Xia, Le (2024) Wireless resource optimization in semantic communication-based cellular networks. PhD thesis, University of Glasgow.
Full text available as:
PDF
Download (7MB) |
Abstract
Recent advances in artificial intelligence (AI) have made semantic communication (SemCom) a promising solution that can yield significant benefits in guaranteeing high spectrum resource utilization, information interaction efficiency, and transmission reliability. Compared with conventional bit communication (BitCom), which guarantees the precise reception of transmitted bits, the accurate delivery of semantics implied in desired messages becomes the cornerstone of SemCom. Nevertheless, the unique semantic coding and background knowledge matching mechanisms make it challenging to achieve efficient wireless resource optimization for multiple mobile users (MUs) in SemCom-enabled cellular networks. To this end, the objectives of this thesis are to investigate different optimal wireless resource management strategies in different SemCom network scenarios. Specifically, a total of four differing scenarios are taken into account here, i.e., general SemCom-enabled networks (SC-Nets), energy efficient SemCom-enabled networks (EE-SCNs), hybrid semantic/bit communication networks (HSB-Nets), and SemCom-enabled vehicular networks (SCVNs).
For the general SC-Net scenario, we address two fundamental problems of user association (UA) and bandwidth allocation (BA) on the downlink side, where two different knowledge-matching states of all MUs in the SC-Net are identified. Most importantly, a concept of bit-rate-to-message-rate (B2M) transformation is developed along with a new metric, namely system throughput in message (STM), to measure the overall network performance in a semantic manner. By formulating a joint STM-maximization problem for each SC-Net case, the corresponding optimal solution is then proposed. As for the EE-SCN scenario, we focus on jointly addressing the power allocation and spectrum reusing problems involving the device-to-device (D2D) SemCom approach, in which the energy efficiency model of SemCom is dedicatedly defined. To maximize the average energy efficiency of all cellular users (CUEs) and D2D users (DUEs), a fractional-to-subtractive problem transformation method, a heuristic algorithm, and a Hungarian method are employed together to obtain the optimal solutions. In terms of the HSB-Net scenario, the UA, mode selection (MS), and BA problems are jointly optimized on the uplink side, where two modes of SemCom and BitCom are available for all MUs’ selection. By leveraging the B2M method, the unified performance metric of both modes is identified. Then, we specially develop a knowledge matching-aware two-stage tandem packet queuing model and theoretically derive the average packet loss ratio and queuing latency. Based on the corresponding formulated problem, an optimal resource management strategy is proposed by utilizing a Lagrange primal-dual transformation method and a preference listbased heuristic algorithm with polynomial-time complexity. Finally, in line with the next-generation ultra-reliable and low-latency communication (xURLLC) requirements, we identify and jointly tackle two inevitable problems of knowledge base construction (KBC) and vehicle service pairing (VSP) in the SCVN scenario. In this case, we first derive the knowledge matching based queuing latency specific for semantic data packets, and then formulate a latency-minimization problem subject to several KBC and VSP related reliability constraints. Afterward, a SemCom-empowered Service Supplying Solution (S4) is proposed along with the theoretical analysis of its optimality guarantee and computational complexity. Numerical results in each of the four scenarios demonstrate significant superiority and reliability of our proposed solutions in terms of various performance metrics compared with multiple benchmarks. All the works presented in this thesis can serve as pioneers in exploring the potential of applying SemCom to wireless cellular networks while ensuring optimal resource management.
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, Zhang, Professor Lei and Imran, Professor Muhammad |
Date of Award: | 2024 |
Depositing User: | Theses Team |
Unique ID: | glathesis:2024-84571 |
Copyright: | Copyright of this thesis is held by the author. |
Date Deposited: | 27 Sep 2024 09:34 |
Last Modified: | 27 Sep 2024 10:25 |
Thesis DOI: | 10.5525/gla.thesis.84571 |
URI: | https://theses.gla.ac.uk/id/eprint/84571 |
Related URLs: |
Actions (login required)
View Item |
Downloads
Downloads per month over past year