基于牛顿正交匹配追踪的高特征分辨率水下弱目标参数估计方法

Newtonized Orthogonal Matching Pursuit based High-Resolution Parameter Estimation for Underwater Weak Targets

摘要:

目标特征参数估计是海洋感知的关键环节。传统水下参数估计方法多采用线性信号处理手段，具有计算复杂度低、实现简单等优点。然而，在低信噪比和临近目标特征情况下，传统方法存在估计精度受限于采样间隔、弱目标被强目标旁瓣遮掩以及特征分辨率受限等问题。格点化压缩感知方法构建字典矩阵，减轻了强目标旁瓣对弱目标遮蔽的影响。然而，由于时延、多普勒均为连续取值的参数，格点化压缩感知算法存在模型错配问题。为解决上述问题，本文提出一种基于牛顿正交匹配追踪（Newtonized orthogonal matching pursuit，NOMP）的参数估计方法。该方法将时延、多普勒视作连续取值的参数，并通过粗网格搜索获取候选目标时延、多普勒粗估计信息，然后使用牛顿迭代实现精细化修正，结合全局正交化与循环优化，有效抑制能量泄漏与互干扰，提升了目标特征的估计精度与分辨能力。数值实验结果表明，相比于现有信号处理方法，在低信噪比和多邻近特征条件下所提方法具有更高的估计精度和分辨率。

Abstract:

Target parameter estimation plays a vital role in ocean sensing tasks. Traditional underwater parameter estimation methods mainly rely on linear signal processing techniques, which have low computational complexity. However, under low signal-to-noise ratio (SNR) conditions and closely spaced targets scenarios, the estimation accuracy of these methods is restricted by sampling frequency, and the weak targets are masked by closely spaced strong ones. In addition, grid-based estimation algorithms are further constrained by grid mismatch, leading to reduced estimation accuracy. To address these issues, this paper proposes a parameter estimation method based on Newtonized orthogonal matching pursuit (NOMP). By treating the delay and Doppler as continuous-valued parameters, the initial estimates of the delay and Doppler are acquired via coarse detection, followed by refinement using Newton’s method. By incorporating global orthogonalization and cyclic refinement, the proposed method effectively alleviates energy leakage and suppresses mutual interference, thereby improving both estimation accuracy and resolution. Numerical experiments demonstrate that the proposed approach achieves higher parameter estimation accuracy and resolution of weak targets under low SNR and closely spaced multi-target scenarios, compared to the linear approach and OMP.