Please use this identifier to cite or link to this item:
Full metadata record
DC FieldValueLanguage
dc.contributor.authorXiao, Xiaoen_US
dc.identifier.citationXiao, X. (2022). Digital beamforming with data-driven models and effective optimization. Doctoral thesis, Nanyang Technological University, Singapore.
dc.description.abstractDigital beamforming is an emerging technology with wide and critical applications in wireless communications and radar systems. There are increasing demands for more efficient and sophisticated engineering beamforming techniques for larger arrays under a constraint engineering cost for current and future cellular communications and digital array radars. This PhD study explores data-driven models with global optimization and artificial intelligence for fast and flexible digital beamforming of practical values. The classical digital beamforming is largely based on analytical or numerical solutions of specific beamforming problems with limitations in flexibility and performance. Inspired by big-data and artificial intelligence, the idea of data-based digital beamforming is proposed for faster, more flexible and high-performance solutions to sophisticated beamforming problems. The data-driven beamforming utilizes off-line generated weights as the data and establishes the link between the required radiation pattern and weights, carrying out the beamforming that benefits the wireless system. This thesis has made four contributions along with the exploration of the topic. Firstly, sophisticated digital beamforming for linear arrays has been extensively studied. The study compares and evaluates the effectiveness of a few most popular global optimization algorithms. These algorithms are deployed under a 1-dimensional linear array setup and compared in challenging beamforming problems. After extensive experiments, the BA is found to be more effective than other candidate algorithms hence used to generate training samples for the data-based digital beamforming. Secondly, an effective data-driven model is proposed and tested for linear array wide nulling and null steering problems. A data-driven adaptive digital beamforming model using General Regression Neural Network (GRNN) is developed to fit the beamforming training samples generated by the BA. The model considers the adaptive beamforming of the main lobe with steerable nulls of varying widths. The numerical experiments and evaluation show that the proposed data-driven digital beamforming model is effective even without any further optimization. Thirdly, the proposed concept was extended to more challenging beamforming problems. Improvement of conventional pattern optimization approach and corresponding extension into 2D array formation is made. A novel multiplexer-based coding scheme, hybrid cost function and improved BA are proposed and tested under various beamforming problems. The outcome serves as important tool for data generation to 2D planar array data-driven beamforming model and the effectiveness is proven in this chapter. Lastly, the idea of a data-driven model is extended to planar array beamforming of greater demands and practical values, using a combination of the multiplexer-based beamforming model with the improved hybrid cost BA and a Deep Neural Network model trained under supervised and unsupervised schemes. The data-driven model shows a promising performance and provides flexible wide null steering capability.en_US
dc.publisherNanyang Technological Universityen_US
dc.rightsThis work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0).en_US
dc.subjectEngineering::Electrical and electronic engineering::Antennas, wave guides, microwaves, radar, radioen_US
dc.titleDigital beamforming with data-driven models and effective optimizationen_US
dc.typeThesis-Doctor of Philosophyen_US
dc.contributor.supervisorLu Yilongen_US
dc.contributor.schoolSchool of Electrical and Electronic Engineeringen_US
dc.description.degreeDoctor of Philosophyen_US
item.fulltextWith Fulltext-
Appears in Collections:EEE Theses
Files in This Item:
File Description SizeFormat 
Theses_XIAO Xiao_Final.pdf9.88 MBAdobe PDFThumbnail

Page view(s)

Updated on Feb 27, 2024

Download(s) 50

Updated on Feb 27, 2024

Google ScholarTM


Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.