Radar Target Identification by Natural Resonances: Evaluation of SignalProcessing Algorithms

摘要

When a radar pulse impinges upon a target, the resultant scattering process canbe solved as a linear time-invariant (LTI) system problem. The system has a transfer function with poles and zeros. Previous work has shown that the poles are independent on the target's structure and geometry. This thesis evaluates the resonance estimation performance of two signal processing techniques: the Kumaresan-Tufts algorithm and the Cadzow-Solomon algorithm. Improvements are made to the Cadzow-Solomon algorithm. Both algorithms are programmed using MATLAB. Test data used to evaluate these algorithms includes synthetic and integral equation generated signals, with and without additive noise, in addition to new experimental scattering data from a thin wire, aluminum spheres and scale model aircraft. (Author)