Correlated Sparse Bayesian Learning for STAP: Algorithm and Dictionary

doi:10.1109/TAES.2025.3622569

Correlated Sparse Bayesian Learning for STAP: Algorithm and Dictionary

2026-01-01

0

PRE

AI

DOI：10.1109/TAES.2025.3622569

Original

Original request for help

Save

Abstract

En 中文

Space-time adaptive processing (STAP) is a key technique for suppressing clutter. We develop a unified correlated sparse Bayesian learning (CSBL) framework to improve clutter suppression in STAP for arbitrary dictionaries by exploiting intrasample and intersample correlations. We propose an adaptive method that enables intrasample correlation imposition and faster convergence speed. Moreover, the method constructs an adaptively updated dictionary by updating correlated atoms in the initial dictionary to match implicitly estimated clutter ridge, effectively mitigating the off-grid problem while maintaining the dictionary dimension. Our method provides a feasible solution without prior knowledge for CSBL-STAP implementation and off-grid alleviation. For performance comparison with available prior knowledge, we design an advanced prior dictionary that simultaneously matches the clutter ridge and enables a denser atomic distribution in the main clutter region. Both simulations and measured data demonstrate significant performance improvements achieved through correlation utilization and validate the effectiveness of the proposed method.

Keywords:

Correlated sparse Bayesian learning (CSBL)

dictionary design

off-grid problem

space-time adaptive processing (STAP)

Correlated Sparse Bayesian Learning for STAP: Algorithm and Dictionary

Abstract

Journal

Organization