楊佳玲Yang, Chia-Lin臺灣大學:資訊工程學研究所胡森博Hu, Sen-PoSen-PoHu2010-06-022018-07-052010-06-022018-07-052008U0001-0502200814135700http://ntur.lib.ntu.edu.tw//handle/246246/184900目前在現代的行動多媒體系統(Mobile Multimedia System)上，如何在電力消耗(Power Consumption)與系統效能(Performance)達到完美的平衡一直是個受關注的議題。一個用來量測行動多媒體系統用電效率的新標準叫”品質能量比(QE Ratio)” 在本篇論文中被提出，一個有較高品質能量比的系統表示每消耗一單位的電池電量可以提供較多的服務質量(QoS)。了在行動裝置上提升品質能量比，我們設計了一個低功率視訊串流嵌入式系統，結合動態電壓與頻率調節(DVFS)省電技術以及動態影像品質調整(Dynamic Quality Adaptation)。動態電壓與頻率調節技術利用現代中央處理器內部的效能監控單元，及時分析影片撥放程式的處理運算量，調整中央處理器的電壓與頻率而達成省電效果。另一方面，動態影像品質調整則是每間隔一段時間去動態調整影片編碼器的品質參數。個低功率視訊串流嵌入式系統已經被實做在一個以Intel Xscale處理器為基礎的嵌入式實驗評估板，實驗數據顯示了我們的系統在行動裝置上改善10%的品質能量比。Optimal balance between power consumption and performance is a great challenge for mobile multimedia system. In this thesis, a new metric called QE ratio is proposed to evaluate the energy efficiency of a low-power mobile multimedia system. A system with higher QE ratio represents more qualityf service is provided while consuming per unit battery energy. To achieve high QE ratio on mobile device, I construct a low-power video streaming system which combines a frame-based DVFS technique and dynamic quality adaption. The DVFS technique changes CPU speed according to frame-basedorkload prediction by utilizing performance counters in modern microprocessor. With effective DVFS techinque, CPU power consumption can be reduced so as to improve QE ratio. Besides, dynamic quality adaptation is proposed to improve QE ratio by periodically adjusting the values of encoding quality parameters for video bitstream. The proposed low-power video streaming system has been implemented on an Xscale-based embedded platform. The experimental result shows the proposed system gains at least 10% improvement of QE ratio on mobile device.Abstract i Introduction 1 Related Work 4 Low-Power Video Streaming System 9.1 QE Ratio . . . . . . . . . . . . . . . . . . . . 9.1.1 Objective Video Quality . .. . . . . . . . . . 10.1.2 Subjective Visual Quality .. . . . . . . . . . 11.2 Proposed System Overview . . . . . . . . . . . . 13.3 Dynamic Quality Adaptation . . . . . . . . . . . 14.3.1 Problem Formulation . . . . . . . . . . . . . 17.3.2 Proposed Algorithm . . . . . . . . . . . . . . 18.4 Frame-based DVFS . . . . . . . . . . . . . . . . 20.4.1 CPU Workload Predictor . . . . . . . . . . . . 21.4.2 System Performance Model . . . . . . . . . . . 22 Implementation 23.1 Frame-based DVFS . . . . . . . . . . . . . . . . 24.1.1 Workload Recording . . . . . . . . . . . . . . 25.1.2 CPU Speed Scaling . . . . . . . . . . . . . . 25 Experimental Result 27 Conclusion 30ibliography 32application/pdf754721 bytesapplication/pdfen-US視訊串流動態電壓與頻率調節品質能量比影像品質行動裝置嵌入式系統Video StreamingDVFSQE RatioImage QualityMobile DeviceEmbedded System[SDGs]SDG7thesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/184900/1/ntu-97-R94922030-1.pdf