In frequency diverse array (FDA) space–time adaptive processing (STAP) system, the spectrum of fast-moving target is non-well focused in the spatial–temporal plane, which causes a large mismatch between the actual and presumed target steering vector and dramatically degrades the performance of FDA-STAP system. In this paper, we propose a robust FDA-STAP approach based on multiple possible prior information constraints to resolve this issue and improve the performance of FDA-STAP system. In the proposed approach, multiple possible prior information, i.e., multiple large uncertainty regions, where the actual target possibly locate in spatial–temporal domains, are utilized to cover and constraint the estimated steering vector. The multiple uncertainty regions constraints make it possible to accurately estimate the fast-moving target steering vector. Moreover, to mitigate the influence of non-focused spatial–temporal spectrum of fast-moving target, a non-focused constraints is developed using the squared norm of space–time steering vector to avoid the estimated steering vector converge to the non-focused region of fast-moving target. Finally, the proposed robust FDA-STAP is converted into a non-convex quadratically constrained quadratic programming problem, and the semidefinite relaxation technique is employed to obtain the estimated target steering vector. Numerical results indicate that the proposed method has superior performance than other methods, including well-maintained main beam direction and significant performance improvement.
Read full abstract