UWB channel estimation based on compressed sensing using dual-kernel dictionaries

Channel estimation is a key issue for ultra-wideband (UWB) wireless communication systems. In this paper, we investigate UWB channel estimation, where the emerging compressed sensing (CS) is used to bypass the difficulty faced by Nyquist sampling. Particularly, we propose a couple of dual-kernel dictionaries, namely cross-derivative dictionary (CDD) and accessorial-derivative dictionary (ADD), to approximate the UWB signals sparsely, where both the transmitted pulse shape and its derivative are used as generating kernels. Extensive simulation results reveal that the proposed dictionaries provide much sparser approximation than the conventional multipath diversity based dictionary (MDD) and can be used to improve the channel estimation precision effectively.