A Methodology for Predicting Improved Dipole Source Configurations From Near-Field Scan Data

Equivalent sources found from near-electric and magnetic field scans are often used to predict and solve interference problems. While there are many ways to represent the source, the user lacks the ability to determine which of the many possible source configurations is more likely to represent the “true” source, and thus accurately represent field data outside the measurement area and in the presence of typical measurement errors. A methodology is proposed for estimating which of many possible dipole source representations of near-field scan data are likely to give better estimates of fields outside the measurement scan plane when utilizing imperfect measurement data. A quality metric for determining better configurations is proposed, which utilizes the statistical variation of the global difference measure (GDM) in the predicted and measured fields. Equivalent sources are estimated when adding noise to measurement data, and prediction statistics are generated for multiple instantiations of measured noise. Results demonstrate that the better configurations minimize the average plus standard deviation of the GDM. The ability of the technique to identify robust source configurations was tested when measurements were subject to additive measurement noise, cross-field coupling, and systematic errors in probe position, and was evaluated based on its ability to predict fields at points above and to the side of the measurement plane. The method consistently identified better source configurations using both simulated and measured data.