Optimal Time-Domain Training Pattern for ML Channel Estimation in MIMO OFDM System

High peak-to-average power ratio (PAPR) is a disadvantage of comb-type frequency-domain training pattern (FDTP) in MIMO OFDM system. To deal with such problem, an excellent orthogonal time-domain training pattern (TDTP), based on circularly shifting Chu sequence and satisfying that optimal condition derived by us, is proposed for maximum likelihood (ML) channel estimation. From theory analysis and simulation, we find: 1) This TDTP achieves the same optimal mean square error (MSE) performance as comb-type FDTP; 2) For bit error ratio (BER) , it performs as well as comb-type FDTP for low and medium signal-to-noise ratios (SNRs) whereas it behaves slightly better than comb-type FDTP for high SNR Thus, this pattern may be used as an alternative solution for ML channel estimation in MIMO OFDM system.