ABSTRACT
This chapter shows how the robustness of one of the top five techniques for handling out-of-sequence measurements (OOSM) in terms of predictive accuracy for multi-target tracking and also shows how copulas could be used to deal with OOSM and how the use of copulas lead to a significant improvement in classification performance for multi-target tracking. The copulas modelling has found many useful applications in actuarial science, survival analysis, hydrology and finance. A motivation for copulas is that it exists as a multivariate distribution function and allows a consistent and flexible modelling of the dependence structure of dealing with OOSM. The use of copulas also deserves further investigation on a number of fronts, for example, in terms of the training parameters and the combination rules that can be employed. The experiment is carried out to rank individual OOSM methods and also to assess the impact of delayed measurements on a single delay against copula-based OOSM in terms of position error.
