隨著當代行動網路通訊技術的成熟。因行動使用者數量越來越多，帶動無線寬頻的使用量提升。在工業界、企業界、智慧工廠自動化，物聯網( Internet of Things, IoT)的需求越來越多，免執照頻段的無線網路已不敷使用。微營運商 (µO) 就是在這個背景下發展的新興網路服務模式。建置小基地台(Small Cell)溝通設施，來進行區域化服務的遞送。微營運商( Micro Operator , µO )具有可佈署特定區 域特定服務的能力，可管控各 µO 區域的上傳與下載的資料傳輸，只要是場域擁有者皆有機會成為微營運商。 本研究針對區域化服務提出 Cross-domain Network Slicing(CNS)系統，使微營運商用戶可就近存取 µO 的資料，或是特定應用的網路服務需求，也可Tunnel 連結至網際網路的雲端資料中心存取。區域化服務除了可就近提供網路服務減少頻寬資源消耗外，其分散式的區域型服務網路架構可將應用程式、數據資料與服務的運算，由雲端的資料中心節點，移往網路邏輯上的邊緣節點來處理，可降低連網延遲，滿足 5G 網路對改善延遲性的應用需求。 本 論 文 提 出 Desion Tree Based Flow Redirection (DTBFR)機制，藉由蒐集視訊服務伺服器負載資訊，提供給 SDN 控制器針對訊務流重導進行決策。實驗結果顯示僅有 3.2%訊務到資料中心存取服務，顯示本機制在 µO 使用者訊務導向決策上有較好的表現，以 µO 使用者所在區域 µO 為優先、接者是相對近的 µO、最後才會到遠端資料中心存取。 ;With the stable development of mobile network communication technique, the utilization of wireless broadband has grown up due to the increasing mobile users. The unlicensed band has become not enough to be used because of higher requirement in Industry, Enterprise, manufacture automation region, Internet of Things (IoT), etc. As result, Micro operator (μO) proposes one brand-new business model to build a small cell infrastructure for specific local services. The μOs can provide the establishment for deployment of local services and can control & manage the transmission of data uplink and downlink. If you are the owner around area, you have the qualification to be a micro operator. This research proposed Cross-domain Network Slicing (CNS) system which can make the tenants of μOs directly access the data from nearest region or application of specific network service and also can link to cloud data center accessing data by tunneling. Not only local distributed network infrastructure can decrease the wastage of bandwidth but have the advantage of low delay in application program, data information, and service computing because it can move data from the center node to the end node. Meanwhile, it satisfied the need of 5G network in the future. This paper proposed a method called Decision Tree Based Flow Redirection (DTBFR). By collecting the information of streaming server loading data can make Software Defined Network (SDN) controller to make the routing decision. Experiment result shows that only the 3.2% traffic workload directed to the data center to access. This approach has better performance in flow redirection, and the priority in this strategy is the μOs who located in this area, and then near the area, and further more are faraway datacenter.
