Bayesian Multipath Channel Estimation Considering Dense Multipath

In this extended abstract we present a Bayesian estimation method applicable on single-input multiple-output radio channels. In addition to specular multipath components (MPC) also the parameters of a stochastic process are estimated that comprises non-resolvable dense multipath. Exploiting the hierarchical tree structure of a Bayesian graphical model, the parametric channel estimator is able to keep the number of unwanted MPC artifacts to a minimum and is jointly determining the model order. A delay-sum beamformer is used to consider the delay differences of MPCs at the array antenna elements which makes the estimator also applicable on wideband and ultra-wideband channels. An in-depth analysis of the methods is given on the basis of synthetically generated channel impulse responses. Further, a first glimpse is presented on how well the method performs on real data.