陳健輝臺灣大學：資訊工程學研究所朱玄義Chu, Hsiuen-YiHsiuen-YiChu2007-11-262018-07-052007-11-262018-07-052005http://ntur.lib.ntu.edu.tw//handle/246246/53681隨著無線網路技術的進步與蓬勃發展, 許多手持式的個人無線上網裝置越來越普及. 這些手持式的裝置大都配備有計算能力越來越強大中央處理器, 更大的主記憶體, 及彩色螢幕. 這些技術的進步還有廣為佈建無線網路基礎架構, 使得要求能夠隨時隨地都能取得無線上網資源的需求越來越強烈. 但是由於無線網路上的干擾問題, 使得無線網路的實際傳輸量往往因為重傳過於頻繁, 而急遽下降. 有鑑於此, 許多動態調整封包大小的演算法都曾被研究過. 他們根據過往傳送的成功率, 來決定接下來的封包大小. 在本篇論文中, 我們認為以訊噪比作為網路品質指標來做封包大小的決定, 會有更精確的效果, 更能有效提昇傳輸量. 我們提出一個根據訊噪比來動態調整封包大小及傳輸速率的演算法來提昇傳輸量. 在我們的實驗中, 以訊噪比依據的演算法可以比依據過往傳輸量的演算法, 提昇最高達 14.90%.With the successful and widespreading wireless technology, many hand-held devices may feature powerful CPUs, large main memories and disks, color displays, and multimedia capabilities. It enables the vision of pervasive ubiquitous computing where users have network access anytime, anywhere. However, in wireless environment, due to existences of noises and interferences, frame errors occurs quite often which leading to a large number of retransmissions. As a result, the good throughput performance suffers. Lagecy frame length adjustment algorithm tend to use the history of failure/sucess transmission to make frame length decision. We argue that the transmission history cannot reponse the channel quality properly. In this paper, we propose a SNR-Based frame adjustment algorithm to improve the throughputs. Our simulation shows that the SNR based algorithm has better throughputs than legacy history based algorithms.Table of content 1 Introduction ... 4 1.1 Frame length adjustment . . . . . . . . . . . . . . . . . . . . . 4 1.2 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 2 Overview of IEEE 802.11 ... 8 2.1 DCF of IEEE 802.11 MAC . . . . . . . . . . . . . . . . . . . . 8 2.2 MAC/PHY Overhead Analysis . . . . . . . . . . . . . . . . . 10 2.3 Performance Analysis of DCF . . . . . . . . . . . . . . . . . . 11 3 Related Works ... 14 3.1 MAC Layer Frame Length Adaptation . . . . . . . . . . . . . 15 3.2 Heuristic Frame Size Adjustment . . . . . . . . . . . . . . . . 18 3.3 Signal-to-Noise Ratio versus Bit Error Rate . . . . . . . . . . 20 4 SNR Based frame length adjustment ... 22 4.1 System Architecture . . . . . . . . . . . . . . . . . . . . . . . 23 4.2 SNR Estimator . . . . . . . . . . . . . . . . . . . . . . . . . . 24 4.3 Frame Size & Data Rate Selector . . . . . . . . . . . . . . . . 25 5 Simulation ... 31 6 Conclusion ... 37 List of figures List of Figures 1 Basic CSMA/CA operation. . . . . . . . . . . . . . . . . . . . 8 2 Timing of successful frame transmissions under the DCF. . . . 9 3 Timing of frame retransmission due to Ack failure. . . . . . . 9 4 Timing of frame retransmission due to an erroneous data framereception. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 5 Frame format of a data frame. . . . . . . . . . . . . . . . . . . 10 6 Frame format of a Ack frame. . . . . . . . . . . . . . . . . . . 11 7 Frame format of a Ack frame. . . . . . . . . . . . . . . . . . . 11 8 PHY and MAC overheads. . . . . . . . . . . . . . . . . . . . . 16 9 Throughput and Frame Length . . . . . . . . . . . . . . . . . 17 10 Basic Access Method of MAC Fragmentation . . . . . . . . . . 18 11 Intersil HFA3861B chip measured under AWGN environment . 20 12 Design strategy . . . . . . . . . . . . . . . . . . . . . . . . . . 23 13 System architecture of proposed link adapataion . . . . . . . . 24 14 Maximum throughput vs. SNR . . . . . . . . . . . . . . . . . 26 15 Maximum throughput vs. Frame Size using 5.5 Mbps data rate 26 16 Maximum throughput vs. Frame Size using 11 Mbps data rate 27 17 Finite state machine of frame size and data rate selection . . . 28 18 Flow chart of transmission . . . . . . . . . . . . . . . . . . . . 30 19 The measurement platform for our simulation input. . . . . . 32 20 Scenario I: SNR trace record, measured in living room, PER= 27.66%, min = 1.0, ave = 11.2, max = 21.0, stdev = 3.47 . 34 21 Result of Scenario I . . . . . . . . . . . . . . . . . . . . . . . . 35 22 Scenario II: SNR trace record, measured in living room(2), PER = 57.55%, min = -1.0, ave = 8.2, max = 16.0, stdev =3.09 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 23 Result of Scenario II . . . . . . . . . . . . . . . . . . . . . . . 37 24 Scenario III: SNR trace record, measured in kitchen, PER = 77.72%, min = 0.0, ave = 9.1, max = 21.0, stdev = 3.15 . . . 38 25 Result of Scenario III . . . . . . . . . . . . . . . . . . . . . . . 391164691 bytesapplication/pdfen-US訊噪比訊框調適SNRframe adjustmentthesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/53681/1/ntu-94-P91922005-1.pdf