Knowledge‐aided block sparse Bayesian learning STAP for phased‐array MIMO airborne radar
Abstract The phased‐array multiple‐input multiple‐output (PA‐MIMO) airborne radar faces more severe sample shortage problem than the conventional PA radar. Hence, it suffers from severe performance degradation when it adopts the traditional space‐time adaptive processing (STAP) methods. Fortunately,...
Guardado en:
Autores principales: | , , , |
---|---|
Formato: | article |
Lenguaje: | EN |
Publicado: |
Wiley
2021
|
Materias: | |
Acceso en línea: | https://doaj.org/article/b7a3141149df48bd8f9f491160a7fef8 |
Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
Sumario: | Abstract The phased‐array multiple‐input multiple‐output (PA‐MIMO) airborne radar faces more severe sample shortage problem than the conventional PA radar. Hence, it suffers from severe performance degradation when it adopts the traditional space‐time adaptive processing (STAP) methods. Fortunately, the introduction of sparse recovery (SR) brings a novel perspective to reducing the sample requirement. However, most existing SR‐STAP methods only consider the clutter sparsity as prior information. In fact, more useful information, such as radar parameters, ground environment, airplane state, can all be obtained in advance and further utilised to improve the performance of SR‐STAP. In this study, the clutter structure, obtained based on the abovementioned prior knowledge, is utilised in the block SR‐STAP in the PA‐MIMO airborne radar. The clutter suppression performance is poor when the block SR algorithm is directly applied into the STAP because the clutter structure is destroyed by common vectorisation operation. To solve this problem, the vectorisation operation is first redesigned based on the clutter distribution characteristics and a novel dictionary is constructed. A basic block sparse Bayesian learning STAP scheme is then proposed in the PA‐MIMO radar’s application. The numerical experiments illustrate that the proposed method can achieve satisfactory performance with a few training samples and outperforms the state‐of‐the‐art SR‐STAP methods. |
---|