Application of the Singular Spectrum Analysis Technique for Phase Constant Determination of Symmetric Reflecting and Nonreflecting Lines

The singular spectrum analysis (SSA) technique has been applied for the elimination of ripples in the measured phase constant ($\beta$) of transmission lines with small impedance mismatches by the line-line method. Compared with other postprocessing switching error (SW) procedure, the error correction (EC) technique, and the rolling average (RA) approach, it has been found that the SSA technique is more effective in removing the ripples of measured $\beta$ of reflecting and nonreflecting X-band and/or Ku-band waveguide straights with different lengths. As a metric for quantitative analysis, we computed the normalized root-mean-square-error (RMSE) and goodness-of-fit (GoF) values of the retrieved $\beta$ values. Better performances of the SSA technique over the SW procedure, the EC technique, and the RA approach are observed when the RMSE and GoF values are considered. Besides, the effect of artificial noise added to scattering parameters on the overall performance of the SSA technique in $\beta$ determination of symmetric reflecting and nonreflecting lines was examined. Furthermore, we investigated the roles of the window length and the number of principal components in the application of the SSA technique and then proposed a two-step procedure for $\beta$ extraction of dispersive lines considering their roles.