Design and experimental verification of a Fast Reduced-Dimension SAGE algorithm

The maximum likelihood (ML) estimation for multipath component (MPC) parameters requires high computational complexity, resulting in low efficiency in channel measurement campaigns. In this paper, a new method, called Fast Reduced-Dimension SAGE (FRDS) algorithm, is proposed to estimate the parameters such as excess delay, power, and angle-of-arrival/departure (AoA/AoD). By sounding channels using pseudo-noise sequences in the time domain, FRDS utilizes the extracted correlation function of the received array signals to perform iterative ML calculation, instead of using baseband signal samples. Thus the data set size and computational complexity are reduced significantly. A channel sounder has been developed and the FRDS algorithm is validated by field experiments. It is shown that FRDS can effectively improve the efficiency of channel measurement and modeling.