Critical Survey of Approaches to EWR Target Tracking: an endorsement of Kalman filters
Thomas H. Kerr · 2007
New methodologies for target tracking and for evaluating its efficacy have recently emerged, all potentially being a magic bullet. The questionable accuracy benefits, missing rigor (in some cases), and definitely large CPU-time computer loading drawbacks of the new estimation approaches are discussed as compared to a conventional Extended Kalman Filter and the novel Batch Maximum Likelihood Least Squares algorithm, as the well-known previous candidates for use in land-based Early Warning Radar (EWR) target tracking. A reminder is that the existing 30 year old Cramer-Rao lower bound evaluation methodology is already rigorous and adequate for evaluating target tracking efficacy in P d < 1 situations when confined to exo-atmospheric target tracking, as arises in EWR. We also discuss the more challenging and sensitive angle- only filter methodologies, needed to handle target tracking when enemy escort jamming denies radar range measurements and impedes target tracking unless two or more radars synchronously triangulate to thereby enable joint tracking of enemy targets. Finally, supporting technologies (some old, some new) are dis- cussed for enhancing the performance of these EKF's with only a modest increase in computational burden. Motivation for these pursuits is the quest to gain more veracity in on-line filter covariance calculation to mitigate any tendency to be overly optimistic (which could otherwise adversely affect multi-target track associations which typically utilize such EKF covariances within its initial gating stage).