Dynamic value-density for scheduling real-time systems

Scheduling decisions in time-critical systems are very difficult, due to the vast number of systems' parameters and tasks' attributes involved in such decisions. Value-based scheduling heuristics have been found to experience a more graceful degradation under overload situations than various other heuristics. However, currently existing value-based heuristics utilize the tasks' static attributes, and therefore, they derive fixed scheduling priorities. In this paper, we propose value-based scheduling heuristics that utilize the tasks' dynamic attributes in order to enhance the overall system's performance under normal operating loads and to reduce performance degradation under overload situations.