DEMPSTER-SHAFER THEORY APPLIED TO ADMISSIBLE REGIONS

The admissible region approach is often used a bootstrap method to initialize a Bayesian state estimation scheme for too-short-arc measurements. However, there are ambiguities in how prior probabilities are assigned for states in the admissible region. Several approaches have proposed methods to assign prior probabilities, however there are inconsistencies in how the prior probabilities can be manipulated. The application of Dempster-Shafer evidential reasoning theory to the admissible region problem can avoid these ambiguities by eliminating the need to make any assumptions on the prior probabilities. Dempster-Shafer theory also enables the testing of the validity of the assumptions used to construct the admissible region. This paper introduces Dempster-Shafer theory and formulates the admissible region in terms of plausibility and belief which reduce to traditional Bayesian probability once there is sufficient information in the system.