Longtime coherent detection in GP-distributed clutter plus noise

In high-resolution maritime radars, small target detection is an intractable task. As a recognized approach, longtime coherent integration encounters three difficulties: spatially heterogeneous sea clutter, heavy computational burden from pre-Doppler whitening methods, and non-negligible noise. This paper investigates a longtime adaptive post-Doppler coherent detection method in generalized Pareto (GP) distributed clutter plus noise, which requires fewer reference cells and lower computational cost. As the first contribution, the moving target detection (MTD) method in compound-Gaussian clutter is proved to be constant false alarm rate (CFAR) to clutter power and spectrum. In GP-distributed clutter, the cell-average/cell-median MTDs (CA/CM-MTDs) are shown to be worse than the near-optimum adaptive generalized likelihood ratio test linearly data-dependent threshold detector (GLRT-LTD), only when the shape parameter of clutter and integrated pulse number are small. As the second contribution, the CA/CM-MTDs are extended to GP-distributed clutter plus thermal noise, which exploits the effective shape parameter of the mixed interference and threshold dependent on clutter-to-noise ratio (CNR) at each Doppler bin. The CA/CM-MTDs are examined to be approximately CFAR in the mixed interference. Experiments using simulated and measured datasets are conducted to show that the CA/CM-MTDs obtain better performance than existing longtime coherent integration detectors in the mixed interference.