Improving extensibility of discrete event simulation environment by using event control model

Abstract

Discrete event simulation environments often need to be modified because additional questions to systems may become apparent while observing the simulation results repeatedly. It leads to increasing development budget and depreciating the effectiveness of the environment. To avoid the modifications and to generate the altered results, this dissertation regards the simulation models as black-box models by applying the black-box extensibility. The discrete event simulation can be abstracted to the sequences of events, which are mapped to the discretized states transitions. By controlling the events at the simulation time, the existing behaviors can be extended without modifications. For this, this dissertation proposes an Event Control Model (ECM) with control functions that modulate, delete and generate the events at the simulation time. The simulators with the ECMs can have better extensibility because it becomes more adaptable to possibly unanticipated changes and substitutes with the modifications at the code level. It prevents increasing development costs due to modifications or development of new models, and it provides a new alternative step to users. To support the effectiveness of this approach, this dissertation describes relevant examples, which are composed of an initial simulation model based on our empirical studies. The case studies will show that the actual number of modified lines is reduced due to the ECMs. Also, there exist the uncountable benefits because the existing simulator is reused without modifications.