隨著網際網路的進一步發展，現今大多都被高速和與影像相關的服務所佔據，下一代網路將對ISP的網路進行優化。美國國家科學基金計畫提出了新的網路架構Named Data Network（NDN），建立網路資料暫存和基於內容命名的路由機制，來降低傳輸流量、加快相應速度。 本論文將以NDN的架構，提出合適NDN的QoS方法，針對NDN網路中不同類型資料做QoS分類，導入IP封包逾時即丟棄的存活時間(Time To Life, TTL)概念，取代傳統記憶體更新方式-先進先出演算法(FIFO)及最久沒有使用演算法（LRU），給予不同的存活時間，資料在來源端時先做分類，定義各種資料在暫存器中的存活時間，降低路由器暫存器(Content Store, CS)的負載量，在不增加服務延遲時間的前提下，使NDN網路的建置得以減少大量記憶體建置成本，在低網路傳輸的狀況下，減少服務延遲。並且動態調整資料封包存活時間類型，以網路環境參數最佳化暫存時間的長度，最後模擬與驗證中，本論文提出的方法可以在低網路流量時，減少約莫15%服務延遲。 With the further development of the Internet, most of the network bandwidth has been now occupied by high-speed and image-related services, which will be extended to more mobile devices. To optimize transmission performance and enable the content distribution capabilities, the Next Generation Network (NGN) will focus on ISP optimization. For this reason, the Named Data Networking (NDN) project was founded by the U.S. National Science Foundation (NSF). The NDN project aims to enable automatic caching of data and develop a name-based routing mechanism to reduce the throughput and accelerate response time. Based on NDN architecture, this paper proposes a QoS enhancement method for NDN: QoS classification is made based on types of data and the concept of Time To Live (TTL), which indicates whether or not an IP packet has stayed in the network too long and should be discarded, is adopted to replace conventional memory update scheme, like the the Least Recently Used (LRU) algorithm. Our proposed method classifies data from the source and defines different TTL of different data types in the router register to reduce the load of the Content Store (CS). Without increasing the service delay, our proposed method helps NDN to decrease the memory construction cost and reduce the service delay under low throughput conditions. The simulation results prove that our proposed method reduces approximately 15% of service delay under low throughput conditions.
