Empirical discretization errors in sparse representations for motion state estimation with multi-sensor radar systems

This work studies and explains the importance of discretization errors in the reconstruction of target motion states in radar problems using compressed sensing. The statistical attributes of these errors, such as the average, the variance and the probability density function, are estimated empirically by performing a large number of simulations. Furthermore, the performance of basis pursuit de-noising including discretization is compared with standard basis pursuit for the same compressed sensing recovery problem. By including these errors into the model, the success rate and the robustness of the recovery increase.