Energy optimization is a critical design concern for embedded systems. Combining DVFS+DPM is considered as one preferable technique to reduce energy consumption. There are optimal DVFS+DPM algorithms for periodic independent tasks running on uni-processor in the literature. Optimal combination of DVFS and DPM for periodic dependent tasks on multi-core systems is however not yet reported. The challenge of this problem is that the idle intervals of cores is not easy to model. In this paper, a novel technique are proposed to directly model the idle intervals of individual cores such that both DVFS and DPM can be optimized at the same time. Based on this technique, the energy optimization problem is formulated by means of mixed integrated linear programming. We also present techniques to prune the exploration space of the formulation. Experimental results using real-world benchmarks demonstrate the effectiveness of our approach compared to existing approaches.
Complex distributed systems are currently disseminatedover a large range of application domains, particularly in-herent in cyber-physical/embedded systems. These systemsare typically subject to several non-functional constraints,stretching from resource limitations to timeliness, includingsafety and other constraints. Taming complexity in their designis particularly important to ensure a swift development and acorrect software product.To this end, component-based software development ap-proaches are particularly well suited. In such approaches acomplex software system is divided into a number of simplersoftware components. Each component is developed indepen-dently, potentially by different teams. The components areintegrated at the final stage of the development. On the otherhand, for predictable resource provisioning, we can reservea fraction of the resource time for each component. Thistechnique is known as the resource reservation technique project centre in trichy . Moreover, reservations enforce mutual isolation, par-ticularly temporal isolation. Thus, components that run insideadequate reservations can be proved correct independently ofother components possibly running in the system
Comments