Scattering models in signal processing
Abstract
Wave propagation in a layered one dimensional scattering medium is interpreted as a model for various signal processing algorithms. The computational methods for determining the propagated waves from the ones launched into the medium at boundaries and the changes in overall medium properties under cascading of scattering layers with given properties form the conceptual basis shared by many such algorithms. In filtering of signals with given state space models, the Hamiltonian formulation suggests that the estimates and the adjoint variables may be regarded as waves propagating in a medium extending in time, with local properties defined by the time varying signal model. Using this interpretation a new algorithm for limited memory Kalman filtering is derived.
- Publication:
-
Ph.D. Thesis
- Pub Date:
- 1985
- Bibcode:
- 1985PhDT........46B
- Keywords:
-
- Mathematical Models;
- Signal Processing;
- Wave Propagation;
- Wave Scattering;
- Algorithms;
- Hamiltonian Functions;
- Kalman Filters;
- Recursive Functions;
- Simulation;
- Communications and Radar