We present a new hierarchical approximation and scheduling approach for applications and tasks with multiple modes on a single processor. Our model allows for a temporal and spatial distribution of the feasibility problem for a variable set of tasks with non-deterministic and fluctuating costs at runtime. In case of overloads an optimal degradation strategy selects one of several application modes or even temporarily deactivates applications. Hence, transient and permanent bottlenecks can be

more »

... rcome with an optimal system quality, which is dynamically decided. This paper gives the first comprehensive and complete overview of all aspects of our research, including a novel CBS concept to confine entire applications, an evaluation of our system by using a video-on-demand application, an outline for adding further resource dimension, and aspects of our prototype implementation based on RTSJ. 1 Introduction This work summarizes the scientific results of the ARTOS [5] project. Its objective was to develop and integrate adaptive resource-management mechanisms in a generic framework for soft real-time systems supporting a dynamic set of applications and tasks. As target platforms general purpose systems as traditional x86, ARM and PPC based platforms were supposed, which may range from small embedded and mobile devices, to desktop PCs and large scale servers. The result is an open and dynamic execution environment for soft real-time applications, which can be integrated on different system levels and used for a variety of scenarios. *