Scholarlink

<p>This paper addresses robust parameter estimation in linear state space models with outlier-contaminated multisensor data. We develop an expectation–maximization (EM) algorithm integrated with least trimmed squares (LTS) to ensure robustness, using randomized techniques to overcome the combinatorial complexity of exact solutions. The proposed method efficiently handles both small- and large-scale sensor systems, with tailored strategies for each scenario. Monte Carlo experiments demonstrate superior robustness and computational tractability compared to non-robust alternatives. A preliminary version of this work, focusing exclusively on the limited number of sensors problem, appeared in [<CitationRef CitationID="CR17">17</CitationRef>].</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

Resilient parameter inference for multisensor state-space models: Managing outliers in large and small sensor networks

  • 期刊论文

  • 发表日期：

    2025年8月22日

查看全文
  • Jaafar AlMutawa

摘要

This paper addresses robust parameter estimation in linear state space models with outlier-contaminated multisensor data. We develop an expectation–maximization (EM) algorithm integrated with least trimmed squares (LTS) to ensure robustness, using randomized techniques to overcome the combinatorial complexity of exact solutions. The proposed method efficiently handles both small- and large-scale sensor systems, with tailored strategies for each scenario. Monte Carlo experiments demonstrate superior robustness and computational tractability compared to non-robust alternatives. A preliminary version of this work, focusing exclusively on the limited number of sensors problem, appeared in [17].