ESTIMATION USING SAMPLED-DATA CONTAINING SEQUENTIALLY CORRELATED NOISE.
HARVARD UNIV CAMBRIDGE MASS DIV OF ENGINEERING AND APPLIED PHYSICS
Pagination or Media Count:
The paper presents improved filtering, prediction, and smoothing procedures for multi-stage linear dynamic systems when the measured quantities are linear combinations of the state variables with additive sequentially correlated noise. The augmented state procedure suggested by Kalman may lead to ill-conditioned computations in constructing the data processing filter. The design procedure described here eliminates these ill-conditioned computations and reduces the dimension of the filter required. The results include explicit relations for prediction, filtering, and smoothing procedures and the associated covariance matrices. Author
- Numerical Mathematics