Designing parsimonious scheduling policies for complex resource allocation systems through concurrency theory

In a recent work of ours we have proposed a theoretical framework for developing optimized scheduling policies for complex resource allocation systems (RAS). This framework is based heavily on the rigorous modeling of the RAS dynamics in the modeling framework of the Generalized Stochastic Petri Nets (GSPNs), and the employment of this framework towards the establishment of a systematic trade-off between the representational economy of the target scheduling policies and their operational efficiency. In this paper, we enhance the representational economy of the target policies in the aforementioned framework by taking advantage of some notions of “(non-)conflict” in the transitional dynamics of the underlying RAS-modeling GSPNs. A series of numerical experiments demonstrate that the representational gains that are effected by the presented methodology can be very substantial.