Utility-Maximizing Scheduling for Stochastic Processing Networks

Stochastic Processing Networks (SPNs) model manufacturing, communication, or service systems. In such a network, service activities require parts and resources to produce other parts. Because service activities compete for resources, the scheduling in a SPN requires deciding which of a set of conflicting activities should be performed at any given time. The goal is to achieve throughput optimality or maximize the net utility of the production. This paper proposes a deficit maximum weight (DMW) algorithm for the problem.