Bayesian sparse estimation of targets with range-Doppler grid mismatch

Abstract

In this paper, we study the problem of estimating the target scene via a signal sparse representation (SSR) scheme in the range-Doppler domain. As compared to a range-gate by range-gate SSR analysis, this bidimensional approach can take into account targets straddling two range-gates. Here, we propose a robust SSR Bayesian algorithm that considers the well known grid mismatch problem in both the range and Doppler dimensions. Our algorithm implements a bidimensional approach to a previous described algorithm. Numerical simulations are performed with synthetic and experimental data to demonstrate the benefit of estimating the grid mismatch with the proposed technique.978-1-4799-8232-5/151$31.00@2015IEEE

Velocity ambiguity mitigation of off-grid range migrating targets via Bayesian sparse recovery

“…As explained in [7,8], the range and velocity of the targets present in the radar scene does not necessarily coincide with that of the sparsifying dictionary H. Following the approach proposed in [8] two mismatch vectors (ε v , ε r ) are introduced to address the mismatch problem on the velocity and range axis resp. They parametrize the sparsifying dictionary H such that…”

“…A Bayesian framework is set up in order to estimate the parameters of interest x and (ε v , ε r ). The hierarchical Bayesian model adopted is that of [8] so it is briefly recalled in this section. However, note that the sparsifying dictionary in case of migrating targets will demand a new method for the sampling of the mismatch parameters (ε v , ε r ).…”

“…We can see that range and velocity indices are coupled in distribution (13), contrary to that of [8]. Besides, it does not belong to any familiar class of distribution so a Metropolis-Hastings move is added to the Gibbs sampler in order to sample jointly (ε v i , ε r i ).…”

“…In [6] an SSR algorithm was developed within this framework and proven to mitigate velocity ambiguities, provided that target's features belong to the analysis grid. Otherwise, as mentioned in [7,8,9,10] of grid-mismatch might arise and disrupt velocity ambiguity mitigation. In [7,8] two SSR algorithms robustified towards grid mismatch are presented in the case of narrowband radar signal where 1D and 2D Fourier analysis grids are used respectively (resp.).…”

“…Otherwise, as mentioned in [7,8,9,10] of grid-mismatch might arise and disrupt velocity ambiguity mitigation. In [7,8] two SSR algorithms robustified towards grid mismatch are presented in the case of narrowband radar signal where 1D and 2D Fourier analysis grids are used respectively (resp.). Here, the novelty lies into the new sparsifying dictionary that takes into account range migration, and the corresponding estimation scheme; the algorithm is developed within a Bayesian framework.…”