Performance analysis of two sequential Monte Carlo methods and posterior Cramer-Rao bounds for multi-target tracking

The multi-target tracking algorithms generally present two basic ingredients: an estimation algorithm coupled with a data association method. In the last years, the use of sequential Monte Carlo methods has grown in many application domains and in particular in target tracking. The state distribution is then estimated with a finite weighted sum of Dirac laws centered around "particles". Very recently, two new algorithms based on sequential Monte Carlo methods have been proposed independently to solve multi-target tracking. The first one solves the data association as in the joint probabilistic data association (JPDAF) spirit whereas the second uses independent probabilistic assignments. In this paper, we first compare their performance for bearings-only applications. Then, we study how the posterior Cramer-Rao bound, giving a lower bound on the estimation error covariance, can be obtained for multiple targets. Three new bounds are obtained according to the data association assumptions and can be evaluated by Monte-Carlo (MC) simulations.