A new channel estimation model in OFDM system based on improved pilot training sequence

Channel estimation is important to OFDM system coherent-demodulation. Though fading can be partly solved by increasing bandwidth, considering performance and complexity, the general channel estimation algorithms, as MMSE and LS algorithm, are not perfect. Based on the OFDM forward channel statistic model, this paper propose a special pilot training sequence, which, when antenna number is not large, can be used to well estimate the time varying channel. The new pilot training sequence eliminates matrix inverse operation and greatly decreases operation quantity. 16QAM and DMPSK is used in emulation and verifies the new estimated machine.