A factor graph approach to signal modelling, system identification and filtering
S. Korl · Repository for Publications and Research Data (ETH Zurich) · 2005
This thesis concerns model-based signal processing.Inmodel-based sig¬ nal processing a class of signals is described by a stochastic state-space model, in general with unknown parameters.The aim of signal estima¬ tion (filtering,denoising, parameterestimation,etc.) is to determine the 'best' signal (e.g. the most probable signal) out of that class given a set