Return
Misspecified Cramér–Rao Bounds for Near-Field Target Localization With Simplified Propagation Models
Y
高
M
L
M
F
DOI:10.1109/taes.2026.3715787.png)
Abstract
En 中文
Near-field target localization has become increasingly important in array-based sensing systems, such as radar, communication, and sonar, especially when large-aperture arrays and high carrier frequencies are employed. In the radiating near-field region, the conventional far-field plane-wave assumption breaks down, and an exact spherical-wave model is required to accurately capture wavefront curvature and amplitude variations across the array. However, to reduce computational complexity, many existing approaches adopt simplified steering models, most notably the phase-only model and the Fresnel approximation. These simplifications introduce model mismatch with respect to the true propagation and may lead to bias and performance degradation. This article provides a principled performance analysis of near-field localization under steering-model simplification using the misspecified Cramér–Rao bound (MCRB). Closed-form MCRB expressions are derived to quantify the bias and variance induced by the phase-only and Fresnel approximation models when the true data follow exact spherical-wave propagation. The analysis shows that the phase-only model yields unbiased range and angle estimates with limited performance loss, whereas the Fresnel approximation generally introduces noticeable bias at short ranges. Numerical results validate the theoretical bounds and show how model mismatch affects localization performance across different near-field operating regimes. These results, together with the theoretical analysis, provide useful guidelines on using simplified models and also highlight the necessity of using exact near-field models for high-accuracy near-field target localization.
Keywords:
Exact spherical-wave model
Fresnel approximation model
misspecified Cramér–Rao bound (MCRB)
near-field localization
phase-only model
Journal
IF:
5.7
Papers:
651
Citations:
2.4W
