Global cost of assignment in the TBM framework for association of uncertain ID reports

Suppose we are given multiple independent reports on the count and identification (class, type, allegiance) of objects that are present in a certain volume of interest. The reports are uncertain (random and non-specific) and collected with the probability of detection Pd<1 and the probability of false alarm Pf>0. The problem is to estimate the actual number of objects and their true identity. Adopting the formalism of the belief function theory as interpreted by the transferable belief model (TBM), a solution is proposed which maximises the plausibility of the global assignment cost of identification reports. The proposed global cost of assignment is shown by Monte Carlo simulations to outperform its ad hoc alternatives.