Analysis of likelihood approximations for bearings-only measurements

This paper provides an analysis of likelihood approximations for bearings-only measurements. By application of Streit's likelihood function decomposition that provides a likelihood which is linear in target state, Gaussian mixture likelihood approximations for bearings-only measurements based on the work of Musicki and Kronhamn are derived. They only differ in the choice of weights. An analysis as well as simulations show that Musicki's approach provides better estimation results for distant targets, whereas Kronhamn's approach is advantageous for nearby targets.