本篇論文提出一套兩階段的動態頻寬(Two Phase Bandwidth Allocation, TPBA)演算法並結合流量預測機制,應用於乙太被動式光纖網路(Ethernet Passive Optical Networks, EPONs)。此篇論文中所陳述之兩階段的頻寬分配方法,涵蓋有頻寬預測的能力及較公平的頻寬分配機制。TPBA演算法能利用兩階段頻寬分配的方式,改善傳統頻寬分配(Dynamic Bandwidth Allocation, DBA)無法有效利用頻寬的問題,達到提高整體乙太被動式光纖網路的骨幹系統效能為目的。此方法中,同時考量不同流量的傳輸等級,並根據其每個傳輸等級的不同特性來增強預測頻寬的正確性。最後,實驗的模擬結果是採用OPNET來呈現數據比較,經由頻寬分配、整體的輸出效能、端點對端點的平均延遲與佇列平均長度的比較結果顯示,TPBA的方法能有效使得乙太被動式光纖網路整體網路效能提升,達到本篇論文之預期的效果。
This study proposes a novel Two Phase Bandwidth Allocation (TPBA) mechanism incorporated with a prediction-based scheme in Ethernet passive optical networks (EPONs). Two phase dynamic bandwidth allocation mechanisms are designed in the proposed TPBA. In the first phase, the TPBA can resolve the idle time problem in traditional DBA mechanism. Furthermore, the TPBA can predict and allocate the excess bandwidth fairly and improve overall system in the next phase. The traffic characteristic in differential traffic class is also considered in the proposed prediction scheme to provide more accurate prediction. Simulation results using OPNET show that the proposed TPBA mechanism with adaptive prediction scheme performs better than well-known DBA methodologies in terms of wasted bandwidth, throughput, end-to-end delay and queue length.