Testsets Generation for Reactive Scheduling

N. Policella and R. Rasconi

In Workshop on Experimental Analysis and Benchmarks for AI Algorithms (AI*IA-RCRA Working group on Knowledge Representation and Automated Reasoning of the Italian Association of Artificial Intelligence), Ferrara, Italy, 2005

This paper presents some basic ideas for the creation of a benchmark generator for reactive scheduling problem instances. The main motivations behind this work grow out either from the recognized lack (hence the necessity) of benchmark sets for this specific problem as well as from the conviction that the resolution of a scheduling problem consists both in the synthesis of an initial solution (``static'' scheduling) and in the utilization of a number of methodologies dedicated to the continuous preservation of solution consistency. In fact, the occurrence of exogenous events during the execution phase in real working environments, often compromises the schedule's original qualities.