Scholarlink

<p>Direction-of-Arrival (DOA) estimation is a fundamental problem in array signal processing with diverse applications in sonar, communication, radar, and autonomous driving. In environments with non-Gaussian noise, such as impulsive noise, traditional DOA estimation methods like MUSIC and ESPRIT encounter significant challenges due to the lack of finite second-order moments. To address this, we propose an enhanced Sparse Bayesian Learning (SBL) algorithm for DOA estimation in the presence of impulsive noise, utilizing fractional low-order moments (FLOM) to mitigate the noise impact. Our approach introduces a novel technique to improve computational efficiency by employing a logarithmic approximation of the <InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(\ell _0\)</EquationSource> </InlineEquation> norm, which facilitates iterative hyperparameter updates for more efficient DOA estimation. In addition, we propose an off-grid estimator that replaces costly matrix inversions with an accelerated gradient solver, improving stability near ill-conditioned systems. Simulations and experimental results demonstrate that the proposed method outperforms traditional algorithms, particularly in scenarios with limited snapshots and low signal-to-noise ratios (SNR), achieving superior estimation accuracy and computational efficiency. The results highlight robust DOA estimation under impulsive noise with improved accuracy and efficiency.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An Enhanced Off-Grid Sparse Bayesian Strategy for Direction of Arrival Estimation in Impulsive Noise

  • 期刊论文

  • 发表日期：

    2025年11月27日

查看全文
  • Wenchao He,
  • Yiran Shi,
  • Hongxi Zhao,
  • Haoran Wang

摘要

Direction-of-Arrival (DOA) estimation is a fundamental problem in array signal processing with diverse applications in sonar, communication, radar, and autonomous driving. In environments with non-Gaussian noise, such as impulsive noise, traditional DOA estimation methods like MUSIC and ESPRIT encounter significant challenges due to the lack of finite second-order moments. To address this, we propose an enhanced Sparse Bayesian Learning (SBL) algorithm for DOA estimation in the presence of impulsive noise, utilizing fractional low-order moments (FLOM) to mitigate the noise impact. Our approach introduces a novel technique to improve computational efficiency by employing a logarithmic approximation of the \(\ell _0\) norm, which facilitates iterative hyperparameter updates for more efficient DOA estimation. In addition, we propose an off-grid estimator that replaces costly matrix inversions with an accelerated gradient solver, improving stability near ill-conditioned systems. Simulations and experimental results demonstrate that the proposed method outperforms traditional algorithms, particularly in scenarios with limited snapshots and low signal-to-noise ratios (SNR), achieving superior estimation accuracy and computational efficiency. The results highlight robust DOA estimation under impulsive noise with improved accuracy and efficiency.