Scholarlink

<p>The affine projection sign algorithm (APSA) has garnered widespread adoption in robust adaptive filtering, owing to its rapid convergence rate and low computational overhead. Nevertheless, when employing fixed parameters such as the step-size and the projection order, APSA encounters a dilemma balancing filtering accuracy and convergence speed. To remedy this contradiction, we introduce two innovative variable step-size evolving order APSAs (VSS-E-APSAs). Notably, these refined VSS-E-APSAs feature not only an adaptable projection order but also a dynamic step-size that varies with time. The dual variable approach significantly boosts the filtering capability of APSA. Furthermore, we undertake some theoretical analyses of VSS-E-APSAs, encompassing computational complexity, mean convergence analysis, and mean-square convergence analysis. Finally, several simulation studies reveal that our proposed methods outperform relevant APSAs, boasting faster convergence and reduced steady-state misalignment.</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

The enhanced evolving order based affine projection sign algorithm utilizing a variable step-size strategy

  • 期刊论文

  • 发表日期：

    2025年7月23日

查看全文
  • Ji Zhao,
  • Xia Ni,
  • Qiang Li,
  • Lingfei Zhou,
  • Hongbin Zhang

摘要

The affine projection sign algorithm (APSA) has garnered widespread adoption in robust adaptive filtering, owing to its rapid convergence rate and low computational overhead. Nevertheless, when employing fixed parameters such as the step-size and the projection order, APSA encounters a dilemma balancing filtering accuracy and convergence speed. To remedy this contradiction, we introduce two innovative variable step-size evolving order APSAs (VSS-E-APSAs). Notably, these refined VSS-E-APSAs feature not only an adaptable projection order but also a dynamic step-size that varies with time. The dual variable approach significantly boosts the filtering capability of APSA. Furthermore, we undertake some theoretical analyses of VSS-E-APSAs, encompassing computational complexity, mean convergence analysis, and mean-square convergence analysis. Finally, several simulation studies reveal that our proposed methods outperform relevant APSAs, boasting faster convergence and reduced steady-state misalignment.