The treatment of time in distributed simulation

Abstract

Simulation is one of the most important tools to analyse, design, and operate complex processes and systems. Simulation allows us to make a 'trial and error' in order to understand a system and describe a problem. Therefore, it is of great interest to use simulation easily and practically. The advent of parallel processors and languages help simulation studies. A recent simulation trend is distributed simulation which may be called discrete- event simulation, because distributed simulation has a great potential for the speed-up. This thesis will survey discrete-event simulation and examine one particular algorithm. It will first survey simulation in general and secondly, distributed simulation. Distributed simulation has broadly two mechanisms: conservative and optimistic. The treatment of time in these mechanisms is different, we will look into both mechanisms. Finally, we will examine the conservative mechanism on a network of transputers using Occam. We will conclude with the result of the experiments and the perspective of distributed simulation.