LTE高密度小細胞網路之可擴展性的移動性管理組件系統設計與實作

The Design and Implementation of MME Scalability for Ultra Dense LTE Small Cell Networks

謝進彥(Jin-Yen Hsieh)

指導教授 ： 趙禧綠

國立交通大學/資訊學院/資訊科學與工程研究所/碩士(2018年)

摘要

隨著可攜式手持裝置普及化以及應用程式的多元發展，使用者對於無線通訊的需求愈來愈高。在快速成長的智慧行動裝置數量和無線資料流量的情況下，高密度小細胞網路(UDN)愈來愈受到重視。然而在使用者個數逐日提升再加上高密度小細胞網路的佈建下，演進數據封包核心網路(EPC)處理相關的控制層相關的流程頻率也會提升，像是一般常見的附著（Attach）、去附著（Detach）和換手（Handover）等控制層的流程，因此當使用者人數多到一定的程度後，演進數據封包核心網路處理控制層相關的程序時延遲也會提升。為了解決這種情況，我們將會利用雲端(Cloud)的架構與演進數據封包核心網路做整合，用來應對未來使用者個數上升時核心網路所會遇到的挑戰。
我們結合了Openairinterface (OAI) 和Openstack提出一個優化的架構，其架構整合OAI中的移動性管理組件(MME)和OpenStack的系統，做到負載平衡和動態擴展的功能，但是因為LTE中的基地台（Evovled Node B, eNB）和MME之間連線的網路使用了串流控制傳輸協議（Stream Control Transmission Protocol, SCTP），而在OpenStack中的負載平衡器並不支援SCTP，因此我們使用Linux Virtual Server (LVS)取代OpenStack原先的負載平衡器，解決了Openstack不支援SCTP的問題。最後我們會基於OAI和OpenStack結合的架構下，測試處理訊息的反應時間和延遲，並且藉由測試結果去制定擴展機制，以達到最佳的雲端架構。

關鍵字

移動性管理 組件 ； 自動擴展機制 ； 附載均衡

並列摘要

With the mobile devices being popular and the diverse development of the application, the requirement for the wireless communication is increasing. Ultra-dense small cell network (UDN) is becoming significant with the rapid of growth of the number of smart device and traffic demands of the mobile network. However, with the increasing number of the users and the erection of UDN, the frequency of processes related to the control plane of Evolved Packet Core (EPC) will also increase. When the number of the users to a certain extent, the latency of the processes of EPC will also increase. In order to solve this problem, we will use the cloud to integrate with EPC to confront the challenge of core network in the future with the increasing number of users.
We combine Openairinterface (OAI) and Openstack to propose an optimized architecture. This architecture integrates the Mobility Management Enity (MME) in EPC and system of OpenStack for load balancing and auto scaling. However, the connectings between EPC and Evovled Node B (eNB) in LTE use the Stream Control Transmission Protocol (SCTP), the load balancer in OpenStack doesn’t support SCTP. Therefore, we use Linux Virtual Server (LVS) instead of the original load balancer of OpenStack to solve this problem. Finally, we will simulate the response time and delay of processes related to the control plane based on this architecture, and use the simulation results to set the scaling policy for achieving better architecture.

並列關鍵字

Mobility Management Enity ； Auto Scaling ； Load Balancing

參考文獻

[1] C. V. N Index, "Cisco Visual Networking Index: Global Mobile Data Traffic Forecast Update, 2016–2021," White Paper, 2017.

[2] David Lopez-Perez, Ming Ding, Holger Claussen and Amir Jafari, "Towards 1gnps/ue in cellular systems: Understanding ultra-dense small cell deployments," IEEE Communications Surveys & Tutorials, vol. 17, no. 4, pp. 2078–2101, 2015.

[3] Naga Bhushan, Junyi Li and Durga Malladi, "Network densification: The dominant theme for wireless evolution into 5g," IEEE Communications Magazine, vol. 52, no.2, 52‒89, 2014.

[4] Amir Jafari, David Lopez-Perez, Ming Ding and Jie Zhang, "Study on scheduling techniques for ultra dense small cell networks," IEEE Vehicular Technology Conference (VTC), 2015.

[5] Nareg Adalian, Georgi Ajaeiya, Zaher Dawy, Imad H. Elhajj, Ayman Kayssi, and Ali Chehab, "Load balancing in LTE core networks using SDN," 2016 IEEE International Multidisciplinary Conference on Engineering Technology (IMCET), pp. 213-217, 2016.