Hybrid sparse/diffuse UWB channel estimation

The problem of channel estimation in ultra wideband (UWB) systems is investigated. While much prior work modeled the UWB channel as sparse, recent propagation studies suggest that, due to certain forms of scattering, a hybrid model, composed of both sparse and so-called diffuse components, is more appropriate. This new model is exploited to develop new channel estimation strategies based on the expectation-maximization algorithm. A theorem is provided which states the convergence of the iterative estimator for the sparsity level. The newly developed algorithms are compared to other schemes including recently popularized sparse approximation methods. Simulation results show that the new methods outperform the prior techniques which do not exploit the intrinsic nature of UWB propagation.