Doubly-selective MIMO-OFDM channel identification using superimposed training

In order to estimate doubly-selective MIMO-OFDM channel meanwhile improve bandwidth efficiency, a superimposed training (ST) method is considered. The time-varying channel is assumed to be approximated by a complex exponential basis expansion model (CE-BEM). A periodic (non-random) training sequence is arithmetically superimposed at a low power to the information sequence at the transmitter, channel parameters could be obtained without loss of bandwidth. The unknown information sequence can be interference to the ST channel estimation method, in this paper an iterative ST (IST) channel estimation method is presented to improve channel estimation performance exploiting equalized information symbols. From the result of computer simulations, we show that the proposed method can achieve good MSE and BER performance.