Yue Dai

EECS Department, University of California, Berkeley

Technical Report No. UCB/EECS-2024-30

May 1, 2024

http://www2.eecs.berkeley.edu/Pubs/TechRpts/2024/EECS-2024-30.pdf

The massive multiple-input-multiple-output (MIMO) wireless communication technology is one of the critical parts in the 5G and 5G+ communication systems, which offers an energy- and cost-efficient, secure, and robust solution for the next generation of communication systems. This dissertation focuses on developing integrated signal processing for massive MIMO systems to improve the performance and efficiency of massive MIMO for future communication systems. It provides a comprehensive analysis of the design and prototyping of scalable mmWave massive MIMO testbeds from both algorithmic and hardware architecture perspectives. It focuses on achieving better data aggregation, linear scaling of computation complexity, cost and energy efficiency, and sustainable development. It demonstrates the feasibility of such systems by designing an algorithm-adaptable, scalable, and platform-portable generator for massive MIMO baseband processing systems, which can be customized for different MIMO systems and hardware configurations. The generator is evaluated by using various channel estimation methods, such as the flat fading, line-of-sight, Rician, and QuaDRiGa channel models, and various hardware parameter values, demonstrating its algorithmic adaptability and effectiveness. The results show that the generator has competitive power consumption and can significantly improve the demodulation error vector magnitude by integrating beamspace methods. Additionally, to show the power of machine learning in the integrated signal processing of massive MIMO systems, a complex-valued neural network aided channel estimation method for massive MIMO systems with interference is proposed and evaluated. The network is trained with a wideband interference, and the method is evaluated with different design and training choices. The result shows an up to 3.16 times improvement in channel estimation performance with a shorter pilot overhead compared to the traditional method, demonstrating the effectiveness of the proposed method.

Advisors: Borivoje Nikolic

---

BibTeX citation:

@phdthesis{Dai:EECS-2024-30,
    Author= {Dai, Yue},
    Title= {Integrated Signal Processing for Massive MIMO Systems},
    School= {EECS Department, University of California, Berkeley},
    Year= {2024},
    Month= {May},
    Url= {http://www2.eecs.berkeley.edu/Pubs/TechRpts/2024/EECS-2024-30.html},
    Number= {UCB/EECS-2024-30},
    Abstract= {The massive multiple-input-multiple-output (MIMO) wireless communication technology is one of the critical parts in the 5G and 5G+ communication systems, which offers an energy- and cost-efficient, secure, and robust solution for the next generation of communication systems. This dissertation focuses on developing integrated signal processing for massive MIMO systems to improve the performance and efficiency of massive MIMO for future communication systems. It provides a comprehensive analysis of the design and prototyping of scalable mmWave massive MIMO testbeds from both algorithmic and hardware architecture perspectives. It focuses on achieving better data aggregation, linear scaling of computation complexity, cost and energy efficiency, and sustainable development. It demonstrates the feasibility of such systems by designing an algorithm-adaptable, scalable, and platform-portable generator for massive MIMO baseband processing systems, which can be customized for different MIMO systems and hardware configurations. The generator is evaluated by using various channel estimation methods, such as the flat fading, line-of-sight, Rician, and QuaDRiGa channel models, and various hardware parameter values, demonstrating its algorithmic adaptability and effectiveness. The results show that the generator has competitive power consumption and can significantly improve the demodulation error vector magnitude by integrating beamspace methods. Additionally, to show the power of machine learning in the integrated signal processing of massive MIMO systems, a complex-valued neural network aided channel estimation method for massive MIMO systems with interference is proposed and evaluated. The network is trained with a wideband interference, and the method is evaluated with different design and training choices. The result shows an up to 3.16 times improvement in channel estimation performance with a shorter pilot overhead compared to the traditional method, demonstrating the effectiveness of the proposed method.},
}

EndNote citation:

%0 Thesis
%A Dai, Yue 
%T Integrated Signal Processing for Massive MIMO Systems
%I EECS Department, University of California, Berkeley
%D 2024
%8 May 1
%@ UCB/EECS-2024-30
%U http://www2.eecs.berkeley.edu/Pubs/TechRpts/2024/EECS-2024-30.html
%F Dai:EECS-2024-30