DFT‐based channel estimation refinement by clustering in FBMC‐OQAM system

Abstract

The conventional DFT-based channel estimation method can improve the performance of the least square (LS) or minimum mean square error (MMSE) channel estimation by removing the noise outside the maximum channel delay while the noise inside channel delay still remain. Here, a novel DFT-based pilot-aided channel estimation method for filter bank multicarrier with offset quadrature amplitude modulation (FBMC-OQAM) is proposed. Using clustering and discriminant analysis, this method can separate automatically the noise inside channel delay from channel impulse response (CIR) coefficient of the conventional DFT-based channel estimation. Theoretical analysis and simulation results show that the accuracy of channel estimation can be improved remarkably. Compared with conventional DFT-based channel estimation method and DFT-based channel estimation method with threshold, the proposed DFT-based channel estimation method with clustering has a better bit error ratio (BER) performance in additive white Gaussian noise (AWGN) channel and flat fading channel.