Multitarget Tracking in Clutter based on Generalized Data Association: Performance Evaluation of Fusion Rules

The objective of this chapter is to present and compare different fusion rules which can be used for Generalized Data Association (GDA) for multitarget tracking (MTT) in clutter. Most of tracking methods including Target Identification (ID) or attribute information are based on classical tracking algorithms such PDAF, JPDAF, MHT, IMM, etc. and either on the Bayesian estimation and prediction of target ID, or on fusion of target class belief assignments through the Dempster-Shafer Theory (DST) and Dempster’s rule of combination. The main purpose of this study is to pursue our previous works on the development of a new GDA-MTT based on Dezert-Smarandache Theory (DSmT) but compare it also with standard fusion rules (Dempster’s, Dubois & Prade’s, Yager’s) and with the new fusion rules: Proportional Conflict Redistribution rule No.5(PCR5), fusion rule based on T-Conorm and T-Norm Fuzzy Operators(TCN rule) and the Symmetric Adaptive Combination (SAC) rule. The goal is to assess the efficiency of all these different fusion rules for the applied GDA-MTT in critical, highly conflicting situation. This evaluation is based on a Monte Carlo simulation for a particular difficult maneuvering MTT problem in clutter. This work is partially supported by MONT grants I1205/02, MI-1506/05 and by Center of Excellence BIS21++ grant (FP6-2004-ACC-SSA-2).