Intelligent resource management for next-generation wireless networks

Cheng, Runze (2024) Intelligent resource management for next-generation wireless networks. PhD thesis, University of Glasgow.

Abstract

The ongoing development of the fifth-generation (5G) wireless communication networks is reaching a commercial stage, which is capable of providing high reliability, low latency, and high transmission rate services. However, the fast development in terminal devices accompanying the surge of mobile users brings tremendous traffic. Meanwhile, diverse service requirements are starting to rise from emerging applications. These requirements encompass global coverage, ultra-high data rate transmission, ultra-low latency, ultra-dense connectivity, ultra-reliable and secure links, pervasive intelligence, etc. Given the constraint of licensed bandwidth, it may struggle to support diverse requirements of novel applications solely through physical layer advancements in the future. In this background, a promising avenue is to improve the utilization of available resources by investigating new communication paradigms and developing intelligent resource management schemes for the next-generation network. We first review and investigate two categories of potential paradigms and their corresponding resource management schemes in the next-generation network, including heterogeneous symbiotic communication (Het-SC) network and semantic communication (SemCom) network. Additionally, we separately investigate different resource management schemes under two specific Het-SC scenarios and two SemCom scenarios. Concretely, for Het SC scenarios, we propose a deep reinforcement learning (DRL)- based resource management scheme to trade-off different communication resources (especially bandwidth) via service/resource exchanges among symbiotic devices (SDs). In addition to communication resource management, we investigate and propose a DRL-based communication caching resource trade-off scheme for heterogeneous devices in a D2D communication network, which can improve the utilization of both communication and caching resources, as well as provide satisfied services for user equipments (UEs). Furthermore, as a preliminary investigation of resource management in SemCom scenarios, a DRL-based communication-computational resource trade-off scheme is proposed for end-to-end (E2E) SemCom-empowered artificial intelligence (AI)-generated content (AIGC) transmission. The scheme is capable of ensuring required latency and content quality with limited communication resources while improving computation resource utilization. In parallel, to maximize the utilization of communication and computational resources and provide high-quality services in multimodal SemCom (M SemCom) networks, we propose a DRL-based communication-computational resource trade off scheme to adjust the computing workload and transmitted data size in SemCom. Through comprehensive simulations, intelligent resource trade-off schemes are examined within diverse promising scenarios. Numerical results validate the performance gains achieved by the DRL-based resource trade-off schemes in terms of service quality and resource utilization.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Subjects: T Technology > T Technology (General)
• Colleges/Schools: College of Science and Engineering > School of Engineering
• Supervisor's Name: Sun, Dr. Yao, Liu, Professor Bo and Imran, Professor Muhammad
• Date of Award: 2024
• Depositing User: Theses Team
• Unique ID: glathesis:2024-84083
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 20 Feb 2024 09:06
• Last Modified: 22 Feb 2024 12:58
• Thesis DOI: 10.5525/gla.thesis.84083
• URI: https://theses.gla.ac.uk/id/eprint/84083

Related URLs

• Article DOI
• Article DOI
• Article DOI
• Article DOI
• https://doi.org/ 10.1109/GLOBECOM46510.2...
• https://doi.org/10.1109/ICC45041.2023.10...
• https://doi.org/10.1109/ICCT56141.2022.1...
• https://doi.org/10.1109/ICCWorkshops5795...