The Design Of Power Managerment For Mobile Financial Payment Terminal Based On Dynamic Voltage And Frequency Scaling

In recent years, the increasing popularity of the Android smart devices makes the application software increased rapidly. With the growing of its functionality and performance requirements,the power consumption is also rising quickly.However,the development process of the battery technology is far behind the requirements of these devices.Designing low-power dynamic policy on system level to reduce system power comsumption has become an important research content of the design of mobile devices.The thesis analyze the Dynamic Voltage and Frequency Scaling.designed its managerment strategies and implemented it on a mobile financial payment terminals.This article focused on the analysis and design of the DVFS strategy, the core strategy is to adjust CPU voltage and frequency dynamically based on system load rate. In order to solve the inverter delay issues in conventional policy,the thesis quantitatively analyzed the actual test results, designed AVG algorithm model as load prediction algorithm to forcast the system load factor for the next time period.In terms of customized strategy in frequency mechanisms,a Hotplug strategy was proposed combining multi-core CPU's hot swap technology.The strategy dynamically adjust the system CPU running cores and operating frequency According to system requirements. According to the Hotplug strategy make the system run at the highest frequency too much, the degree of user interaction Android specific customized strategy optimization criteria.Finally, in aspects of system implementation,the thesis designed and implemented the whole system framework of the frequency conversion mechanism by client/server architecture.Finally,Verification and test results showed that the design of the power management based on the MT6582 platform achieved the desired requirements. The AVG model prediction algorithm was in line with the actual CPU situation, leading a 5% increasing in hit rate compared to Ondemand policy. System functioned properly, dynamic voltage frequency scaling technology reduced the power consumption of normal state to 528mA,the power consumption of early suspend state is 242mA, the power consumption of sleep state is 13mA.The average consumption in normal operation declined 15.7%.