Performance Appraisal of Estimation Algorithms and Application of Estimation Algorithms to Target Tracking

This dissertation consists of two parts. The first part deals with theperformance appraisal of estimation algorithms. The second partfocuses on the application of estimation algorithms to targettracking. Performance appraisal is crucial for understanding,developing and comparing various estimation algorithms. Inparticular, with the evolvement of estimation theory and theincrease of problem complexity, performance appraisal is gettingmore and more challenging for engineers to make comprehensiveconclusions. However, the existing theoretical results areinadequate for practical reference. The first part of thisdissertation is dedicated to performance measures which includelocal performance measures, global performance measures andmodel distortion measure. The second part focuses on applicationof the recursive best linear unbiased estimation (BLUE) or lineaeminimum mean square error (LMMSE) estimation to nonlinearmeasurement problem in target tracking. Kalman filter has beenthe dominant basis for dynamic state filtering for several decades.Beyond Kalman filter, a more fundamental basis for the recursivebest linear unbiased filtering has been thoroughly investigated in aseries of papers by Dr. X. Rong Li. Based on the so-called quasirecursivebest linear unbiased filtering technique, the constraintsof the Kalman filter Linear-Gaussian assumptions can be relaxedsuch that a general linear filtering technique for nonlinear systemscan be achieved. An approximate optimal BLUE filter isimplemented for nonlinear measurements in target tracking whichoutperforms the existing method significantly in terms ofaccuracy, credibility and robustness.