Simulation

The Berkeley Simulation Group (BSG) is a research group headed by Prof. Lee Schruben. Our research interests primarily in the field of discrete event simulation. We mainly use the event scheduling-approach to simulation modeling. This is an extremely powerful modeling paradigm that is very efficient even for large, congested systems. An event graph for a G/G/s queueing system is shown below. The nodes represent the events that can cause changes to the system state. Arrows between the nodes indicate that one event may cause another to happen, possibly after a time delay. Some of the edges are conditional, i.e. the next event will only be scheduled if the condition is true. The variable Q represents the number of jobs waiting for service, and S is the number of available servers. We can model arbitrarily large (or small) systems using the same event graph with different variable initializations.

Current projects include using event graph models of queueing systems to generate math programming models, investigating the relationship between Petri Nets and event graph models, efficient estimation and generation of Non-Homogeneous Poisson Processes for simulation modeling, and developing approximation methods to allow large, congested systems to be simulated more efficiently.

Our current funding comes from grants from the National Science Foundation, and from the Semiconductor Research Consortium. Our web site is www.ieor.berkeley.edu/~BSG.