Research And Implementation Of Embedded Systems Power Management

With the rapid development of integration craft of electronic circuit manufacturability, the size of computer systems become smaller, and the performance go stronger. Meanwhile, the growth of mobile and communication technology make these electronics more portable, a lot of them adapt battery for power supply. As we know, strong performance means high power consumption, however, the poor battery technology and more environmental awareness make the problem between performance and dissipation more serious. The PM (power management) technology alleviate the situation by effectively allocate power to devices. PM technology is very common in desktop computer and server. However, common PM technology is fall behind in embedded field, for example it has little common PM software and no standard, since the application of embedded system usually just for special situation. In order to introduce common PM technology into embedded filed, static power management and dynamic power management are researched in this thesis. In the aspect of static power management, the thesis firstly analyzed the existing relevant technology, and then presented a scheme that suitable for embedded system, finally implemented "suspend" and "hibernate" of static power management on experimental prototype, In dynamic power management aspect, several current strategies applied so far include Greedy, Timeout, Predictive, Stochastic Model were analyzed, and the existing issues of these strategy were pointed out, at last, presented a dynamic self-adaptive strategy that include index average self-adaptive algorithm and classified sliding window average algorithm. The thesis verified the feasibility and practical applicability though implement the aforementioned "suspend to ram" and "suspend to flash", proved the superiority of dynamic self-adaptive by analyzing the results of simulation of dynamic power management.