Abridged Petri Nets

TL;DR

Abridged Petri Nets (APNs) offer a simplified, more transparent graphical framework for modeling complex stochastic systems, enabling efficient performance evaluation through direct linking of system components and hierarchical construction.

Contribution

This paper introduces Abridged Petri Nets, a novel graphical modeling framework that simplifies SPNs by directly linking places and using hierarchical submodels, enhancing model clarity and efficiency.

Findings

APNs are more compact than SPNs.

APNs facilitate more transparent models.

APNs enable efficient performance evaluation.

Abstract

A new graphical framework, Abridged Petri Nets (APNs) is introduced for bottom-up modeling of complex stochastic systems. APNs are similar to Stochastic Petri Nets (SPNs) in as much as they both rely on component-based representation of system state space, in contrast to Markov chains that explicitly model the states of an entire system. In both frameworks, so-called tokens (denoted as small circles) represent individual entities comprising the system; however, SPN graphs contain two distinct types of nodes (called places and transitions) with transitions serving the purpose of routing tokens among places. As a result, a pair of place nodes in SPNs can be linked to each other only via a transient stop, a transition node. In contrast, APN graphs link place nodes directly by arcs (transitions), similar to state space diagrams for Markov chains, and separate transition nodes are not needed. Tokens in APN are distinct and have labels that can assume both discrete values ("colors") and continuous values ("ages"), both of which can change during simulation. Component interactions are modeled in APNs using triggers, which are either inhibitors or enablers (the inhibitors' opposites). Hierarchical construction of APNs rely on using stacks (layers) of submodels with automatically matching color policies. As a result, APNs provide at least the same modeling power as SPNs, but, as demonstrated by means of several examples, the resulting models are often more compact and transparent, therefore facilitating more efficient performance evaluation of complex systems.