No abstract

No abstract

Radar tracking of a ballistic target is a topic presenting very complex issues especially when the characteristics of the objects\ud flying into the Earth’s atmosphere are poorly known. As the target identification is required to obtain good performance, a technique\ud frequently employed in the practical applications is the multiple model filter. This study, thought the increasing interest for this class of\ud estimators, aims to identify the key parameters of the problem by making recognisable their effects on the solution. Thus – along with\ud the ideal operative conditions – the relative geometry target-sensor, the presence of missing plots, the measurement uncertainties and\ud the credibility of filter predictions are investigated. In order to have a strict benchmark of evaluation the multiple model approach\ud performance is compared with the optimal estimate given by posterior Cramér–Rao lower bounds

No abstract

Radar tracking of a projectile flying in the Earth's atmosphere is a very complex issue to cope with, due to the need of (suboptimal) nonlinear filtering techniques. Almost all cases found in literature assume that the target trajectory is observable from the firing point to the impact point on the ground, namely the trajectory observation gets under way from the first available measurement. The radar track initiation time is actually a stochastic quantity that has to be treated by means of a statistical procedure. In this paper a preliminary analysis of the effect of a more realistic filter initialization is proposed. ©2010 IEEE