Towards a tropical automaton product minimizing global completion times

Abstract

The main contribution of this paper is oriented towards system modeling with resource conflicts, in which a set of tasks must be completed at the earliest. Mutual exclusion between tasks with underlying resource conflicts is modeled as subsystems through local tropical automata. Parallel task execution is explored through a new framework for a synchronous product of tropical automata directly minimizing the global completion time of a set of tasks. A dater-based analysis is proposed to determine an optimal qualitative schedule and relies on proposed definitions of global, private, and synchronizing daters. This approach drastically narrows the solution spectrum, and allows finding the optimal task schedule. The approach is applied to a flow network as a case study, but can be applied to systems of different nature.