Recursive Versus Batch Processing Algorithms For Bearings-Only Tracking

At long ranges, bearings-only tracking becomes increasingly difficult because the bearing rate approaches zero.Under such conditions,system observability is severely degraded and linearization errors take on added significance. To minimize their effects, iterative batch processing estimation algorithms have been proposed, even though they are computationally cumbersome. Recent studies have shown, however, that recursive algorithms formulated in modified polar coordinates possess substantially better performance characteristics than their traditional Cartesian counterparts.Consequently,this paper addresses the question of whether iterative batch type schemes are actually necessary; specifically, do they provide sufficient performance gains to justify the additional computational burden. Recursive and iterative batch algorithms are developed utilizing modified polar coordinates. Analysis reveals that the primary difference between these two estimation schemes is their linearization procedure.The effects of linearization errors on performance is depicted via simulation experiments. Optimal system performance (i.e., the Cramer-Rao bound) is also determined by employing an "idealized" filter, which subsequently serves as a basis for comparison and evaluation.