Mitigation of Artifacts in Multistatic & Passive Radar Imaging Using Microlocal Analysis

TL;DR

This work addresses crosstalk artifacts in multistatic and passive radar imaging by modeling the radar data as a sum of Fourier integral operators and applying microlocal analysis to locate and characterize artifact formation under backprojection. It identifies two complementary mitigation strategies: (i) designing the data acquisition geometry so the artifact surface lies outside the region of interest, and (ii) applying a calibrated sequence of Fourier integral operators to progressively displace artifacts away from the ROI while preserving true scatterers. Theoretical analysis of diagonal and mixed backprojections reveals that artifacts originate from the mixed term $F_1^{*}F_2$, with the artifact surface tied to ellipsoidal geometries and a receiver-dependent plane $\pi$, which guides both approaches. Numerical experiments in MATLAB validate that both methods substantially reduce crosstalk artifacts, with the geometry-based approach trading data completeness for artifact-free ROI and the displacement approach offering artifact attenuation without sacrificing data usage when feasible, also extending to beam-forming strategies and multi-emitter setups to ensure full ROI coverage in practice.

Abstract

In the analysis of many synthetic aperture radar (SAR) experiments the possibility of passive background signals being recorded simultaneously and corrupting the image is often overlooked. Our work addresses this by considering the multistatic experiment where two stationary emitters are "always on" so there is "crosstalk" between their signals. The model for the radar data is given by a Fourier integral operator, and we assume that the data cannot be separated into contributions from individual emitters. Using techniques of microlocal analysis, we show that "crosstalk" between emitters leads to artifacts in the image and we determine their locations relative to the scatterers that produced the data. To combat the harmful effects of crosstalk, we develop methods that allow us to create an image of a region of interest (ROI) that is free from such artifacts. The first method makes use of a carefully designed data acquisition geometry to localise artifacts away from a ROI, and the second is an image processing technique that displaces artifacts away from a ROI. These methods are verified via numerical implementation in MATLAB. The analysis carried out here is valuable in bistatic and multistatic radar experiments, where an unwanted, passive source is also being detected, as well as in passive imaging, where one wishes to produce a high-quality image purely from uncontrolled sources of illumination.

Figures (11)

• Figure 1: Propagation of singularities as a result of the artifact relation, $\mathcal{C}$, associated with the operator $F_1^{*}F_2$.
• Figure 2: Experimental design involving a single point scatterer at $\mathbf{x}$ that results in artifacts (red) that are underneath a ROI.
• Figure 3: Experimental design with a single point scatterer at $\mathbf{x}$ that results in artifacts (red) that are above a ROI.
• Figure 4: Illustration of the process by which an artifact is displaced from $z$ to $\tilde{z}$ by scaling along the ray connecting $\gamma(r)$ and $x$.
• Figure 5: An experimental design where artifacts on the same side of $\pi$ as $\gamma(r)$ will be associated to a value of $c<1$.

Theorems & Definitions (8)

• Definition 2.1
• Theorem 3.1
• Theorem 3.2
• proof
• Proposition 4.1
• proof
• Theorem 4.1
• Definition 4.1