標題: | 乙太被動光纖擷取網路之上鏈路排程機制 Uplink Scheduling Schemes in EPON Access Network |
作者: | 彭崇禎 Chun-Chen Peng 張仲儒 鄭瑞光 Chung-Ju Chang Ray-Guang Cheng 電信工程研究所 |
關鍵字: | 乙太被動光纖擷取網路;排程;公平性;EPON;scheduling;fairness |
公開日期: | 2003 |
摘要: | 寬頻服務在家庭及小型公司的需求激增已經成為擷取網路技術日益進步的主要因素。而如今,乙太被動光纖擷取網路(EPON)更被認為是下一代擷取網路的明日之星,原因是它整合了既有的光纖基礎建設以及目前已發展成熟並且成本低廉的乙太設備。除此之外,另一個讓乙太被動光纖網路備受注目的原因是它解決了寬頻服務所帶來對頻寬需求激增的問題,像是IP電話、隨選視訊(VoD)等服務無不催促著網路操作者致力於發展能提供全服務性質的擷取網路。在本篇論文中,我們提出了一個能適用於乙太被動光纖擷取網路的上鏈路排程方法,我們針對及時性的服務(例如語音服務、視訊服務等)設計了以延遲為考量的排程機制;另外,也針對非及時性的服務(例如一般資料服務)而設計了兩個以公平性作考量的排程機制,分別為Hybrid LQF-QLP機制以及Hybrid EQL-QLP機制。模擬的結果顯示我們所提出的排程演算法確實能夠讓語音封包的平均延遲限制在一個可接受的範圍之內,並同時對於資料服務考量了封包延遲與封包阻隔機率的公平性。
再者,我們提出了一個利用預測的方式來排程的機制,其中預測器是採用了移動平均(Moving Average)的技巧。我們發現當加入一個預測器之後,最大週期時間將可延長並且能提高系統的封包流量。模擬的結果顯示這種預測的方式的確能夠有效的提升系統的效能。 Rapid deployment of broadband services in the residential and small business area has played an important role in the evolution of access networks. Currently, Ethernet passive optical networks (EPON) are being considered as a promising solution for the next generation access network, due to the convergence of low-cost Ethernet equipment and low-cost of fiber infrastructure. In addition, the growing demand of broadband services such as IP telephony, video on demand has urged the network operator to accelerate the deployment of full-service access networks. In this thesis, we proposed a delay-considered scheduling scheme for real-time services, i.e. voice and video service, and two fairness-considered scheduling schemes, i.e. Hybrid LQF-QLP scheme and Hybrid EQL-QLP scheme, to support non-real-time data service. The goal of the scheduling algorithm is to meet the delay bound of voice service, and to simultaneously maintain the fairness of both packet delay and packet blocking probability for non-real-time data service. Simulation results show that the proposed scheduling method can meet our goal. In addition, we proposed a prediction-based scheduling method, in which we adopt a Moving Average technique. We find that by implementing a predictor, the maximum cycle time can be extended and the system throughput can be improved. Simulation results show that the proposed scenario can improve performance well. |
URI: | http://140.113.39.130/cdrfb3/record/nctu/#GT009113513 http://hdl.handle.net/11536/45968 |
顯示於類別: | 畢業論文 |