Parameter Estimation for a Compound Radar Clutter Model With Trimodal Discrete Texture
Stephen Bocquet, Luke Rosenberg, Christoph H. Gierull
Abstract
The trimodal discrete (3MD) radar clutter model is a simple and accurate model for the interference distribution, which facilitates the computation of the probability of false alarm as a function of the detection threshold. Four estimation methods for the 3MD parameters are presented: two method-of-moments estimators-one based on an analytic solution to the moment equations for integer order from zero to five and the other using a nonlinear least-squares (LS) fit to 50 sample moments of fractional order; a nonlinear LS fit to the complementary cumulative distribution; and the maximum likelihood estimator (MLE). The methods were tested on both measured radar sea clutter data and simulated data and compared with the Cramér-Rao lower bound. The accuracy of all four methods is satisfactory, with the MLE the most accurate, but also the most computationally intensive. The analytic solution to the moment equations is the fastest method, two orders of magnitude faster than the MLE, but it is intractable for more than three modes. The other methods can all be used to estimate parameters for more than three modes if required.