Radar target identification techniques applied to a polarization diverse aircraft data base

The report investigates the performance of several types of multi-frequency radar systems employing a data base of measured monostatic radar signatures as descriptors for Radar Target Identification (RTI). The approach, a Monte-Carlo computer simulation, enables the evaluation of radar systems exploiting various aspects of RTI techniques. These aspects, examined by misclassification percentage curves, are: the number of interrogation frequencies, operating bandwidths, classification algorithm types, larger target aspect zones, and target descriptors called feature vectors. The data base of radar signatures was obtained at The Ohio State University ElectroScience Laboratory Compact Range facility, and consisted of scale model monostatic calibrated radar measurements from five commercial aircraft. It is shown that the fully-coherent radar feature vectors HH, VV, RR, LL, and RL perform very effectively for target identification with signal to noise ratios of O dB. It is also shown that the low-frequency sector of the data base provides good classification performance versus noise power, and that the number of frequencies within a given bandwidth is optimized by Shannon's sampling theorem.