Generalized frequency division multiplexing decoders under imperfect channel estimation

Generalized frequency division multiplexing (GFDM) is a waveform that satisfies the data rate and latency requirements of next-generation communication systems, and has been considered for agricultural, civilian, and home applications. Existing studies have attempted to quantify the effect of the channel estimation error on the bit error ratio (BER) of GFDM, but the scenarios under consideration are limited to the fluctuating two ray channel. In this study, we quantify the effect of the channel estimation error on the BER of GFDM, covering several decoding techniques, signal to noise ratios, and the number of propagation paths in Rayleigh fading channel. The method employs a simulation of a typical GFDM transmission and reception with imperfect channel estimations. Two decoding methods were considered: zero forcing and matched filter. This study found that the number of paths has an adverse effect on BER when the channel estimation is imperfect: the greater the number of paths, the worse the BER. The BER increases rapidly as the GFDM receiver transitions from no channel estimation error to a small channel estimation error when there is more than one propagation path. Moreover, given the channel estimation error, the matched filter decoder is found to be more reliable than the zero forcing decoder, except in the extreme case of one propagation path and no additive noise. This study provides insight into the adverse effect of channel estimation errors on the reliability of GFDM transmission and helps select a decoding technique that is robust to channel estimation errors.