- 學位論文
於正交分頻多重進接網路中最大化處理能力與省電效益之下行頻寬分配最佳化之研究
Optimization of Downlink Bandwidth Assignment for Maximum Throughput and Energy-Efficiency in OFDMA Networks
摘要
由於移動設備的迅速發展，更多的用戶使用其行動設備隨時隨地觀看視頻。為滿足日益增長的用戶需求，有限的下行鏈路頻寬的利用率需要被提升。行動設備的另一個關鍵瓶頸是電池壽命。在新一代的正交分頻多重進接（OFDMA）無線系統中，如何分配的下行頻寬以產生最大吞吐量和能量效率已成為重要的挑戰。在這項研究中，我們分為三個階段分別研究下行頻寬分配問題以達到最大吞吐量與節能：1）channel diversity，2）power saving class，3） layer-encoding。 對於channel diversity，我們在全球互通微波存取(WiMAX)網路中探討視頻群播問題。因為WiMAX網絡的MAC層是連線導向，每個用戶的連接被建立之前都需要從基地台獲得頻寬。然而，網絡的效率會因為較差的傳輸通道品質而減少。而一個基地台分配子信道給用戶，信道衰落的差異常常影響傳輸速率。因此，將具有良好傳輸通道品質的子信道分配給用戶，以提高帶寬利用率。在依信道差異做群播信道分配最佳化的章節，我們提出了幾種傳輸調度方案，以避免信道差異的影響並改善傳輸性能。我們的策略是依用戶信號與噪聲的比率（SNR）將用戶分群組。目標是找到一個最優的調度，以減少視頻流的傳輸延遲和最大化的基地台的容量。模擬結果表示，我們提出的方法FF和FFRSG具有較高效率的群播調度的結果。 對於power saving class，我們研究了即時流量調度的QoS保證基於IEEE802.16e的WiMAX網絡的優化問題。每個移動站可以具有有特定的QoS要求，包括頻寬要求和延遲邊界的一個或多個即時串流。因為串流具有不同的到達率，所以當它不需要對接收或發送時，移動台可以關閉天線收發機制並進入睡眠模式以節省能源。為了最小化能量消耗，我們需要根據QoS要求用以高效率的排程所有串流。為了最小化收發訊框的數量，我們將排程問題數學化成Integer Linear Programming (ILP)模型。並且提出一個啟發式演算，稱為自適應帶寬預留（ABR）方法，用以降低計算複雜度。我們在論文中分析並證實了ABR算法在最壞情況是2倍解。我們的方法不僅保證QoS的即時流量，而且最大限度地減少移動台的能量消耗。所提出的方法對於移動台提供了對吞吐量和能量儲蓄有顯著改善。實驗結果表示，ABR算法與實驗比較對象有較優的節能效益，較高的頻寬利用率和較低的連線建立失敗率。 對於layer-encoding，我們研究最小化OFDMA的用戶站的無線系統的能量消耗的資源分配問題。當群播用戶接收以圖層(layer)為基礎的視頻編碼的視頻內容，不僅其頻寬要求必須得到保證，而且消耗的能量需要被最小化。因此，排程過程分為兩個階段。第一，為了優化耗電問題，我們提出Layer Determination Scheme For SVC(LDS)最小化要求時槽的數目。所提出的LDS方案選定要被傳送的圖層，並決定選定圖層的Modulation和分配可用時槽。在第二階段中，互補的匹配策略應用於提出一個互補匹配算法（Complementary Matching Algorithm, CMA）來最小化在OFDMA無線系統中active symbol的總數。通過LDS和CMA算法的分析，我們證實LDS和CMA算法在減少active symbol的近似比為（1+4/7）。在模擬中結果，所提出的LDS和CMA算法與比較對象相比，LDS和CMA算法在能源消耗和吞吐量方面都有較佳的性能。
並列摘要
Since mobile devices have developed rapidly, more users use their mobile devices to access video streams anytime and anywhere. To meet the needs of growing number of users, the utilization of limited downlink bandwidth needs to be enhanced. Another key bottleneck for mobile devices is battery life. In the new generation of orthogonal frequency division multiple access (OFDMA) wireless systems, how to allocate the downlink bandwidth with maximum throughput and energy efficiency has become an important challenge. In this study, we investigated in three phases the downlink bandwidth assignment issues for maximum throughput and energy saving : 1) channel diversity, 2) power saving class, and 3) layer-encoding. For the channel diversity phase, we consider the problem of video stream multicasting over worldwide interoperability for microwave access (WiMAX) networks. Since the MAC layer of WiMAX network is connection-oriented, each subscriber gains the bandwidth from the base station before connection is established. However, network efficiency may be reduced by poor channel quality. While a base station allocates subchannels to subscribers, the diversity of channel fading often affects the transmission rate. Therefore, the subchannels are allocated to subscribers with the good channel condition to enhance bandwidth utilization. In the optimization of multicast subchannel assignments with channel diversity chapter, several transmission scheduling schemes are proposed to avoid the effect of channel diversity and to improve the transmission performance. Our strategy is to divide subscribers into several groups depending on the signal-to-noise rate (SNR) of allocatable channels. The objective is to find an optimal schedule to minimize the transmission latency of the video stream and to maximize the capacity of the base stations. Our simulations show that the proposed methods FF and FFRSG yield efficient multicast scheduling and outperform previous results. For the power saving class phase, we study an optimization problem for real-time flow scheduling with QoS guarantee over IEEE 802.16e WiMAX networks. Each mobile station may have one or more real-time flows that have specific QoS requirements, including bandwidth requirement and delay bounds. Since flows have different arrival rates, the mobile station turns off the transceiver and enters sleep mode to save energy only when it does not need to either receive or send traffic. To minimize the energy consumption, we required the efficient scheduling of all flows under QoS requirements. Therefore, the scheduling problem is formulated as an Integer Linear Program (ILP) in order to minimize the total number of active frames to reduce energy consumption. A heuristic algorithm, called adaptive bandwidth reservation (ABR), is also proposed to improve the computing efficiency. The approximation factor of 2 is proved in the worst case analysis of the ABR algorithm. Our approaches not only guarantee the QoS for real-time flows, but also minimize energy consumption of mobile stations. The proposed approach provides a significant improvement on throughput and energy saving for all mobile stations. The experiment results demonstrate that the ABR algorithm outperforms the previous approaches in terms of energy saving, bandwidth utilization and drop rate. For the layer-encoded phase, we study a resource allocation problem for minimizing the energy consumption of subscriber stations in Orthogonal Frequency Division Multiplexing Access (OFDMA) wireless systems. Not only does the requirement have to be guaranteed when multicast users receive video content in layer-based video coding, but also the energy consumption needs to be minimized. Hence, the scheduling process is divided into two phases. First, the number of requested tiles is minimized by the proposed scheme Layer Determination Scheme For SVC (LDS) for the issue of optimizing energy consumption. The proposed LDS scheme determines the modulation of the selected layer and assigns available tiles to the selected layer. In the second phase, a complementary matching strategy is applied to propose a Complementary Matching Algorithm (CMA) to minimize the total number of active symbols in an OFDMA wireless system. To find the complementary subset for a specific request, the problems of minimizing the total number of active symbols is reduced to the subset sum optimization problem. Through the analysis of the LDS and CMA algorithm, we demonstrate that the approximation ratio of the proposed LDS and CMA algorithm is (1+4/7 ) for minimizing the total active symbols in OFDMA wireless systems. In the simulation results, the proposed LDS and CMA algorithm were found to have better performance in terms of energy consumption and throughput.