Research On Power Management For Embedded Operating System

With the rapid development of embedded computing technology, the embedded systems are increasingly exquisite and small, while becoming more powerful. A miniature CMOS chip can complete diverse functions, such as wireless communication, image process and multimedia. However, the continual working time, produce and run cost, encapsulation bulk, reliability, life circle, and the use of high performance components of the embedded systems are greatly affected due to the power limitation. The power is becoming one of the first constraints of embedded system design. With the development of microelectronics, many advanced hardware power control technology appeared. The software power management technology has become the important means for controlling and optimizing the power of computer system, consequently was heatedly discussed in literature recently.This dissertation analyzed the influence factors of power consumption of embedded systems, and discussed the key position and advantage of operating system in the computer power management. The dissertation introduced the state of art of the power management technologies in the operating system, and then pointed out that unpredictability is the basic characteristic of service requests of computer system. Stochastic analysis is an effective method to study the power management in the operating system. Therefore, this dissertation researched the stochastic model, management policy, control algorithm of the dynamic power management, and the non-realtime and real-time task scheduling for DVS processors. The main contributions of this dissertation are as follows:1. Using the stochastic decision model, prove that the optimal dynamic power management policy is the determinate MARKOV policy, namely the time-out policy. The result reduces the space for searching the optimal solution and lowers the computational complexity, which provides the theory basis to simplify the control algorithm for dynamic power management. The result is also of great significance for the clock frequency determination and the slack...