No abstract

Recently it was discovered that the Pareto distribution is a good model for low-grazing-angle high-resolution radar sea clutter returns. Validation of it as a model for high-grazing-angle radar clutter is provided.Introduction: Good models of radar clutter returns are an important component in the performance analysis of radar detection schemes. During the earlier days of low-resolution radar, it was found that the Gaussian assumption was a reasonable approximation [1]. As radar designers began to improve radar resolution, it was found that new models were required to reflect the observed radar returns [2]. In the context of maritime surveillance radars, clutter models have been introduced to account for the sharp deviation from the Gaussian assumption. This has resulted in the introduction of Weibull [3], lognormal [4] and K-distribution [5] clutter models. The latter has enjoyed considerable success as a high-resolution sea clutter model. This is because it has both a theoretical and a practical basis; it models observed sea clutter features through dual processes. One process models the observed fast fluctuations of clutter using a conditional Rayleigh distribution, while the second process models the underlying modulation through a Gamma distribution [6].Although the K-distribution has proven to be a good model, [7] introduced a mixture distribution based on it, in the amplitude domain, in order to account for spikiness in the horiziontally polarised sea clutter that has been observed. This spikeness has meant that the standard K-distribution has been a poor fit in the distribution's upper tail region. The mixture distribution, known as the KK-distribution, has been found to improved the fit to horiziontally polarised clutter.The difficulty with the increasing complexity of clutter models is that it has resulted in more degrees of freedom, implying that more parameter estimations are required. Also, and much more importantly, the model complexity has made the production of optimal and suboptimal detectors a mathematically challenging exercise.

No abstract

A necessary and sufficient condition is established, which can be used to compare the performance of two radar detection schemes. This means that the performance of detectors can be measured relative to a benchmark detector. In the context of coherent multilook detection in a Pareto clutter environment, the whitening matched filter (WMF) detector, which is optimal for Gaussian clutter, has been observed to perform very well as a suboptimal detector. Based upon this, variations of the WMF are constructed and analysed.

No abstract