多切片下第五代行動通訊增強型行動寬頻演算法優化

Optimization of Scheduling Algorithms in Multi 5G eMBB Slicing

高竹儀(Chu-Yi Kao)

指導教授 ： 曹恆偉

國立臺灣大學/電機資訊學院/電信工程學研究所/碩士(2022年)

https://doi.org/10.6342/NTU202201821

摘要

隨著科技的演進，社會上對於無線通訊的服務要求也日益增加，全世界也在2020年開始從第四代行動通訊技術 (4G LTE)往第五代行動通訊技術(5G NR)邁進。5G NR相較於4G LTE主要在三個面向進行了大幅度的改善，分別為大規模機器類型通信(Massive Machine Type Communication, mMTC)、超高可靠與低時延通信(Ultra-reliable and Low Latency Communications, URLLC)及增強移動寬頻(Enhanced Mobile Broadband, eMBB)。傳統的第三代合作夥伴計劃以及這幾年由全世界主要電信商所推廣的開放式無線電接取網路(Open Radio Access Network, O-RAN)針對5G基地台的功能性以及架構進行了詳盡且明確的規範定義，但在規範中如何針對不同的環境以及服務需求選取最洽當的規範參數則仍是有待各家電信廠商自行開發及研究，因此也造就了同樣規範下，不同廠商產品會有不同的效能以及服務品質。
MAC層的排程器又是其中影響基地台服務效能的重大關鍵，其主要的功能為針對不同的用戶安排不同的頻譜資源以及編碼的選擇。但是大多數文獻中對排程器都是以C++或是Python語言自行開發的模型進行行為模擬，聚焦在一定範圍內，在不同場景下給定已知狀態的需求進行資源調度及排程以達到針對性的分析優化，或是以整個系統包含手機基地台和核心網進行概念性的資源分配分析。前者會簡化定義範圍外的行為流程以及不同硬體架構下演算法運算複雜度的限制，後者則是會忽略實際上物理層安排資源的限制，例如：在分析上行資源排程的時候，忽略了物理層的物理下行控制通道(PDCCH)和物理上行共享通道(PUSCH)所需的時間差異(k2)等。
本論文不同於上述兩者，以O-RAN 架構為核心透過真實的5G端對端系統針對增強移動寬頻這項服務目標提出一個可實作並有效的MAC層排程器，進行包含多用戶下的吞吐量及延遲性進行優化及分析。
本論文首章為論文簡介，第二章為背景知識介紹，從第三章開始為本論文主要貢獻，也就是可實際應用的MAC層排程器，包含演算法的系統架構、基地台吞吐量和資料延遲分析，以及考量實際基地台系統的穩定度和硬體限制所進行的C語言離線演算法數據模擬等，並在第四到第六章進行分析模擬及實際測試結果，第七章為結語與未來展望，最末章為附錄。

關鍵字

第五代無線通訊規範 ； 媒體網路控制排程器 ； 服務質量 ； 網路切片 ； 延遲性 ； 吞吐量

並列摘要

As technology evolves rapidly, so does the company's need for wireless communication services. The world has also been shifting from the fourth generation (4G LTE) to the fifth generation (5G NR) since 2020. Compared to 4G LTE, 5G NR has been substantially improved in three aspects: Massive Machine Type Communication (mMTC), Ultra-reliable and Low Latency Communications (URLLC), and Enhanced Mobile Broadband (eMBB). The traditional third-generation partnership program and the Open Radio Access Network (O-RAN) announced by massive telecom companies worldwide have detailed the functionality and architecture of 5G base stations and a clear specification definition. Nevertheless, choosing the most appropriate specification parameters for various environments and service requirements is difficult for each telecommunications company to develop and investigate. Therefore, different manufacturers will have different levels of efficiency and quality of service under the same specification.
The MAC layer scheduler is one of the critical factors affecting the base station's service efficiency. Its primary function is to organize different spectrum resources and coding choices for different users. However, in most research papers, the scheduler is either a self-developed model in C++ or Python for behavior simulation or the entire system, including mobile phone base stations and core network for conceptual resource allocation analysis. The former will simplify the behavioral flux outside the defined scope and the limitation of the computational complexity of the algorithm under different hardware architectures. The latter ignores the requirements of physical layer allocation of resources; for example: when analyzing the uplink resource scheduling, it may miss the time difference (k2) between the physical downlink control channel (PDCCH) and physical uplink shared channel (PUSCH).
In this master thesis, we take O-RAN architecture as the core. We use a whole 5G End-to-End system and introduce a feasible and effective MAC scheduler algorithm to optimize the throughput and latency for users under the eMBB service.
Chapter 1 will briefly introduce our thesis motivation and the main contribution. Chapter 2 shows some basic background information for 5G system architecture, features, and some physical setting usage. Also, we introduce some traditional scheduling algorithms for LTE for reference. Chapter 3 analyzes the traffic and latency the 5G network needs to deal with and proposes a testing platform gather processing time consumption logs when UE performs data transmission. Chapter 4 introduces our proposed scheduling algorithm and its system architecture. Besides, we have an analysis of our scheduling algorithm. In chapters 5 and 6, we did some experiments on the End-to-End system and the offline simulation model to obtain more testing results for our scheduling algorithm.

並列關鍵字

5G ； MAC Scheduler ； Quality of Service ； Network Slicing ； Latency ； Throughput

參考文獻

[1] Qualcomm Tech, Future of 5G, 2020 February

[2] 3GPP TS 23.501 version 16.6.0 Release 16 5G; System architecture for the 5G System (5GS), ETSI TS 123 501 V16.6.0 (2020-10)

[3] 3GPP TS 38.300 version 16.4.0 Release 16, 5G; NR; NR and NG-RAN Overall description; Stage-2, ETSI TS 138 300 V16.4.0 (2021-01)

[4] O-RAN.WG1.O-RAN-Architecture-Description-v06.00 (2022-03)

[5] 3GPP TS 38.321 version 16.1.0 Release 16, 5G; NR; Medium Access Control (MAC) protocol specification, ETSI TS 138 321 V16.1.0 (2020-07)