Time-varying channel estimation for SISO/MIMO-OFDM Systems Using Superimposed Training

We address the problem of time-varying channel estimation for orthogonal frequency-division multiplexing (OFDM) systems by using superimposed training. Channel estimation is composed of three steps. Firstly, we split one OFDM symbol time interval into equi-spaced time slots, wherein the channel variation can be assumed to follow a linear fashion. Then, some temporary channel estimates within each time slot is obtained by using a sequence of impulse train. Finally, an interpolation method is employed to smooth the final estimation. Unlike conventional ST schemes, the effects due to the unknown data on channel estimation are fully cancelled by introducing certain data distortion corresponding to the placement of embedded pilots. At receiver, an iterative reconstruction based symbol detection scheme is carried out to mitigate the introduced distortion (thus to enhance the BER performance). The performance of the proposed approach is shown to significantly outperform existing methods based on superimposed training.