- 學位論文
支援多速率訊務無線區域網路之產能最佳化
Throughput Optimization in Wireless Local Area Networks to Support Multi-Rate Traffic
摘要
在無線區域網路環境中，當行動裝置(Mobile station)與無線基地臺的距離相對遙遠時，傳輸訊號將受到環境干擾而衰減，使得傳輸訊號品質愈來愈差。此時，行動裝置會藉由調整其調變方式，也就是調降其傳輸速率，來克服此問題。當無線區域網路處於多種傳輸速率並存之狀況時，使用傳輸速率較低的行動裝置將會使得其它使用高速率傳輸的行動裝置的產能下降至與低速相同。之所以會有此異常現象，在於無線區域網路中，各行動裝置擁有相同存取媒體的機率。當使用較慢速的無線裝置取得存取媒體的使用權時，由於其傳輸速率較慢，所以必定將佔據較多的時間來傳送資料，因此將使得使用較高速的行動裝置產能降低。 在本論文中，我們除了分析此異常問題外，同時提出兩個解決方案來解決此問題，而這兩方案主要是籍由使用時間公平性及產能公平性的觀點來解決。我們透過動態調整不同傳輸速率行動裝置之競爭視窗(Contention window)及傳送封包長度來達成公平性要求。同時提出兩個非線性整數規劃模型來調整相關參數，使得系統能夠達到產能最佳化同時滿足時間及產能公平性的要求。 基本上，我們所提出的方式改善了系統的效能同時也解決了效能異常的問題。我們將這兩個非線性整數規劃問題變成非線性實數規劃問題，同時使用 penalty method 搭配 gradient-based 的方法來解決此問題。在本論文中，我們做了一些實驗來展現此解決方案的效能。另外也對一些相關參數的調整做了適當的建議。
並列摘要
In an infrastructure configuration of Wireless Local Area Networks (WLAN), a mobile station that is far away from the access point (AP) will suffer from signal fading and interference. Hence the mobile station changes it modulation scheme, i.e., it degrades its data rate, to accommodate the fading environment. In such case, a mobile station with lower bit rate will reduce the throughput of all other transmitting stations with higher bit rates since all mobile stations have the same priority to access the wireless medium. When a mobile station with lower bit seize the wireless medium, it will occupy the wireless medium a long time because of its low bit rate and hence penalize other mobile stations with higher bit rates. In this paper, we analyze the anomaly problem and propose a solution based on two criteria: time fairness index and throughput fairness index. We dynamically assign different sizes of minimum contention window and transmitted payload lengths to stations with different data rates. Two nonlinear and integer programming problems are formulated to optimal the total system throughput in the constraints of time fairness index and throughput fairness index. Basically, the solution improves the system performance and avoids the anomaly problem. We relax theses two problems and provide solutions based on penalty method with gradient-based approach. Some experiments are demonstrated to show the effectiveness of the proposed solutions. The guidelines for adjusting the related parameters are also presented in the paper.