| dc.contributor.advisor | 蔡子傑 | zh_TW |
| dc.contributor.advisor | Tsai, Tzu-Chieh | en_US |
| dc.contributor.author (Authors) | 黃奕晟 | zh_TW |
| dc.contributor.author (Authors) | Huang, Yi-Cheng | en_US |
| dc.creator (作者) | 黃奕晟 | zh_TW |
| dc.creator (作者) | Huang, Yi-Cheng | en_US |
| dc.date (日期) | 2024 | en_US |
| dc.date.accessioned | 4-Oct-2024 10:47:35 (UTC+8) | - |
| dc.date.available | 4-Oct-2024 10:47:35 (UTC+8) | - |
| dc.date.issued (上傳時間) | 4-Oct-2024 10:47:35 (UTC+8) | - |
| dc.identifier (Other Identifiers) | G0111753164 | en_US |
| dc.identifier.uri (URI) | https://nccur.lib.nccu.edu.tw/handle/140.119/153916 | - |
| dc.description (描述) | 碩士 | zh_TW |
| dc.description (描述) | 國立政治大學 | zh_TW |
| dc.description (描述) | 資訊科學系 | zh_TW |
| dc.description (描述) | 111753164 | zh_TW |
| dc.description.abstract (摘要) | 隨著5G網路的快速發展，網路流量和用戶數也呈現爆炸性的增長，許多新型應用也應運而生，在不同場景下，用戶對網路的需求各不相同，但主要都離不開三大類：增強型行動寬頻（enhanced mobile broadband, eMBB），提供更大頻寬容量。極低延遲的可靠通訊（ultra-reliable and low latency communications, uRLLC），提供小於1毫秒及更可靠的通訊。巨量物聯網通訊（massive machine type communications, mMTC），提高連線數，滿足每平方公里最少有一百萬的連線裝置數；最終提升用戶使用5G網路時的服務品質(QoS, Quality of Service)。
在本研究中，我們模擬多個UE (User Equipment, 即使用設備)在固定數量的基地台之間移動，在網路功能虛擬化(Network Function Virtualization，NFV)與軟體定義網路(Software Defined Networking，SDN)的基礎上，結合聯邦學習技術，預測下一個時間點可能有最多UE的基地台，由會話管理功能網元(SMF)指派該區域的用戶平面功能網元(UPF)為上行分流器(Uplink Classifier, ULCL)，針對UE送到基地台的流量進一步分類，根據延遲與頻寬需求引導至不同的錨點(Anchor UPF)，最後送往數據網路，當使用者需要較即時性的反應，如視訊直播、車聯網，將流量引導至較近的Edge UPF，由多接取邊緣運算(Multi-access Edge Computing, MEC)裝置就近處理，減少大量數據在回程(Backhaul)網路上傳輸，不但減少延遲，也解決不同營運商在核心網路中因為數據隱私而造成無法有效整合的問題。 | zh_TW |
| dc.description.abstract (摘要) | With the rapid development of 5G networks, network traffic and user numbers have experienced explosive growth, and many new applications have emerged. In different scenarios, users have varying network requirements, which can be broadly classified into three main categories: enhanced Mobile Broadband (eMBB), providing larger bandwidth capacity; Ultra-Reliable and Low Latency Communications (uRLLC), offering sub-millisecond and more reliable communication; and massive Machine Type Communications (mMTC), increasing the number of connections to support at least one million connected devices per square kilometer. The ultimate goal is to enhance the Quality of Service (QoS) for users when utilizing 5G networks.
In this research, we simulate multiple User Equipment (UE) moving between a fixed number of base stations. Building upon the foundations of Network Function Virtualization (NFV) and Software Defined Networking (SDN), we integrate federated learning techniques to predict the base station that may have the most UE at the next time point. The Session Management Function (SMF) designates the User Plane Function (UPF) in that area as the Uplink Classifier (ULCL) to further classify the traffic sent by UE to the base station. Based on latency and bandwidth requirements, the traffic is guided to different Anchor UPFs and finally sent to the data network. When users require more real-time responses, such as video streaming or vehicle-to-everything (V2X) communication, the traffic is directed to a closer Edge UPF, where it is processed by nearby Multi-access Edge Computing (MEC) devices. This approach not only reduces latency by minimizing the transmission of large amounts of data over the backhaul network but also addresses the issue of ineffective integration between different operators in the core network due to data privacy concerns. | en_US |
| dc.description.tableofcontents | 第一章 緒論 1
1.1 研究背景與動機 1
1.2 研究目的 2
1.3 論文架構 4
第二章 相關研究 5
2.1 5G核心網路 5
2.1.1 各網元功能介紹 5
2.1.2 5G與4G的不同之處 6
2.1.3 5G系統下兩種主要的UE換手方式 8
2.2 流量分類的相關研究 10
2.3 不同應用對於延遲及頻寬的需求比較 13
2.4 Adaptive Bitrate Streaming 自適性串流 14
2.5 Network Data Analytics Function (NWDAF) 15
2.5.1 NWDAF提供的服務 16
2.5.2 分散式佈署NWDAF網元 18
2.6 聯邦學習介紹 19
2.6.1 聯邦學習分類 20
第三章 設計與實作 24
3.1 環境建置 25
3.1.1 在Kubernetes佈署5G核心網路 25
3.1.2 加入網路數據分析功能網元 27
3.2 場景設計 28
3.3 數據蒐集 29
3.4 實驗一：傳統集中式訓練方法 32
3.5 實驗二：聯邦學習過程 32
3.5.1 不同時間點之上行分流器(ULCL)位置 33
3.5.2 客戶端使用長短期記憶模型(LSTM) 35
3.5.3 伺服器端聯邦聚合過程 35
3.6 實驗三：不進行移動性預測之上行分流效能 37
第四章 實驗結果與效能分析 39
4.1 網路頻寬利用率 39
4.2 延遲時間評估 41
4.3 斷線次數比較 43
4.4 減少交換資料次數 43
第五章 結論與未來展望 45
第六章 參考資料 46 | zh_TW |
| dc.format.extent | 3414841 bytes | - |
| dc.format.mimetype | application/pdf | - |
| dc.source.uri (資料來源) | http://thesis.lib.nccu.edu.tw/record/#G0111753164 | en_US |
| dc.subject (關鍵詞) | 網路功能虛擬化 | zh_TW |
| dc.subject (關鍵詞) | 軟體定義網路 | zh_TW |
| dc.subject (關鍵詞) | 聯邦學習 | zh_TW |
| dc.subject (關鍵詞) | 上行分流器 | zh_TW |
| dc.subject (關鍵詞) | 多接取邊緣運算 | zh_TW |
| dc.subject (關鍵詞) | 服務品質 | zh_TW |
| dc.subject (關鍵詞) | Network Function Virtualization | en_US |
| dc.subject (關鍵詞) | Software Defined Networking | en_US |
| dc.subject (關鍵詞) | Federated Learning | en_US |
| dc.subject (關鍵詞) | Uplink Classifier | en_US |
| dc.subject (關鍵詞) | Multi-access Edge Computing | en_US |
| dc.subject (關鍵詞) | Quality of Service | en_US |
| dc.title (題名) | 用於優化5G核心網路上行鏈路流量轉發的聯邦學習之研究 | zh_TW |
| dc.title (題名) | A Study on Federated Learning for Optimizing Uplink Traffic Forwarding in 5G Core Networks | en_US |
| dc.type (資料類型) | thesis | en_US |
| dc.relation.reference (參考文獻) | [1] "The 5G Core Network Demystified"[Online]. Available:
https://infohub.delltechnologies.com/en-us/p/the-5g-core-network-demystified/
[2] Devaki, C., Rainer, L.& Juho, P.(2019) “5G for the Connected World. ” John Wiley & Sons Ltd.
[3] Dennis Lanov ,(March 2022). “Shortest Path Assisted User Plane Function Selection On 5g Session Management Function”. In Technical Disclosure Commons.
[4] Tze-Jie, T., Fu-Lian, W., Wei-Ting, H., Jyh-Cheng, C., Cheng-Ying, H. (September 2020). ”A Reliable Intelligent Routing Mechanism in 5G Core Networks” In MobiCom '20: Proceedings of the 26th Annual International Conference on Mobile Computing and Networking.
[5] Arthur, C., Carlos,K., G´eza, S., Bal´azs P´eter Gero, Judith, K., Stˆenio, F., and Djame, S. (2009). “A survey on internet traffic identification,” IEEE communications surveys & tutorials, vol. 11, no. 3.
[6] Michelle, C., Lars, E., Joe, T., Magnus, W., Stuart, C. (2011). “Internet assigned numbers authority (iana) procedures for the management of the service name and transport protocol port number registry,” Tech. Rep.
[7] Ping, D., Akihiro, N., Lei, Z., Jing, M., Ryokichi, O., (September 2021). “Service-aware 5G/B5G Cellular Networks for Future Connected Vehicles” In 2021 IEEE International Smart Cities Conference (ISC2).
[8] “The 5G Guide - A Reference For Operators” (2019)[Online]. Available: https://www.gsma.com/wp-content/uploads/2019/04/The-5G-Guide_GSMA_2019_04_29_compressed.pdf
[9] “Best Video Bitrate for Streaming in 2022”, [Online]. Available: https://www.dacast.com/blog/best-video-bitrate-for-streaming/
[10] Wiki, "Adaptive bitrate streaming"[Online]. Available: https://en.wikipedia.org/wiki/Adaptive_bitrate_streaming
[11] “3GPP TS 23.501” [Online]. Available: https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=3144
[12] “Applied Machine Learning, Federated Learning. ”[Online]. Available: https://smartnets.yale.edu/research/applied_ml
[13] “multus-cni” [Online]. Available: https://github.com/k8snetworkplumbingwg/multus-cni/tree/master
[14] “mnc_NWDAF” [Online]. Available:
https://github.com/net-ty/mnc_NWDAF
[15] Understanding LSTM Networks:
http://colah.github.io/posts/2015-08-Understanding-LSTMs/.
[16] “edge-computing”, [Online]. Available: https://www.3gpp.org/technologies/edge-computing | zh_TW |
