透過您的圖書館登入
IP:18.225.72.104
  • 學位論文

在軟體定義網路中負載平衡的影音群播系統

A Load-Balanced Video Multicast Routing System in Software-Defined Networks

指導教授 : 徐正炘

摘要


由於傳統的群播技術(IP multicast)需要高效能的路由設備以及許多 複雜的設定, 對於網路營運商而言,傳統的群播技術成效不彰。 在 本篇論文中,我們實做了一個基於SDN架構的群播路由系統。針對 群播路徑的建立, 我們討論影音串流的多路徑群播樹於軟體定義網 路(Software-Defined Networks)的問題。我們的目標是建立一個健全、 負載平衡、具可適性且相容於軟體定義網路的影音群播路由。 我們 將此群播路由問題轉換為一個最佳化的數學問題,得出一個針對鏈結 負載的min-max最佳解,並取名此演算法為Robust Multipath Multicast Routing (RMMR*),但此演算法為了求得最佳解而需耗時較久,所以 我們另外設計了一較有效率的啟發式演算法。 我們將這兩個演算法 實作在群播路由系統並安裝在OpenFlow控制器中。 我們在真實系統架 設的測試平台以及利用Mininet模擬器擬真出的網路中分別做實驗, 測 試群播系統的可行性、效能以及可擴展性。 在測試平台的環境中, 我們測量到的系統對於群播流的設置時間都小於5毫秒, 而對於群播 收看者的偵測時間皆小於0.15秒。 在模擬器中所做的實驗,以下幾點 能發現我們的演算法更優於IP Multicast: (i) 降低了19%到95%的幀遺失 率、 (ii) 影像品質提高了4 dB到15 dB、 (iii) 影片收看者的輸送量增加 了25%到66%、 (iv) 最大鏈結使用率降低了15%到50%。 另外我們也權 衡了我們設計的兩個演算法的最佳性以及所費時間,找出他們所適合 的網路環境。 最佳解演算法較適合比較小而且更穩定的網路環境, 而 啟發式演算法則較適合使用於更大且常有變動的網路環境中。

並列摘要


IP multicast in traditional networks, dictates high-end routers and incurs high administrative overhead, which is no longer suitable for deployment due to its complicated operations. In this thesis, we implement a multicast routing system based on SDN framework. For computing the multicast routes, we study the problem of establishing multipath multicast routing for streaming videos in Software-Defined Networks (SDNs). The objectives of the considered problem are robustness, load balance, adaptiveness, and SDN compatibility. We formulate the multicast routing problem into a mathematical optimization problem and propose a min-max link load multicast routing algorithm, called Robust Multipath Multicast Routing (RMMR∗). We further design a heuristic algorithm to obtain the multicast trees efficiently. We implement our proposed algorithm in our multicast routing system on OpenFlow controller. We conduct the experiments in real testbed and Mininet emulator to demonstrate the practicality, performance and scalability. In the experiment of real testbed, we measure the response time of our system: (i) all operations of flow-entries insertion are completed in less than 5 milliseconds, (ii) the detection time of all clients are no more than 0.15 second. The results of experiment in emulator show the merits of our algorithms over the IP multicast, e.g., we observe: (i) frame loss rate reduction between 19% and 95%, (ii) video quality improvement between 4 dB and 15 dB, (iii) sink throughput increase between 25% and 66%, and (iv) maximal link utilization reduction between 15% and 50%. We also show the tradeoff between optimality and run time of the two proposed algorithms: one of them is more suitable for smaller and more static networks, and the other one is more suitable for larger and more dynamic networks.

參考文獻


[9]  A. De Gante, M. Aslan, and A. Matrawy. Smart wireless sensor network management based on software-defined networking. In Communications (QBSC), 2014 27th Biennial Symposium on, pages 71–75. IEEE, 2014. 

[10]  E. W. Dijkstra. A note on two problems in connexion with graphs. Numerische mathematik, 1(1):269–271, 1959. 

[11]  C. Diot, B. Levine, B. Lyles, H. Kassem, and D. Balensiefen. Deployment issues for the ip multicast service and architecture. Network, IEEE, 14(1):78–88, Jan 2000. 

[12]  W. C. Fenner. Internet group management protocol, version 2. 1997. 

[14]  N. Freris, C. Hsu, J. Singh, and X. Zhu. Distortion-aware scalable video streaming to multi-network clients. IEEE/ACM Transactions on Networking, 21(2):469–481, April 2013. 


延伸閱讀