by Fredrik Hillnertz & Max Roth
This essay is a qualitative study of two simulation paradigms: event-based and time-based
simulation. The purpose is to find positive and negative characteristics in implementations where
one of the simulation paradigms has been applied. Two models were implemented, a queue model
and a percolation model. It appeared to be a bit harder to understand event-based simulation
while the time-based simulation was easy to implement. Time-wise it took about the same amount
of time to implement the queue model with both simulation paradigms while the percolation
model required more time to implement with event-based simulation. This was mostly due to the
fact that it was harder to interpret the percolation model in terms of events.
By carefully observing the execution time while at the same time changing the parameters it was
possible to distinguish characteristics of the different implementations. Event-based simulation
had the best performance in most cases, which was influenced a lot by what time-step was chosen
for the time-based simulation. It was also found that both implementations had very different
execution times depending on which parameters were variated. The overall conclusion is that
event-based simulation is the most time-efficient solution but also the most difficult one. One
should have the approach to attempt implementation of an event-based simulation and use timebased
simulation in case it is not possible. If you have the time you could also try to implement
both variants to get the best of both.