Electronic Thesis and Dissertation Repository

Thesis Format

Monograph

Degree

Master of Engineering Science

Program

Electrical and Computer Engineering

Supervisor

Wang, Xianbin

Abstract

The introduction of new advanced technologies such as higher carrier frequencies, ultra-wide bandwidth, and increased transmission rate in 5G to support ever growing quality-of-service (QoS) demands have brought new challenges such as transmitter-receiver pair specific and domain specific non-orthogonality induced among spatial, time-frequency, and delay-doppler domain radio resource blocks and nonlinear distortions induced among multiple-input multiple-output (MIMO) antennas in spatial domain. In such conditions, current communication systems encounter severe performance degradation and incur higher operational cost. Based on this observation, this thesis aims at creating new multi-dimensional modulation techniques and nonlinear predistortion architectures to achieve higher communication performance with less operational cost.

Firstly, a customized, situation-aware multi-dimensional modulation (MDM) technique is developed with the goal of achieving maximized data rate under joint non-orthogonality conditions in spatial and time-frequency domains. The proposed MDM scheme is designed to take into account the non-orthogonality degrees induced in those domains and jointly optimize the radio resource block attributes to achieve the goal. Secondly, to minimize receiver side operational cost while supporting required data rate under joint non-orthogonality conditions in spatial, time-frequency, and delay-doppler domains, a user-centric multi-dimensional modulation (UC-MDM) technique is developed. The proposed situation-aware and cost-aware UC-MDM is designed to take into account the transmitter-receiver pair specific non-orthogonality degrees and the receiver side operational cost, and intelligently utilize optimum radio resource combination through MDM to achieve the goal. Finally, to reduce complexity of current multi-input digital predistortion (DPD) models for nonlinear distortion compensation, a less complex, decomposed, scalable cross-correlation based DPD (CC-SISO DPD) architecture is proposed for massive MIMO systems.

Through simulation results and analysis, it is demonstrated that the proposed novel, customized, domain specific solutions can achieve higher communication performance with reduced operational cost in future wireless networks.

Summary for Lay Audience

5G and beyond wireless networks are becoming densified and operated with advanced communication technologies to support diverse QoS requirements. However, such advanced technologies have also brought new challenges in multiple domains which need to be closely investigated to efficiently achieve higher performance communication. Therefore, this thesis mainly focuses on those challenges such as transmitter-receiver pair specific non-orthogonality induced among radio resources in spatial, time-frequency, and delay-doppler domains and the nonlinear distortions induced in spatial domain.

First of all, to achieve maximized communication data rate under joint non-orthogonality in spatial, and time-frequency domains, a multi-dimensional modulation (MDM) technique is proposed. The main premise behind the proposed MDM is to simultaneously utilize spatial, and time-frequency domain radio resources through multi-dimensional modulation and also jointly optimize the spatial-time-frequency domain radio resource block attributes to minimize the overall non-orthogonality degree in those domains and thus achieve maximized data rate.

Then, focusing on transmitter-receiver pair specific non-orthogonality degrees induced among radio resources in spatial, time-frequency and delay-doppler domains and the domain specific receiver side operational costs for orthogonality restoration and demodulation, a user-centric multi-dimensional modulation (UC-MDM) technique is proposed. The situation-aware and cost-aware UC-MDM is designed to intelligently utilize the optimum combination of radio resources with optimum radio resource separation through MDM in either spatial-time-frequency or spatial-delay-doppler domains to minimize the user-device operational cost while achieving the required communication data rate.

Finally, considering the severity of nonlinear distortions induced by nonlinear and reverse crosstalk among MIMO antennas and the exponentially increasing operational cost of current multi-input DPD models for nonlinear distortion compensation in MIMO, a decomposed cross-correlation based single-input DPD (CC-SISO DPD) architecture is proposed. The proposed CC-SISO DPD architecture can estimate the nonlinear and reverse crosstalk with high accuracy and with significantly less operational cost than multi-input DPD models. Furthermore, the proposed CC-SISO DPD eliminates the requirement for signal feedback in transmit path, and thus reduces the overall hardware implementation complexity in massive MIMO arrays.

Download

Share

COinS