A comparison of 5G candidate waveforms subject to phase noise impairment at mm-wave frequencies

Sammanfattning: Frequencies above 6 GHz are being considered by mobile communication industry for the deployment of future 5G networks. Large channel bandwidths above 6 GHz are likely to be used to serve diverse use cases and meet extreme user requirements. However in the higher carrier frequencies, especially the millimeter-wave frequencies (above 30 GHz), there can be severe degradations in the transmitted and received signals due to Radio-Frequency (RF) impairments such as phase noise introduced by the local oscillators. 5G radio interface, operating higher carrier frequencies, has to be robust against phase noise. In this thesis, the effect of phase noise has been investigated for three different multi-carrier waveforms, namely Orthogonal Frequency Division Multiplexing (OFDM), Offset QAM Filter-Bank Multi-Carrier (OQAM-FBMC) and QAM Filter-Bank Multi-Carrier (QAM-FBMC). These waveforms have been considered as potential candidates for 5G radio interface. We develop analytical tools to evaluate the performance of these three waveforms subject to phase noise in terms of Signal-to-Interference Ratio (SIR). The derived tools can be used to obtain SIR under any phase noise model with a known phase noise spectral density. We have used a specific phase noise model (the mmMagic phase noise model) in our waveform evaluations and comparisons at three different carrier frequencies: 6 GHz, 28 GHz, and 82 GHz. The theoretical results are further verified using Monte Carlo evaluations for SIR and SymbolError Rate (SER). The evaluation results reveal that OFDM is relatively more robust to phase noise than OQAM-FBMC and QAM-FBMC. It has been also observed that the choice of the overlapping factor in FBMC based waveforms can play a role in the performance. For the given phase noise model, OFDM has been observed to be robust to phase noise even at very high frequency (up to 82 GHz), which makes it a strong candidate for 5G radio interface.