New versions of MCDA/ICMDA algorithms applied in a nonlinear context

Over-the-horizon (OTH) radars provide a survey of wide areas, using ionospheric reflections of the electromagnetic waves. Most of the time they have to face multipath problems: state estimation has to be done with measurements involving different observation models. To tackle this measurement-to-observation-model association problem, the Monte Carlo data association (MCDA) algorithm, and a derivative one, the iterated conditional mode data association (ICMDA) have been developed. They only apply in a linear context. We propose new versions of these algorithms, well adapted to nonlinear problems. Our two algorithms are applied, through numerical simulations, to a concrete case: target tracking with the French OTH radar Nostradamus, in clutter environment