基于原子范数的非均匀阵列到达角估计  

Non-uniform array DOA estimation based on atomic norms

作　　者：龙伟军 徐艺卓 张玉禄 赵清华 LONG Weijun;XU Yizhuo;ZHANG Yulu;ZHAO Qinghua(School of Electronics and Information Engineering,Nanjing University of Information Science and Technology,Nanjing 210044,China)

机构地区：[1]南京信息工程大学电子与信息工程学院,江苏南京210044

出　　处：《现代电子技术》2024年第17期10-18,共9页Modern Electronics Technique

基　　金：国家自然科学基金资助项目(62071440)。

摘　　要：在雷达系统中,天线阵列至关重要。非均匀阵列能减少天线阵元使用,带来更高的自由度,但由于其阵元位置不规则排布在到达角(DOA)估计、空间频谱计算等方面存在难题,难以满足雷达目标跟踪、声源估计等场景中高精度和快速DOA估计的需求。针对该问题,文中在ℓ_(0)原子范数的基础上提出基于Circle混沌映射的并行无网格DOA估计算法。该算法通过并行处理目标函数的同时进行群体寻优,引入自适应策略进行模型精英学习和状态评估,随后增加变异算子实现全局最优解寻找,从而实现了非均匀阵列信号快速DOA估计。仿真验证表明,相对于常见DOA估计算法,该算法可以获得较小的均方根误差(RMSE),同时实现更精确的角度估计和更少的计算复杂度。Antenna arrays are crucial in radar systems.Non-uniform arrays can reduce the use of antenna elements and bring higher degrees of freedom.Due to the irregular arrangement of array elements,non-uniform arrays have difficulties in direc-tion of arrival(DOA)and spatial spectrum calculation,which are difficult to meet the needs of high-precision and fast DOA estima-tion in radar target tracking and acoustic source estimation.In view of the above,a parallel gridless DOA estimation algorithm based on Circle chaotic map is proposed on the basis of atomic normℓ_(0).This algorithm performs group optimizing while processing the objective function problem in parallel.An adaptive strategy is introduced for model elite learning and state evaluation.And then,mutation operators are added to find out the global optimal solution,so as to realize fast DOA estimation of non-uniform array signals.Simulation experiments show that the proposed algorithm can obtain a smaller root mean square error(RMSE),and achieve more accurate angle estimation and less computational complexity in comparison with the existing common DOA estimation algorithms.

关 键 词：DOA估计 ℓ_(0)原子范数 非均匀阵列 Circle混沌映射 自适应策略 变异算子 

分 类 号：TN911.7-34[电子电信—通信与信息系统]