Distributed Simulation

Conservative simulation algorithms are based on the restriction that an event cannot be executed unless it is proved that no other event shows up during execution. This restriction is fulfilled by using strict synchronization between logical processes. Different methods for the analysis of event-based systems have been introduced for revealing the parallel properties of the several applications, algorithms, as well as environments. To handle parallel-event traces from a more methodological point of view, the distributed-simulation method was been introduced. Distributed simulation is used for real-world system analysis, where events have to be processed in a concurrent way. This results in a speed up of the simulation task. The basic ideas behind distributed simulation are mapping and scheduling, which means that an event can only be executed if it is proved that this event is independent of other prospective executable events. For this reason guarantee messages are exchanged. This concept finally results in Classical simulation methods Optimistic simulation methods Hybrid methods Speculative simulation methods Deterministic tie-breaking methods Distributed shared memory methods Fuzzy-based methods One of the most important points in event-oriented simulation is that every event will be processed in a deterministic way. Consider a tie-breaking method that does not sequentialize simultaneous events. Hence one can assume that sequencing pays attention to the transitive generation sequence. Consider that the simultaneous events e and é of the same logical process can be processed in such a way that e can be executed before é if é can be directly or indirectly generated from e. This constraint of sequentialized simulation is a simple boundary, meaning é does not exist before e has been executed and the event execution is not repeated, which is not possible at any time for an optimistical method. The constraint which has to be fulfilled is called deterministic tie-breaking.