An algorithm unifying IMM and JPDA approaches
Ljudmil Bojilov, Kiril M. Alexiev, Pavlina Konstantinova · 2002
Abstract. In this paper, an attempt to unify the IMM and the JPDA approaches in a common algorithm is presented. For tracking manoeuvring target in clutter, IMM-PDA algorithm can be implemented. It solves successfully the problems of tracking single target in clutter. When multiple targets in clutter have to be tracked, the JPDA algorithm is one of the cheapest solutions in terms of computational load. If we assume manoeuvring targets in clutter, however, we will need of an algorithm with features of IMM and JPDA algorithms at the same time. We consider the more simple case for IMM algorithm – two models, one of which is embedded in the other. The main drawback arises when the association probabilities have to be calculated. To avoid combinatorial explosion due to the two state predictions for every track, we accept the next trade-off- we merge these two predictions, calculate association probabilities and the respective super innovation and finally go back to IMM case spliting the super innovation into two innovations.