Optimal superimposed training for estimation of OFDM channels

We design an optimal superimposed training (SIT) scheme for orthogonal frequency division multiplexing (OFDM) systems. Linear minimum mean square error (MMSE) and least square (LS) channel estimator are developed. The proposed optimal SIT consists of two parts, optimal training sequences deduced with respect to the channel estimate's mean square error (MSE) and optimal power allocation between training and data derived by minimizing the averaged channel capacity lower bound. The relationship between bit error ratio (BER) and averaged capacity bound is investigated under the optimal SIT. Simulation results validate our optimum design.