基于DPM和DVFS的嵌入式系统低功耗优化技术研究
发布时间：2018-08-01 14:08
【摘要】：随着计算机技术和移动互联网技术的迅速发展,嵌入式移动终端(例如手机、平板)已经成为人们日常生活中不可缺少的一部分。嵌入式移动便携设备依靠电池作为能量补给,而电池的容量却受体积和重量的限制很难有效提高,因此嵌入式移动设备的功耗问题成为当今研究的一个关键问题。 降低嵌入式系统的功耗有多种方法。其中动态电源管理(Dynamic Power Management, DPM)和动态电压频率调整技术(Dynamic Voltage and Frequency Scaling, DVFS)是降低系统功耗的两种关键技术。本文从动态电源管理和动态电压频率调整技术两种策略降低优化移动嵌入式系统的功耗。 动态电源管理主要依据设备工作负载的变化切换其工作状态以达到系统功耗最小化。例如当外围设备的空闲期足够长时,使其关闭或进入休眠状态以节约电池电量。本文首先从应用层、应用框架层、Linux内核层分析了Google Android的电源管理机制原理,然后从小流量模式(10Kb/分钟)和大流量模式(2Mbit/分钟)两种情况实验并分析了Android网络模块在系统总功耗中的重要作用,接着在此基础上提出了一个对Android网络模块的动态电源管理优化解决方案。该方案从网络流量、电池电量、时间、是否下班高峰阶段四个因素考虑,实现Android系统在网络模块上的动态电源管理。由实验结果可以看出,手机移动网络模块(GPRS/WIFI)在手机系统功耗中占据了十分重要的位置,因此该解决方案对提高移动终端的能量利用率、降低移动终端的功耗具有重大的意义。 动态电压频率调整技术是降低嵌入式系统功耗的另一种重要方法。DVFS根据系统任务的紧迫程度来动态调节处理器的供电电压和时钟频率,以达到任务响应时间和系统功耗之间的平衡。嵌入式移动终端中使用的处理器通常都具备了动态电压频率调整能力,在操作系统中如何高效使用处理器提供的这一能力降低功耗成为一个值得研究的问题。本文针对以Linux为内核的移动操作系统,提出了一种优化的、细粒度的DVFS算法,并在Android操作系统上加以实现。该算法基于历史数据,使用线性预测的方法,指导电压频率调整的方向和幅度,线性预测中的参数通过回归方法确定。实验结果表明,本文所提的优化细粒度DVFS算法相比于Linux内核默认使用的粗粒度调频算法,能降低系统13.55%的功耗,提升了移动终端的续航时间。
[Abstract]:With the rapid development of computer technology and mobile Internet technology, embedded mobile terminals (such as mobile phones, tablets) have become an indispensable part of people's daily life. Embedded mobile portable devices rely on batteries as energy recharge, but the capacity of batteries is limited by volume and weight. Therefore, the power consumption of embedded mobile devices has become a key issue. There are many ways to reduce the power consumption of embedded systems. Dynamic power management (Dynamic Power Management, DPM) and dynamic voltage frequency adjustment (Dynamic Voltage and Frequency Scaling, DVFS) are two key technologies to reduce power consumption. In this paper, the power consumption of mobile embedded system is reduced by two strategies: dynamic power management and dynamic voltage frequency adjustment. Dynamic power management is mainly based on the variation of the device workload to switch the working state to minimize the power consumption of the system. For example, when a peripheral device has a long idle period, turn it off or go to sleep to save battery power. Firstly, this paper analyzes the principle of Google Android power management mechanism from the application layer and application frame layer. Then, the important role of Android network module in the total power consumption of the system is analyzed by two experiments: small traffic mode (10Kb/ minute) and large traffic mode (2Mbit/ minute). Then a dynamic power management optimization solution for Android network module is proposed. Considering four factors, such as network flow, battery power, time, whether or not the off-duty peak stage, this scheme realizes the dynamic power management of Android system on the network module. It can be seen from the experimental results that mobile network module (GPRS/WIFI) occupies a very important position in the power consumption of mobile phone system, so this solution has great significance to improve the energy utilization of mobile terminal and reduce the power consumption of mobile terminal. Dynamic voltage frequency adjustment is another important method to reduce the power consumption of embedded system. DVFS dynamically adjusts the power supply voltage and clock frequency of the processor according to the urgency of the system task. In order to achieve a balance between task response time and system power consumption. The processors used in embedded mobile terminals usually have the ability to adjust the dynamic voltage frequency. How to efficiently use the processor to reduce the power consumption in the operating system is a problem worth studying. In this paper, an optimized, fine-grained DVFS algorithm for mobile operating system with Linux kernel is proposed and implemented on Android operating system. Based on historical data, the method of linear prediction is used to guide the direction and amplitude of voltage frequency adjustment. The parameters of linear prediction are determined by regression method. The experimental results show that the proposed optimized fine-grained DVFS algorithm can reduce the power consumption of the system by 13.55% compared with the coarse-grained FM algorithm used by the Linux kernel by default and improve the mobile terminal life.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP368.1
本文编号：2157823
[Abstract]:With the rapid development of computer technology and mobile Internet technology, embedded mobile terminals (such as mobile phones, tablets) have become an indispensable part of people's daily life. Embedded mobile portable devices rely on batteries as energy recharge, but the capacity of batteries is limited by volume and weight. Therefore, the power consumption of embedded mobile devices has become a key issue. There are many ways to reduce the power consumption of embedded systems. Dynamic power management (Dynamic Power Management, DPM) and dynamic voltage frequency adjustment (Dynamic Voltage and Frequency Scaling, DVFS) are two key technologies to reduce power consumption. In this paper, the power consumption of mobile embedded system is reduced by two strategies: dynamic power management and dynamic voltage frequency adjustment. Dynamic power management is mainly based on the variation of the device workload to switch the working state to minimize the power consumption of the system. For example, when a peripheral device has a long idle period, turn it off or go to sleep to save battery power. Firstly, this paper analyzes the principle of Google Android power management mechanism from the application layer and application frame layer. Then, the important role of Android network module in the total power consumption of the system is analyzed by two experiments: small traffic mode (10Kb/ minute) and large traffic mode (2Mbit/ minute). Then a dynamic power management optimization solution for Android network module is proposed. Considering four factors, such as network flow, battery power, time, whether or not the off-duty peak stage, this scheme realizes the dynamic power management of Android system on the network module. It can be seen from the experimental results that mobile network module (GPRS/WIFI) occupies a very important position in the power consumption of mobile phone system, so this solution has great significance to improve the energy utilization of mobile terminal and reduce the power consumption of mobile terminal. Dynamic voltage frequency adjustment is another important method to reduce the power consumption of embedded system. DVFS dynamically adjusts the power supply voltage and clock frequency of the processor according to the urgency of the system task. In order to achieve a balance between task response time and system power consumption. The processors used in embedded mobile terminals usually have the ability to adjust the dynamic voltage frequency. How to efficiently use the processor to reduce the power consumption in the operating system is a problem worth studying. In this paper, an optimized, fine-grained DVFS algorithm for mobile operating system with Linux kernel is proposed and implemented on Android operating system. Based on historical data, the method of linear prediction is used to guide the direction and amplitude of voltage frequency adjustment. The parameters of linear prediction are determined by regression method. The experimental results show that the proposed optimized fine-grained DVFS algorithm can reduce the power consumption of the system by 13.55% compared with the coarse-grained FM algorithm used by the Linux kernel by default and improve the mobile terminal life.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP368.1
【参考文献】
相关博士学位论文 前1条
1 卜爱国;嵌入式系统动态低功耗设计策略的研究[D];东南大学;2006年
，本文编号：2157823
本文链接：https://www.wllwen.com/kejilunwen/jisuanjikexuelunwen/2157823.html
