A first-order statistical method for time-variant MIMO channel estimation

Abstract

We propose a superimposed periodic pilot scheme for time-variant channel estimation of a multiinput and multioutput (MIMO) system. As a premise, the time-varying channel is assumed to follow a complex exponential basis expansion model. Estimation of such channel can be achieved only with a simple first-order statistics, and especially, mean-value uncertainty of additional noise is ignorable as is different from traditional method. Also, there is no loss of information rate except for a controllable increase in transmission power. Further, the problem of high peak-to-average symbol power ratio caused by superimposed pilot is also addressed. A detailed analysis is given. At last, numerical simulations demonstrate the effectiveness of this proposed method.