全球微波存取互通介面(WiMAX)是目前無線網路中可以實現廣泛的傳輸距離以及高速的接取服務之技術，其中於IEEE 802.16e標準訂定了使用者移動的規範與支援，而於IEEE 802.16m中更開啟了使用者於高速移動時可以高速傳輸的大門，使之成為目前最引人矚目之無線網路技術。然而WiMAX雖具有上述之特性，但使用者移動或者環境改變時，仍會有由於頻道狀況改變而造成資料傳輸率下降之問題。目前的研究指出，對於此問題可以提出可調適調變的技術來解決，當使用者與基地台之間的頻道狀況差時，可以使用較為強健的調變來做傳輸；而當頻道狀況佳時，則可以使用傳輸速率較快的調變來做傳輸，然而目前的研究對於如何預測頻道之品質並沒有加以探討，或者是利用使用者目前以及之前頻道的狀況來做為頻道未來品質之預測，前者將會使得當使用者回報頻道品質時，可能會造成資訊的延遲而產生不適當的調變調整，後者將會由於環境的改變，導致預測之不準確，以致於傳輸錯誤率提昇。本篇論文提出一個頻道預測演算法來做為改進，將使用者之移動速率以及方向也包含在預測所需的資訊之中，並且使用多天線之技術(MIMO)來收集周圍使用者所量測到的頻道訊號強度。藉由使用者移動之資訊可以獲得下一時間點使用者之確實位置，而收集周圍基地台所量測得的頻道品質可以作為頻道預測之演算法的資訊，即可預測出使用者於下一時間點所在位置之頻道品質，接著即可調整調變的方式，來達到傳輸效率的改善。 本篇論文透過QualNet來模擬，模擬之結果顯示使用所提出之演算法會使得總吞吐量改善17%、延遲獲得15%的改善，也代表了預測之準確性有明顯的提昇。 Worldwide Interoperability for Microwave Access (WiMAX) can realize the long distance transmission and high speed access in current wireless network. IEEE 802.16e standard specifies the support of user mobility. IEEE 802.16m standard specifies the high speed transmission under high mobility. Although WiMAX has above characteristics, the transmission rate decreases when a user moves to another location or when the radio environment changes due to different channel quality in a cell area. Adaptive modulation is used to address the problem caused by channel quality variation. Adapting a robust modulation to transmit data when the channel quality becomes worse and high speed modulation when the channel quality becomes better can reduce the variance of the delay and packet loss of the network. However, it takes time to measure the channel quality, to determine a new modulation, to propagate the information of the new modulation and to adapt to a new modulation. Channel quality prediction is thus used to decide new modulation in advance. The thesis proposed a novel channel prediction method to improve the precision of the channel prediction. The method takes the position and speed of a user into account and collects the signal strengths measured by neighboring users by multi-input multi-output (MIMO) technology. According to the user position and speed, the future position can be estimated. The collected signal strengths of the channel from other users can be used for prediction, and the channel quality of the user future position can be predicted and the modulation can be adjusted in advance. Therefore, the variation of packet loss and delay and the jitter can be reduced, and the network quality is thus improved. Simulations are done with QualNet to evaluate the proposed method. The results of the simulations indicate the improvement of 17% throughput and 15% delay time. Therefore the accuracy of the channel quality prediction has been improved.
