Date of this Version
Battery powered embedded system plays an important role in daily life. Dynamic voltage and frequency scaling (DVFS) techniques are used to dynamically alter the voltage and frequency levels of the central processing unit (CPU). In this thesis, DVFS is proved to be a highly effective method of achieving low power consumption while meeting the performance requirements.
This thesis aims at developing an energy- and temperature-aware dynamic voltage and frequency scaling scheme which can prolong battery life in an embedded system. The contribution of this research in this scheme includes:
1. Studying the battery performance under different circumstances.
2. Exploring the relationship of energy and CPU frequency. Finding out the best frequency and voltage set for the battery usage based on the Imote2 platform.
3. Developing an energy- and temperature-aware DVFS algorithm to prolong the battery life and maximize the usage of battery energy.
4. Linux programming based on the kernel function of CPU scaling to achieve the algorithm. Analyzing and estimating the energy usage in a specific program.
5. Designing experiments to verify the algorithm.
The combination of the DVFS and battery model provides a method capable of maximizing the battery durability and optimizing the embedded system performance.
Adviser: Song Ci