Numerically Stable and Efficient Implementation of a Continuous-Discrete Multiple-Model Estimator
Mirosław Sankowski, Wojciech Buda · International Journal of Electronics and Telecommunications · 2013
This paper deals with the problem of implementing adaptive radar tracking filters based on continuous-time models of target motion and on discrete-time models of measurement process.The particular difficulties addressed include: nonlinear and non-stationary target movement models with uncertain parameters, and low data rate due to a rotating radar antenna.The proposed tracking filter relies basically on the continuous-discrete variant of the extended Kalman filter (EKF), the probabilistic data association (PDA) technique and the interacting multiplemodel (IMM) state estimation scheme.Numerical properties of the algorithm are discussed and a software implementation is developed using the open-source BLAS library.Several design concepts are combined to assure numerical stability, convergence and efficiency of the estimator.