研究生: |
陳純美 |
---|---|
論文名稱: |
延遲敏感和損失敏感的訊務的GPS方式排程器 An Integrated GPS-based Scheduler for Time-sensitive and Loss-sensitive Traffic |
指導教授: |
蔡榮宗
Tsai, Jung-Tsung |
學位類別: |
碩士 Master |
系所名稱: |
資訊教育研究所 Graduate Institute of Information and Computer Education |
論文出版年: | 2000 |
畢業學年度: | 88 |
語文別: | 英文 |
論文頁數: | 45 |
中文關鍵詞: | 服務品質 |
英文關鍵詞: | QoS, GPS |
論文種類: | 學術論文 |
相關次數: | 點閱:122 下載:7 |
分享至: |
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報 |
在未來高速網路中將可以針對擁有不同服務品質(Quality of Service, QoS)的傳輸要求提供服務品質的保證,而通常是利用在網路的節點中加上一個排程器(scheduler)來保證這些服務品質。GPS(Generalized Process Sharing)方式是流體模式(fluid model)下一個理想的排程器,因為它擁有某些特性:包含保證每個連線(connection)在每個網路節點中能得到一定的最小傳輸速度且不受其他連結情況的影響。另外,使用GPS的方式,一個連結的訊務(traffic)如果受到漏水式水桶(leaky bucket)的限制,則它的端點到端點間(end-to-end)的最長傳輸延遲可以用一個簡易的算式計算出來。然而在GPS中,每個連結所得到的權重(weight)是固定的,所得到的傳輸速度也成為固定比例的模式,並沒有針對目前網路狀況調整封包排程的機制,因此不適用於有延遲保證要求的即時性資料傳輸。在我們所提出的系統中,網路的訊務交通被分成兩大類,一類是延遲敏感(delay-sensitive);一類是損失敏感(loss-sensitive)。在有限的緩衝器(buffer) 大小限制下,我們使用具有即時調整機制的排程演算法來設定兩種訊務型態在GPS中的權重以避免緩衝器滿溢(overflow)。和GPS不同的是,在我們所提出的排程演算法中,延遲敏感連線的個別延遲要求為排定封包傳送順序的主要考量。最後,我們利用模擬程式來探討所提出的排程演算法中,各項參數對系統所造成的影響。
Future high-speed networks are expected to support guarantees of QoS requirements for different traffic types. The approach to QoS guarantee is to embed schedulers in networks. One of the schedulers is GPS which is an idealized fluid discipline with a number of very desirable properties, including the provision of minimum bandwidth guarantees to each connection regardless of the behavior of other connections, and the provision of deterministic, easily-computable end-to-end delay bounds to connections whose traffic is leaky-bucket constrained. It, however, has some inherent limitations for delay-sensitive traffic, which has strict delay requirements. Furthermore, GPS is unable to solve the problem of the strong correlation between rate and delay without state dependent adjustment during service. It is thus not recommended to use GPS for real-time service because of lacking independent hard guarantees of individual delay. This thesis, however, proposes an adaptive GPS weight assignment scheme which incorporates two traffic types: delay-sensitive and loss-sensitive. Under the assumption of finite buffer for each traffic type, this scheme is mainly to prevent the buffer for loss-sensitive traffic from overflow. Individual delay requirement of each delay-sensitive connection is also considered in our scheduler. Results from simulations for the two traffic types under our system are shown and discussed.
[1] H. Zhang, “Service disciplines for guaranteed performance service in packet-switching networks,” Proceedings of the IEEE, vol. 83, no. 10, pp. 1373-1396, October 1995.
[2] J. C. R. Bennett and H. Zhang, “Hierarchical packet fair queueing algorithms,” IEEE/ACM Trans. Networking, vol. 5, no. 5, pp. 675-689, October 1997.
[3] A. K. Parekh and R. G. Gallager, “A generalized processor sharing approach to flow control in integrated services networks: the single node case,” IEEE/ACM Trans. Networking, vol. 1, no. 3, pp. 344-357, June 1993.
.
.
.
.
.
.