Map-Aided Secondary Data Selection

Abstract

Here, we present the results of an investigation on a secondary data screening approach that uses the National Land Cover Dataset along with the Digital Elevation Model to compute a feature vector for each secondary data range. By combining both knowledge sources, we created a feature vector for each range which is essentially a map of the terrain radar cross section as function of azimuth angle. We present the loss in signal-to-interference plus noise ratio, due to the use of an estimated covariance matrix versus a known covariance matrix, for two scenarios: Los Angeles and KASSPER '02. On one hand, our results reveal that map-aided training does not offer a consistent improvement in performance over selecting secondary vectors based on range from the cell-under-test (CUT). On the other hand, the results also reveal that the use of map-aided training does not degrade performance. Thus, one can use map-aided training without the fear of degrading performance while maintaining the potential of improved capability in scenarios where similarity scoring reveals differences between the feature vectors of the CUT and the secondary data ranges.