Observation Modeling of Reference--Background Residuals in Single-Snapshot FDA-MIMO-GPR

TL;DR

This paper models the residual effects caused by mismatched reference media in FDA-MIMO-GPR imaging, analyzing their impact on anomaly detection and reconstruction errors.

Contribution

It establishes a reference-state observation model for single-snapshot FDA-MIMO-GPR and analyzes the covariance structure of residuals under the distorted Born approximation.

Findings

Pronounced cross-frequency and cross-channel covariance observed under mismatched references.

Tikhonov reconstruction reveals low-dimensional pseudo-anomaly errors.

Residual modeling is crucial for reference selection and background suppression.

Abstract

Reference media are widely used in distorted-Born-approximation-based GPR imaging to represent partially known propagation effects. When the true host background differs from the chosen reference medium, the difference enters the observations and propagates into anomaly estimates. For single-snapshot FDA-MIMO-GPR, this paper establishes a reference-state observation model under the distorted Born approximation and defines that difference as the reference--background medium residual, namely, the effective residual between the reference medium and the physical background medium. Hereafter, this quantity is abbreviated as the reference--background residual. Its response is derived from the Cole--Cole dispersive mapping, the reference propagation kernels, and the FDA frequency--transmit organization. The paper then constructs its observation-domain covariance, analyzes the off-diagonal channel-block structure, and uses a standard Tikhonov estimator to show how the response transfers to reconstruction error and covariance over an anomaly candidate region. Numerical results show pronounced cross-frequency and cross-channel covariance under mismatched reference states. After Tikhonov reconstruction, these structures appear as low-dimensional, concentrated pseudo-anomaly errors. Right-hand-side coherence and inter-channel correlation arise mainly because multiple transmit--receive channels jointly observe the same residual field, while FDA space-frequency coding determines their organization in the observation and reconstruction domains. The reference--background residual should therefore be modeled explicitly in reference-state selection, background suppression, and channel-covariance analysis for single-snapshot FDA-MIMO-GPR.