網際網路(Internet)的迅速發展及使用人數的遞增，有效率地頻寬配置管理和服務品質的提供是目前網際網路所要面臨的重要問題。近幾年，由網際網路標準制定組織(Internet Engineering Task Force，IETF)所提出的多重協定標籤交換技術(Multi-protocol Label Switching，MPLS)是為實現未來寬頻網際網路的可行方法，MPLS整合各種IP交換技術的優點，引入標籤(Label)的觀念，來提升整體網路的運作效率，比起傳統網路更具彈性及擴充性。MPLS明確路由安排機制可使整體網路資源的利用理想化，更可提供較佳的服務品質(QoS)支援。本論文的主要研究重點在探討MPLS網路的標籤交換路徑的安排(LSP Assignment)，並考量不同服務等級路徑特性，將資源作有效的分配，以期達到不同連線服務之需求。LSP路由演算法於本論文中列舉Best-fit Shortest Path Algorithm (BSP)以及Worst-shortest Path Algorithm (WSP)兩種演算法，並針對不同等級的服務分別來做路徑選擇模擬之比較。經由考慮優先權安排LSP，希望可以達到服務品質之保障。而為讓網路中的鏈結(Link)頻寬的分配更具效率，在論文中亦提出伸縮限制頻寬(Elastic Constrained Bandwidth)的配置方式，由模擬結果得知，此配置方式可有效紓解高優先權路徑過於集中之缺點。在文中亦說明利用最小切割演算法計算最小切割來尋求減少網路關鍵路段負載，降低網路壅塞情形產生。 The paper works on bandwidth-constrained QoS routing looks for one feasible path that meets a certain bandwidth requirement. Internet Engineering Task Force (IETF) Multi-protocol Label Switching (MPLS) standard has made explicit routes practicable and improves network utilization. In other words, by using MPLS techniques it is possible to set up explicit routes for data flows that are constrained by path, resource availability and requested Quality of Service. In this article, we discuss two path routing algorithms for MPLS LSP Assignment. We termed them as Best-fit shortest path algorithm (BSP) and Worst-shortest path algorithm (WSP). In order to guarantee high-priority services, we will find the best one of path routing algorithms for setting up the path of high-priority services. We make a comparison between these path routing algorithms for the two types of services (Guaranteed service and Control-load service). Then we also propose a link-sharing method, Elastic Constrained Bandwidth. The experimental results indicate that the method solves the problem that high priority LSPs concentrate on some links. We then describe what the Minimum-Cut algorithm is and how to separate a network into two parts. By using the Minimum-Cut algorithm, we could easily calculate the Minimum-cut links, which we termed critical links, and decrease the complexity of a network. We expect that these above algorithms could accomplish the QoS performance and further the utility of a network.
