林宗男臺灣大學：電信工程學研究所陳路儉Chen, Lu-ChienLu-ChienChen2007-11-272018-07-052007-11-272018-07-052007http://ntur.lib.ntu.edu.tw//handle/246246/58689無線網路在今日的世界中已經非常流行了.許多人經由無線網路來上網,接收電子邮件或是玩線上遊戲.大部分的無線設備都是根據美國電子電機協會所製定的無線網路通訊標準802.11分散式協調功能來製造的.但是在分散式的無線網路中,存取點和每個無線工作平台對無線通道的存取的機會是相同的.這個特性會造成無線網路中上傳和下傳非常嚴重的不公平性,特別是在當無線工作平台增加時.在這篇論文中我們提出一個非常有力的解決方式就是利用類神經網路來達成無線網路上傳和下傳的公平性.我們的方法很容易可以應用在無線網路的存取點,而且不用修改802.11無線網路通訊標準.我們的模擬結果証明我們的方法不僅可以達成無線網路上傳和下傳的公平性,還可以根據無線環境的變化來動態調整上傳和下傳的比例.Today Wireless Local Area Network (WLAN) is very popular. Many people go to Internet, receive E-mail or play on-line game through WLAN. Most of WLAN devices are based on the IEEE 802.11 Distributed Coordination Function (DCF) [1]. However in Institute of Electrical and Electronics Engineers (IEEE) 802.11 DCF-based infrastructure WLAN, the opportunity of accessing wireless channel is the same for each mobile station and access pointer (AP). This feature will cause very serious unfairness problem between the traffic flow of the uplink and downlink, especially when the number of stations increases. In this paper we propose a powerful solution that using neuron network (NN) to solve this problem. Our solution is easy to implement at the AP side without the modification of the IEEE 802.11 standard. The simulation results indicate that our proposed method is not only to achieve fairness between the uplink and downlink traffic flow, but also to dynamic adjust the ratio between uplink and downlink according to the WLAN environment changing.Abstract--------------------------------------------------i Acknowledgments-----------------------------------------iii I.INTRODUCTION--------------------------------------------1 II.BACKGROUD DESCRIPTION----------------------------------5 2.1 Overview the families of IEEE 802.11------------------5 2.2 802.11 MAC Mechanism---------------------------------10 2.3 802.11 WLAN Configuration----------------------------16 III. IEEE 802.11 DCF THROUGHPUT--------------------------19 3.1 Markov chain analytic model with idea channel--------20 3.2 Markov chain analytic model with error prone channel-23 IV. PROBLEM INDENTIFY------------------------------------28 V. NEURAL NETWORKS---------------------------------------32 5.1 Cost-reward function---------------------------------32 5.2 Learning the nonlinear function----------------------34 5.3 Adjusting the back-off parameters--------------------36 5.4 Adaptive NN algorithm--------------------------------38 VI. SIMULATION RESULTS-----------------------------------42 6.1Analytic simulation results---------------------------42 6.2 NS-2 simulation results------------------------------57 VII. CONCLUSION------------------------------------------65 REFERENCES-----------------------------------------------661020667 bytesapplication/pdfen-US無線網路上傳下傳公平性類神經網路WLANuplinkdownlinkfairnessneuron networkthesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/58689/1/ntu-96-P93942011-1.pdf