| Summary: | 第三代行動通訊系統雖然提供了更大的頻寬來因應未來的多媒體傳輸，然而多媒體應用對於服務品質要求嚴謹，不僅須較大的頻寬，更有延遲時間(delay)、跳動(jitter)及封包遺失率等限制。為了要讓頻寬的利用更有效率，UMTS依資料延遲特性之不同將資料型態分為conversational、Sreaming、interactive、background四類，並分別訂定其相關之服務品質限制。
本論文以基因演算法來做頻寬配置的最佳化，為改善基因演算法運算耗時之缺點，我們提出回饋式基因演算法以加快運算速度。在頻寬配置同時，為了要達到封包延遲時間之保証，我們以WFQ佇列機制所提供之延遲時間限制式來計算延遲時間。透過基因演算法及WFQ之結合，我們得以在滿足延遲時間限制下達到頻寬配置結果之最佳化。本論文也將服務品質滿意度及營運收益納入為最佳化目標之一，讓電信業者得以根據不同需求來改變系統之設定。
=== Third Generation Mobile Communication Systems (3G) provide higher bandwidth to fulfill applications of multimedia. However transmission of multimedia not only needs bandwidth but also delay, jitter and packet loss guaranteed. In order to satisfy all QoS constraints and improve the system utilization, UMTS classify the traffic into four classes - conversational, streaming, interactive and background.
This thesis uses genetic algorithm to optimize bandwidth allocation. But the genetic algorithm is a time consuming algorithm. So we propose feedback genetic algorithm to speed up computation. On the other hand, we adopt the delay constraint function of WFQ to compute and guarantee delay time. With feedback genetic algorithm and delay constraint function, we can optimize bandwidth allocation with delay guaranteed. Besides, we also take users satisfactions and operators profit into consideration.
|
|---|
