Studying performance of DEVS modeling and simulation environments using the DEVStone benchmark

摘要

The Discrete Event System Specification (DEVS) formal modeling and simulation (M&S) framework (which supports hierarchical and modular model composition) has been widely used to understand, analyze and develop a variety of systems. Numerous DEVS simulators have been developed; nevertheless, evaluating the performance of such simulators is a complex process and it has been usually done using ad hoc methods. DEVStone, instead, is a synthetic benchmark that can be used to automate the evaluation of the performance of DEVS-based simulators. DEVStone generates a suite of models with varied structure and behavior automatically. To do so, it uses a standardized mechanism that can be the basis for comparisons between simulation software tools. As a proof of the concept, we present various tests in which DEVStone was used to study the efficiency of five different simulation engines. In this case, we compared various versions of the CD++ simulator, and then compared its performance with the 'A Discrete Event System Simulator' (ADEVS) M&S tool. This is the first effort in which these simulation tools have been thoroughly compared with a very demanding set of experiments. The use of DEVStone allowed a standardized and exhaustive method to compare different features of the simulation software. We show how the basic ideas used for DEVStone facilitates performance analysis for upgrades and updates of a given simulation engine, while also providing a common metric to compare different M&S environments.