基于距离平滑的多任务稀疏STAP算法  

Multi-task sparse STAP algorithm based on range-smoothness

作　　者：秦立龙[1] 翁呈祥[2] 张峻宁 QIN Lilong;WENG Chengxiang;ZHANG Junning(College of Electronic Warfare,National University of Defense Technology,Hefei 230000,China;Air Force Early Warning Academy,Wuhan 430019,China)

机构地区：[1]国防科学技术大学电子对抗学院,合肥230000 [2]空军预警学院,武汉430019

出　　处：《空天预警研究学报》2022年第1期1-4,9,共5页JOURNAL OF AIR & SPACE EARLY WARNING RESEARCH

基　　金：国家自然科学基金资助项目(61801500,62171450)。

摘　　要：针对空时自适应处理(STAP)算法在非均匀环境下处理有限独立同分布训练样本存在的问题,利用滤波器权矢量稀疏性和距离平滑性,提出了一种基于多任务学习的STAP算法,以降低所需训练样本数.该算法首先令相近待检测距离环的滤波器权矢量保持一致,并附加稀疏正则化约束;然后利用交替方向乘子法对优化问题进行求解.理论分析和仿真结果表明,该算法能够有效实现小样本条件下的杂波抑制,滤波器具有更好的输出信杂噪比(SCNR)性能.In order to solve the problem in using the space-time adaptive processing(STAP)algorithm to process finite independent and identically distributed training snapshots in the nonhomogeneous environment,this pa-per proposes a STAP algorithm based on multi-task learning by using sparsity of weight vectors and spatial smoothness of filter to reduce the number of required training snapshots.In the proposed algorithm,the filter weight vectors of the neighboring range bins of interest are kept consistent first,with sparse regularization constraints added.Then,an alternating direction method of multipliers(ADMM)is used to solve the optimization algorithm.Theoretical analysis and simulation results demonstrate that the proposed algorithm can effectively realize clutter suppress in the condition of small samples,and that the filter has better output signal-to-clutter plus noise ratio(SCNR)performance.

关 键 词：交替方向乘子法 多任务学习 空时自适应处理 稀疏表示 

分 类 号：TN957.51[电子电信—信号与信息处理]