在通信網路中，節點與鏈路的故障保護至關重要，特別是對於一些需要極 低延遲的服務，實現快速且高效的故障恢復機制尤為必要。IP 快速重路由機制 (IP Fast Reroute, IPFRR)的主要目標是將鏈路或節點(路由器)故障後的網路 恢復時間降至最低。使用 IPFRR 機制時，鏈路或節點故障期間的網路恢復時 間僅需幾毫秒。雖然 IPFRR 機制能夠在本地修復點(Point of Local Repair, PLR)啟動修復，但其在操作成本上存在效率問題。為解決此挑戰，本論文提 出了一種基於段路由(Segment Routing, SR)的新型段級恢復框架，將保護應 用於端到端路徑的每個段，而非單一鏈路或節點。 本論文有別於傳統 IPFRR 技術或分段保護的方法，不僅有考慮到節點失效 的情況並加以討論，並且重新設計路徑計算及分段的演算法，最後在模擬時比 較了眾多因素對於整體網路效能的影響。其中，發現本論文的方法在流量丟失 率及平均路徑長度都因標籤深度(Segment List Depth, SLD)增加而降低。模擬 結果顯示，故障導致的流量丟失率因為 SLD 的增加最多可降低 35.47%，並減 少操作路徑長度的增幅高達 8.56%。這些結果證明了此方法能同時實現對多重 故障的有效恢復，在增強網路保護能力及資源利用效率方面的顯著優勢。;Fault protection for nodes and links is critical in communication networks, particularly for latency-sensitive services that require fast and efficient recovery mechanisms. IP Fast Reroute (IPFRR) aims to minimize network recovery time following link or node failures, typically achieving recovery within milliseconds at the Point of Local Repair (PLR). However, IPFRR faces operational efficiency challenges. To address these limitations, this thesis proposes a novel segment-level recovery framework based on Segment Routing (SR), which applies protection to each segment of an end-to-end path rather than to individual links or nodes. Differing from traditional FRR and segment protection methods, this work incorporates node failure considerations and redesigns path computation and segmentation algorithms. Extensive simulations evaluate the impact of multiple factors on overall network performance. Results indicate that increasing the number of SLD reduces both traffic loss rate and average path length. Specifically, failure- induced traffic loss can be reduced by up to 35.47%, while the increase in operational path length is limited to 8.56%. These outcomes demonstrate the effectiveness of the proposed method in achieving robust multi-failure recovery, significantly enhancing network protection and resource utilization efficiency.
