Scholarlink

<p>Accurate vehicle localization is essential for Internet of Vehicles (IoV) applications, especially in dense urban environments where roadside sensing units (RSUs) are required. Direction-of-arrival (DOA) estimation plays a critical role in RSU-assisted localization, and sparse Bayesian learning (SBL)–based methods are widely used due to their robustness under low SNR and limited snapshot conditions. However, existing SBL methods struggle to simultaneously achieve high estimation accuracy and fast convergence. To address these issues, this paper proposes a three-stage off-grid SBL algorithm (TSOGJSBL) for high-precision DOA estimation. In the first stage, the algorithm accelerates convergence by applying adaptive grid densification with a first-order Taylor model, focusing computational resources on potential source regions. The second stage performs off-grid compensation, while the third stage uses local search to refine DOA estimates, improving accuracy by projecting them onto the continuous angle domain. Simulation results demonstrate that the proposed algorithm achieves superior estimation accuracy and computational efficiency compared with existing methods. Moreover, the improved DOA performance enables decimeter-level vehicle localization under low SNR and limited snapshot conditions, highlighting its potential for infrastructure-assisted IoV applications.</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

Three-Stage Off-Grid Sparse Bayesian Learning for High-Resolution DOA Estimation with an Application to IoV Localization

  • 期刊论文

  • 发表日期：

    2026年7月1日

查看全文
  • Jiaqi Zhang,
  • Licui Zhang

摘要

Accurate vehicle localization is essential for Internet of Vehicles (IoV) applications, especially in dense urban environments where roadside sensing units (RSUs) are required. Direction-of-arrival (DOA) estimation plays a critical role in RSU-assisted localization, and sparse Bayesian learning (SBL)–based methods are widely used due to their robustness under low SNR and limited snapshot conditions. However, existing SBL methods struggle to simultaneously achieve high estimation accuracy and fast convergence. To address these issues, this paper proposes a three-stage off-grid SBL algorithm (TSOGJSBL) for high-precision DOA estimation. In the first stage, the algorithm accelerates convergence by applying adaptive grid densification with a first-order Taylor model, focusing computational resources on potential source regions. The second stage performs off-grid compensation, while the third stage uses local search to refine DOA estimates, improving accuracy by projecting them onto the continuous angle domain. Simulation results demonstrate that the proposed algorithm achieves superior estimation accuracy and computational efficiency compared with existing methods. Moreover, the improved DOA performance enables decimeter-level vehicle localization under low SNR and limited snapshot conditions, highlighting its potential for infrastructure-assisted IoV applications.