PAPR reduction and digital predistortion for non-contiguous waveforms with well-localized spectrum

One important direction in advanced communication waveform studies for future wireless communications is improved spectrum localization, i.e., minimization of the power leakage outside very narrow guardbands around the useful signal band. This helps to improve the spectrum efficiency and facilitates asynchronous frequency-division multiple access with small overhead. These are central targets, e.g., in the 5G mobile system development. Various multicarrier and single-carrier waveforms with effective spectrum localization are available, including filter bank based waveforms and filtered OFDM, but their spectral characteristics have so far been investigated mostly in the digital processing domain. For practical implementation, it is necessary to study the effects of transmitter power amplifier (PA) nonlinearities on the spectrum localization of these waveforms. In this context, power amplifier linearization and peak-to-average power ratio (PAPR) reduction methods have a crucial role in the design of energy efficient and cost-effective transmitters. This paper focuses on these issues, by combining a generic low-complexity PAPR reduction method based on peak windowing with linearized PA based on digital predistortion (DPD). It is demonstrated that the combined DPD and PAPR reduction allows the transmitter to significantly improve the spectrum localization without sacrificing the inband waveform quality, while operating very close to the PA saturation level, thus achieving high power efficiency as well. The results are generally applicable to all spectrally localized waveforms.