PeabraiN: A PIPE Extension for Performance Estimation and Resource Optimisation

Many discrete systems with shared resources from different artificial domains (such as manufacturing, logistics or web services) can be modelled in terms of timed Petri nets. Two studies that may result of interest when dealing with such a systems are the performance evaluation (or completed jobs per unit of time) and the resource optimisation. Exact performance evaluation, however, may become unachievable due to the necessity of an exhaustive exploration of the state-space. In this context, a solution can be to estimate the performance by computing bounds. Resource optimisation leverages a budget and distributes resources in order to maximise the system performance. In this paper, we present Pea brain, a collection of PIPE tool-compliant modules for performance estimation and resource optimisation based on bounds computation for Stochastic Petri Nets. The algorithms supporting the modules make an intensive use of linear programming techniques and therefore their computational complexity is low. Besides, other PN properties, such as structural enabling bound at a transition, structural marking bound at a place or visit ratios computation, are added to PIPE tool as well.