Impact of “Better than Nyquist” pulse shaping in GFDM PHY with LTE-compatible frame structure

Abstract

The shifting towards New Radio waveforms for future cellular networks will require a smooth transition from the existing 4G physical layer (PHY), which is based on the LongTerm Evolution-Advance (LTE-A) system. This will allow the service provider to develop less cost multi-standard devices that rely on the same master clock compatibility and with the same time-frequency grid representation of the LTE-A. The main characteristic of generalized frequency-division multiplexing (GFDM) modulation is low out-of-band (OOB) emission, accomplished by using flexible pulse shape filtering in the timedomain of individual subcarriers. In this paper, first we focus on the integration of GFDM in the time-frequency grid of the LTE-A system and then we analyze the impact of “Better than Nyquist” pulse shaping filters on OOB emission and symbol error rate. We consider different “Better than Nyquist” pulse shaping filters that were originally introduced in single-carrier modulation schemes to reduce the sensitivity to symbol timing error. By using the results available in the literature, the performance achieved by “Better than Nyquist” pulse shaping filters is compared to that achieved by conventional Nyquist ones. The concept of the wavelet, based on the Meyer auxiliary function, along with pulse shaping filter will be also addressed. The results are presented for 16-QAM in the case of transmission over AWGN, time-varying frequency flat, and static frequency selective fading channels. Monte Carlo simulations are shown in order to validate the accuracy of the analytical expressions.