Performance evaluation of four nonlinear filters for two radar applications


  1. Ding, Z.
  2. Balaji, B.
Corporate Authors
Defence R&D Canada - Ottawa, Ottawa ONT (CAN)
In this report, we reviewed the nonlinear filtering algorithms for radar tracking applications, and presented the radar tracking models, the state-space model and the measurement model. Four nonlinear filters were selected for a comparison study for two radar tracking applications. The four filters included the extended Kalman filter (EKF), the converted measurement EKF (CMEKF), the unscented Kalman filter (UKF) and the particle filters (PFs). We used two recorded real datasets and two simulated datasets. The first recorded dataset was collected from an air traffic control (ATC) radar experiment with several aircraft. The second recorded dataset was collected from a high frequency surface wave radar (HFSWR) trial that was characterized by a very long integration time and a limited set of manoeuvre types. Posterior Cramer-Rao low bound (PCRLB), root mean square error (RMSE), normalized estimation error squared (NEES) and normalized innovation squared (NIS) were used as measures of performance. We found that the performances of the four nonlinear filters were very close with each other. All the filters performed consistently. Their performances were all below the PCRLB. The PFs, occasionally diverged, had slightly lower RMSE for simulated HFSWR data than other three filters. Further confirmed was PFs’ much higher computational requirement. For sensitivities to the track initialization and the sampling interval, the CMEKF was found to be the most robust one.

Il y a un résumé en français ici.

Report Number
DRDC-OTTAWA-TM-2010-246 — Technical Report
Date of publication
01 Dec 2010
Number of Pages
Electronic Document(PDF)

Permanent link

Document 1 of 1

Date modified: