Nonlinear signal processing for efficient physical layer design in 5G beyond

Download
2023-7-12
Salman, Murat Babek
Nonlinear power amplifier (PA) distortion is a significant hardware impairment that imposes limitations on communication systems, hindering their performance. This thesis presents a comprehensive analysis framework for hardware impairments and develops nonlinear signal processing techniques to mitigate the detrimental effects of nonlinear distortion on communication system performance. The evaluation is based on several criteria, including bit-error-rate (BER), achievable information rate (AIR), and out-of-band radiation. The thesis's contribution can be divided into two independent parts. The first part focuses on uplink communication, where the user terminal (UT) is equipped with a nonlinear PA. For uplink communication, the investigation encompasses the effects of frequency-selective communication channels and symbol-rate signal processing. A critical phenomenon named as 'distortion amplification', which significantly degrades detection performance, is discovered. To overcome distortion amplification, the use of a bank of nonlinear processing units for different fractional delays is proposed for both single-input-single-output (SISO) and multiple-input-multiple-output (MIMO) systems, leading to substantial improvements in detection performance. The second part of thesis examines downlink communication scenarios, where base stations (BS) are equipped with nonlinear PAs. Firstly, a comprehensive analysis of received distortion power is conducted to investigate the effects of multi-user diversity and frequency selectivity. Then, based on the obtained distortion power, a power allocation scheme is proposed to enhance spectral efficiency for linear precoders. Lastly, a low-complexity digital pre-distortion (DPD) design is developed for two-stage hybrid beamforming systems with a fully connected analog beamforming structure. The proposed DPD structure achieves sufficient linearization performance with reduced complexity compared to state-of-the-art methods.
Citation Formats
M. B. Salman, “Nonlinear signal processing for efficient physical layer design in 5G beyond,” Ph.D. - Doctoral Program, Middle East Technical University, 2023.