基于流形分离的伪空间平滑解相干方法
Pseudo space smoothing decorrelation method based on manifold separation
摘要
经典空间平滑解相干方法不能直接用于不规则阵列,已有的任意阵列解相干算法对先验参数要求较高.为同时解决上述两个问题,文章直接利用经典流形分离技术构建伪导向矢量,其表示形式与均匀线阵导向矢量表示形式相同,并将原本仅适用于均匀线阵或部分均匀线阵的空间平滑-多目标分辨算法用于解相干处理.该方法即为基于流形分离的伪空间平滑(manifold-separation-based pseudo-spatial smoothing,MSPSS)算法.无偏性和有效性分析表明,该算法为渐近无偏,方差无限接近克拉默-拉奥界,性能较为突出:在负信噪比(-30 dB)下依旧保持了高角度分辨率(-3 dB主瓣宽度为 0.8°)及低绝对误差(10-15 数量级).相比以往解相干算法,该算法真正实现了任意阵列流形解相干,且无上限地打破了现实矩阵的"瑞利限",为后续衔接其他高性能波达方向估计算法提供了基础.
The classical spatially smooth decorrelation method can not be directly applicable to arbitrary arrays,and the existing arbitrary array decorrelation algorithms have high requirements of prior parameters.To address both issues simultaneously,this paper employs classical manifold separation technology to construct pseudo guide vectors in the same form as the directional vector representation of a uniform linear array(ULA),and applies the spatial smoothing-multiple signal classification(SS-MUSIC)algorithm,originally limited to uniform or partially uniform linear arrays,to achieve decorrelation processing.This approach is referred to as the manifold-separation-based pseudo spatial smoothing(MSPSS)algorithm.The unbiasedness and effectiveness analysis of the algorithm is conducted,demonstrating its asymptotic unbiasedness and variance approaching the Cramér-Rao bound.Simulation results indicate outstanding performance with high angle resolution(-3 dB main lobe width of 0.8°)and low absolute error(10-15 order of magnitude)at negative signal-to-noise ratio(SNR)(-3 dB).The algorithm truly achieves arbitrary array manifold decorrelation processing and surpasses the"Rayleigh limit"of real matrices without limitation,laying a foundation for subsequent integration with other algorithms.
年,卷(期):
2026, 45(5)
分类号:
TN911.7
机标分类号:
TN911.7
P747
TP391
在线出版日期:
2026年11月4日
基金项目:
2021YFC3100900:国家重点研发计划,