On the Applicability of
Principal Component Inverse (PCI) to Rapidly Adaptive Suppression of Terrain Scattered
|Brian E. Freburger and Donald W. Tufts
University of Rhode Island
Kingston, RI 02881
Abstract Reduced-rank weight calculations are investigated to provide the faster adaptation required for effective suppression of Terrain Scattered Interference (TSI). Both element space Principal Component Inverse (PCI) and reduced-rank least squares for the generalized sidelobe canceller (GSLC) in beamspace are compared with each other and with their Sample Matrix Inverse (SMI) counterparts. More specifically, the case where training data is limited due to the non-stationarity of the jammer signal is considered. It is demonstrated using the Mountaintop data that in non-stationary environments such as an airborne jammer, using the reduced rank methods with training on the same data interval on which the weight vector will be applied yields better residual signal to jammer ratio than training on a nearby data interval. In fact the performance in this case is better than training on the entire pulse repetition interval using a full-rank or reduced-rank method. In addition the reduced-rank methods outperform the full-rank versions in the case where training data is limited by the availability or stationarity of the noise. Both reduced-rank methods investigated have similar performance, however, the element space PCI does not require recalculation of the weight vector for different matching vectors giving it a computational advantage.
Direct comments and questions to: firstname.lastname@example.org
© MIT Lincoln Laboratory. All rights reserved.