Parameter Estimation in a Constrained State Space

摘要

This paper examines the problems of state estimation for a moving acoustic source confined to a well-defined, simply-connected or multiply-connected region of state space. A multiple-parameter constrained estimator that provides enhanced performance and that permits determination of the most probable solution is presented. The estimator is a batch processor that yields both dynamic and residual classifiers, the behavior of which is shown to be dependent on source-observer geometry. The proposed realization is well-suited for solution selection through hypothesis testing. Experimental results showing estimator performance are presented and solution quality is discussed. (Author)