University of Houston

In last decades, mobile computing becomes more and more popular. However, one limitation which could hold back its development is that all mobile devices nearly rely on their batteries. Low-power designs to prolong the life of the battery operating in these mobile devices will play an important role in the blossom of mobile computing It also attracts a lot of interests in the real-time research community. Two kinds of techniques to reduce the energy consumption of processor were studied at those works in OS level, dynamic voltage scaling (DVS) and CPU power down methods. Due to a lack of a unified testing framework, most of the proposed algorithms were evaluated via simulations under their own experimental scenarios. However, finding their performance in real processors is essential if these algorithms are to be used in practice. In this thesis, we design a unified framework in Real-Time Linux to evaluate these power-aware scheduling algorithms based on a real Intel PXA255 XScale processor. Several DVS and CPU power down algorithms are implemented and compared on our framework. The framework includes everything needed for evaluating real-time power-aware algorithms on an embedded system, from flexible hardware setup to extendible software design. As far as we know, this is the first implementation on a real embedded system which can evaluate both DVS and the power down algorithms in real-time applications.