"Soft assignment algorithm" in maneuvering targets tracking

Abstract

The optimal solution of the "assignment problem" in maneuvering target tracking (MTT) is how to associate measurements with tracks correctly. It is critical in MTT. The earlier "traditional assignment algorithm" has poor performance because it does not consider missing and new targets. The modern augment assignment algorithm overcomes the two shortcomings above, but at the same time, the risk of error association probability becomes higher. This paper presents a new assignment algorithm-"soft assignment". It need not consider missing and new targets while it is an optimal solution based on the following two criteria: (1) the correct association probability is high; (2) the number of associations is near to the actual ones. The two criteria are presented in the form of function and the optimal solution using "soft assignment" can also be obtained.