Real time simulation framework for RT-DEVS models

Abstract

Discrete event simulation is widely used not only to predict performance but also to analyze behavior of complex systems. Real time, discrete event simulation is a means to verify a real time system in which simulation models may interact with an external environment, such as software components, hardware components or human operators. In such simulation, a simulator must handle external events from its environment in a timely manner. Simulation methodologies developed in non-interactive simulation, such as performance simulation, do not have such capabilities in event processing. This is because such methodologies employ virtual (logical) time, not real (physical) time, in event scheduling and processing. In real time simulation, it is vital to ensure that the timing requirements of the simulation models are satisfied because the models operate based on real time and compete with one another for processor time. Scheduling feasibility analysis verifies that the component models of a target system are able to timely perform their behavior. This analysis inherently depends on an underlying computational model and a scheduling policy. Early analysis is essential because a later runtime checking is risky and costly, and besides, manual reasoning is error prone. This thesis proposes a scheduling policy and its feasibility analysis method for the real time DEVS(RT-DEVS) models. The RT-DEVS model provides features such as object-orientation, state description of behavior, and formal semantics for executability. The event driven scheduling policy has been devised for the RT-DEVS models. The RT-DEVS models with the scheduling policy produce a set of tasks with a unique relation, for which we devised a scheduling feasibility analysis method. In order to improve the tractability of formal analysis, we decided to sacrifice the generality by assuming a periodic state sequence of a target system. The proposed analysis method consists of two phases. The first phase is state s...