摘要

Radar processors may suffer from performance loss when the heavy clutter is not sufficiently suppressed in a heterogeneous environment. In order to achieve the clutter suppression and further improve the detection performance, the clutter nulling method is widely addressed in radar systems, especially for the low-rank clutter in space-time adaptive processing, where the rank of the clutter covariance matrix is smaller than the length of test data vector. For the ubiquitous clutter-plus-noise environment in practice, where it is assumed as the superposition of the white Gaussian noise and the low-rank compound-Gaussian clutter without the accurately prior information of the texture, this paper develops a clutter nulling approach, whose kernel and emphasis are to obtain the maximum-likelihood estimation of the orthonormal basis vectors of the clutter subspace. Precisely in processing, the proposed clutter nulling method is mainly derived with the application of the Lagrange multiplier method and adaptively implemented using iteration method with the training data. Finally, the results on numerical data validate the advantages of the proposed nulling approach, in comparison with the existing method.