna, versus the length in wavelengths, as obtained from a series of overlapping rational-function-fitting models. This used d = n = 3 (the solid line), based on 20 generating-model samples spaced 0.15 apart in Llwavelength, which are shown as open crosses and joined by a dashed line [26]. A comparison of the fitting-model values with the generating-model samples at other frequencies reveal's a numerical agreement of 1% or better. -'Editor's note: Part I of this article appeared in the February issue of the Magazine; Part I11 will appear in the June issue. 0 F 2-3 Length/Wavelength Figure 2b. The input susceptance of the monopole antenna, as obtained by the method used for Figure 2a. are connected by the dashed lines, to show the result of straightline interpolation; the MBPE results are the solid line.The fitting model (FM) used a d = n = 3 rational function, which requires seven data points. The first fitting model, M I , was used to develop an estimate out to the fourth data sample, as shown by the solid line. The second model, M,, was "slid" up in frequency by adding the eighth data sample and dropping the first. It was used to plot the solid line between the fourth and fifth data samples. This process was continued until model M14 was reached, which completed the fitting-model plot from data points 16 to 20. This particular procedure is not unique, as other approaches might be used, but it produces results within a few per cent of the generating model, between the data samples actually used to quantify the fitting model. Extensions of this idea include comparing the fitting models for mutual consistency in their regions of overlap, to estimate the relative uncertainty of the fitting-model result, and to determine whether additional generatingmodel samples are needed.A similar use of MBPE for a more complicated problem, representing the input impedance of a log-periodic dipole array (LPDA) due to de Beer and Baker [27], is demonstrated in Figure 3. (Note that the rational-function fitting model can be applied equally well to admittance functions, as shown in Figure 2, or to