謝宏昀臺灣大學:電信工程學研究所陳明芳Chen, Ming-FangMing-FangChen2010-07-012018-07-052010-07-012018-07-052008U0001-1308200816282100http://ntur.lib.ntu.edu.tw//handle/246246/188222對於行動裝置而言，如何省電是一個重要的議題；在IEEE 802.16e標準中，也特別針對行動裝置訂定了第一類、第二類及第三類等三種不同省電機制。這三種省電機制各具有不同的特性及參數，以適用在不同的流量模型 (traffic model) 下達到省電之目的。為了研究各省電機制在不同參數或是流量類型下的效能表現，許多相關的研究都提出了不同的分析模型。在這些相關的研究當中，大部分都假設網路流量為Poisson或是CBR流量模型。然而以這樣的假設所分析的省電效能模型，在真實的網路環境與流量模型下，將無法準確地表現出802.16e省電機制的效能。因此，在本論文中我們首先針對802.16e第一類省電機制提出一個可廣泛適用在不同流量模型下的分析模型，包含Poisson、Pareto以及4IPP等流量模型。其中，4IPP流量模型是WiMAX標準所推薦用來模擬HTTP/FTP資料流量之模型。本論文所提出之分析模型除了可更精準地分析802.16e省電機制在不同流量模型下的效能之外，亦可推廣至第二類省電機制之效能分析。過分析模型的研究，我們歸納出不同的省電參數對於行動裝置省電效能之特性與影響。我們發現雖然省電機制可延長移動式裝置供電時間，但也因此造成封包傳送時間的延遲。此外，封包傳送延遲時間的長短亦與省電機制中多項參數有關，包括最小、最大睡眠時間及網路流量之負載。因此，為了控制封包之延遲時間，必須能依據不同的網路流量特性去改變省電參數之設定，以能動態在省電效能與封包延遲取得平衡點。基於我們所提出之省電機制分析模型，在本論文中，我們設計了一可適性省電演算法。此演算法藉由即時觀察網路流量統計特性來求得最佳之省電參數設定，使得行動裝置之省電機制可隨著網路特性之變化而調整參數，以控制封包傳送延遲時間，達到最佳之省電效能。模擬結果顯示，我們設計之可適性省電演算法確能比相關文獻上的方法達到更好的省電效率，且有較佳的適用性與彈性。Power saving is an important issue for mobile stations (MSs). IEEE 802.16e defines three types of power saving classes (PSCs) for supporting the sleep mode operations on MSs with different types of traffic. Related work has developed analytical models to evaluate the performance of power saving operations. Most of them employ Poisson process or CBR as the traffic model, and hence their capability is limited in capturing the characteristics of realistic traffic. In this thesis, we first propose a generic analytic model for capturing the behaviors of IEEE 802.16 sleep mode operations under arbitrary traffic distribution, including Pareto and 4IPP traffic models for describing the characteristic of HTTP/FTP data traffic. While we focus primarily on the operation of PSC of type I, we also show that the proposed model can be extended to PSC of type II. Simulation results show that the proposed model has better flexibility and can achieve higher accuracy compared with existing models.hile power saving operation can prolong the lifetime of MSs, one significant tradeoff is that they may potentially increase the packet transmission delay. Based on the proposed analytical model, we observe that the delay depends on setting of minimal/maximal sleep window size. If we can dynamically change the size of window depending on the traffic conditions, we can avoid delay from increasing with varying traffic distributions. To address the tradeoff between energy efficiency and packet delay, we propose an adaptive power saving algorithm for maintaining packet delay on MSs under different traffic loads. Through on-line observing the distribution of arriving packets, the algorithm can determine the optimal power saving parameter setting and adaptively change them according to varying traffic conditions. Simulations show that the proposed algorithm can indeed achieve minimal energy consumption and satisfy the desired delay constraint.ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viHAPTER 1 INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1HAPTER 2 BACKGROUND . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1 Sleep Mode Operations in IEEE 802.16e . . . . . . . . . . . . . . . . . . . . . . . . 5.1.1 Power saving class of type I . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1.2 Power saving classes of type II . . . . . . . . . . . . . . . . . . . . . . . . . . 8.1.3 Power saving classes of type III . . . . . . . . . . . . . . . . . . . . . . . . . . 10.2 Traffic Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.2.1 Poisson traffic model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2.2 Pareto traffic model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2.3 4IPP traffic model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.3 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17HAPTER 3 ANALYTICAL MODEL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20.1 Analytical Model for Power Saving Class of Type I . . . . . . . . . . . . . . . . . . 20.2 Numerical and Simulation Results for Poisson Traffic Model . . . . . . . . . . . . . . 26.2.1 Comparison with other models . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27.2.2 Simulation results of static operation . . . . . . . . . . . . . . . . . . . . . . . 31.3 Numerical and Simulation Results for Pareto Traffic Model . . . . . . . . . . . . . . 35.3.1 Analytical model for Pareto traffic model . . . . . . . . . . . . . . . . . . . . . 35.3.2 Static simulation for Pareto traffic model . . . . . . . . . . . . . . . . . . . . . 38.3.3 Comparison with power saving performance under Poisson and Pareto traffic model . . 40.4 Numerical and Simulation Results for 4IPP Traffic Model . . . . . . . . . . . . . . . 42.4.1 Impact of previous packet delay Dp . . . . . . . . . . . . . . . . . . . . . . . . . 42.4.2 Simulation results of static operation . . . . . . . . . . . . . . . . . . . . . . . 44.5 Analytical Model for Power Saving Class of Type II . . . . . . . . . . . . . . . . . . 49.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54HAPTER 4 DESIGN OF ADAPTIVE POWER SAVING ALGORITHMS . . . . . . . . . . . . . . . . . . . 58.1 Problem Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58.2 Design of Adaptive Power Saving Algorithms . . . . . . . . . . . . . . . . . . . . . . 61HAPTER 5 PERFORMANCE EVALUATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69.1 Evaluation of Historical Length . . . . . . . . . . . . . . . . . . . . . . . . . . . 69.2 Performance Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73.3 Adaptability Verification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75.4 Comparison with Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77HAPTER 6 CONCLUSIONS AND FUTURE WORK . . . . . . . . . . . . . . . . . . . . . . . . . . 84PPENDIX A 4IPP PARAMETRIC MODEL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86PPENDIX B MARKOV-MODULATED POISSON PROCESS . . . . . . . . . . . . . . . . . . . . . . . 90EFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 982123519 bytesapplication/pdfen-US802.16省電機制802.16 power saving[SDGs]SDG7thesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/188222/1/ntu-97-R95942103-1.pdf