基於動態預測模型之手持式串流播放裝置省電設計

Low power design in handheld media streaming playback devices based on a dynamic prediction model

蔡明諺(Min-Yen Tsai)

指導教授 ： 梁文耀

國立臺北科技大學/電資學院/資訊工程系研究所/碩士(2011年)

https://doi.org/10.6841/NTUT.2011.00226

摘要

本文主要針對手持式串流媒體播放裝置提出低功耗的電源管理設計，在之前的研究中涵括對緩衝區容量的分析以及傳輸協定,如時間片段(Time Slicing)所作省電設計。有別於前者，我們的考量點是從系統層面著手，藉由動態預測下一個時間點的功耗狀況，使用動態電壓調整達到降低功耗的目的。

為了動態的預測功耗狀況，需要一模型協助我們找出裝置狀態與電壓頻率之間的關係。為了建構此模型，我們先使用軟體預估的方式。將作業系統所提供的參數，建立回歸模型估算出裝置運作時所消耗的功率。再將媒體撥放時的狀態，其中包含功率消耗、幀率、工作負載與網路使用的資料建立成表格，並利用此表格評估撥放時下一個時間區段的狀態。最後將執行狀況與功率消耗關係利用回饋修正的方式映射到適當的電壓頻率，並進行調整使系統維持最適合的功率消耗，以達到Low Power目的。

關鍵字

串流媒體撥放 ； 動態電壓調整 ； 動態預測模型

並列摘要

In this thesis, we focus on low power design of handheld media streaming playback devices. The design issues of the research include buffer size analysis and transmission protocol. Unlike previous works, our approach is to address the problem at system level. By predicting power consumption for the next interval during playback, reduction in power consumption could be achieved by using DVFS.

In order to predict power consumption dynamically, a predict model is needed which help us find the device status and the relationship between voltage and frequency. To construct the model, the software estimation method which captures operating system’s parameters as regression model is used to estimate device power consumption at first.

Then, we record the state of media playback which includes power consumption, frame rate, workload, and network status as a table and use it to predict the state of next time interval. Finally, the predicted state is mapped into a combination of frequency and voltage, which are scaled accordingly, to make the system more power efficient.

並列關鍵字

Low Power ； Power Predict ； dynamic prediction model

參考文獻

[1] M. Horowitz, T. Indermaur, and R. Gonzalez, “Low-power digital design,” IEEE Symp. on Low Power Electronics, 1994, pp. 8-11

[2]洪志銘，“具有彈性的自動化嵌入式系統功率測量平台之設計與實作”，台北科技大學碩士論文，

[8] Canturk Isci, Gilberto Contreras and Margaret Martonosi ,“Live, Runtime Phase Monitoring and Prediction on Real Systems”, MICRO 39 Proceedings of the 39th Annual IEEE/ACM International Symposium on Microarchitecture.

[10] P. Bellasi, W. Fornaciari, and D. Siorpaes, ”Predictive Models for Multimedia Applications Power Consumption based on Use-Case and OS Level Analysis,” in Proc. of DATE, pp.1446-1451, April 2009.

[11] C. Isci and M. Martonosi, ”Runtime Power Monitoring in High-end Processors: Methodology and Empirical Data,” in Proc. of Internatinal Symposium on Microarchitecture, pp.93-104, 2003.