隨著智慧型手機的普及，人們對行動網路的需求大增，因此促進行動網路科技高速發展，然而昂貴的電信網路設備增加電信營運商的建置成本。在專用的硬體設備與軟體下，電信營運商難以依照使用情境提供與擴充電信網路設備與功能。這項問題在5G 的架構下有所改進，透過 5G Service Base Architecture (SBA) 架構下，電信營運商可以將核心網路部署至一般伺服器中並降低成本。 在本篇論文中，先以 5G SBA 架構為基礎，將 5G 核心網路各個網路元件容器化，並藉由 Kubernetes 雲端運算技術打造出一套以雲原生 (Cloud Native)技術為基礎之5G 核心網路平台。接著，導入開源交換機 SONiC來傳輸與乘載使用者經由基地台至5G 核心網路間的資料流，模擬電信營運商真實的營運架構，並利用 Cloud Native 的架構，自動化管理、測量與監控 5G 核心網路元件的資源使用量，本實驗藉由模擬完整的虛擬化 5G 核心網路，分析與測量不同案例下 5G 核心網路各網路元件的資源使用量。;The impressive demand for mobile data access with mobile phones promotes the rapid development of mobile network technologies and services. Telecommunication operators conventionally deploy dedicated hardware and software to provide mobile services, which results in huge construction costs by not only telecommunications equipment but also service maintenance telecom operators. In addition, this conventional way is difficult for telecommunication operators to expand network infrastructure and customize network functions in response to various scenarios and usages. To remedy this problem, the 5G architecture takes advantage of the 5G Service Base Architecture (SBA) architecture, which enables telecommunication operators to deploy core networks to commercial off-the-shelf server equipment and then reduce the costs. In this thesis, we develop a Cloud Native based 5G core network platform. Our study is based on the 5G SBA architecture, where each network component in 5G core network is containerized with the building support of the Kubernetes and Cloud computing technologies. Also, our study takes in the open source networking switch (SONiC) to transmit data flows from the radio base station to the 5G core network. By simulating the operations and scenarios of a telecommunication system, the proposed Cloud Native architecture can automatically deploy, manage, and monitor resource usages of each network component in 5G core system platform. With the prototypical implementation, our study clearly shows the functionality of the Cloud Native based 5G core network platform. The resource information of containerized network components in 5G core network can be promptly reported under different experimental cases.
