Knowledge-based Solutions as They Apply to the General Radar Problem

This tutorial provides an introduction to the application of knowledge-based processing to the general radar problem. We interpret knowledge-based processing as the use of adaptivity and the exploitation of prior knowledge in such a way as to choose the optimum processing method in each case, and we interpret the general radar problem as the detection, classification and tracking of targets against a background of clutter and interference. As such the tutorial attempts to describe the nature of the general radar problem and the basic processing techniques that are used, and to show why knowledge-based signal processing may be advantageous, setting the scene for the subsequent tutorials covering CFAR detection, space-time adaptive processing, tracking, and emerging technologies. The fundamental concepts of matched filtering, superresolution and adaptive filtering are described, emphasizing the equivalence of time/frequency and aperture/angular domains, and introducing the concept of Space-Time Adaptive Processing. A description is given of some of the statistical clutter models in common use (Rayleigh, Ricean, Lognormal, Weibull and Compound-K), with practical examples of sea clutter and of land clutter which demonstrate that clutter is in general non-Gaussian and non-stationary, both in time and space. Two examples are given of the application of adaptive techniques to the suppression on nonhomogeneous clutter, showing that the performance of the adaptive Doppler filtering algorithm is severely compromised at clutter edges, due to incorrect estimation of the clutter covariance matrix, and how in Space-Time Adaptive Processing a nonhomogeneity detector can be used in the choice of the most appropriate STAP algorithm, forming the socalled Knowledge-Based STAP (KB-STAP) processor.