A novel joint channel and multi-user interference statistics estimator for UWB-IR based on Gaussian mixture model

Ultra-wideband impulse radio communications have been growing rapidly over the last few years, as a promising technique for high bit-rate and multi-user transmissions over the 3-10 GHz unlicensed spectrum. In literature, multi-user interference is often approximated with a white Gaussian process and embodied in the overall noise term. However, it has been found that, in typical indoor environments, such approximation is inadequate. In this paper, we propose a new approach to characterize the interference, based on the Gaussian mixture model, which allows to derive a joint channel and multi-user interference statistics estimator, based on the iterative space-alternating generalized expectation maximization algorithm. The effectiveness of the proposed estimator is shown by means of numerical examples.