DFT-Based Channel Estimation for GFDM on Multipath Channels

Abstract

The DFT-based channel estimation is proposed to be applied for generalized frequency division multiplexing (GFDM) system. In order to implement this method, maximum channel impulse response needs to be known. By eliminating the interference and noise outside of the channel impulse response parts, improvement in the estimation can be obtained. The aid of pilot symbols is needed to operate the least square (LS) channel estimation. Once the LS channel estimation is obtained, the DFT-based algorithm is performed. The results show that by employing the DFT-based channel estimation, the performance of mean square error (MSE) improves significantly at low SNR$\sim 10$ dB compared to the LS estimation, while at the same time the SER of DFT-Based channel estimation also gives valuableimprovement.