Interval Branching

We propose a new simulation method, called interval branching, which aims to facilitate efficient simulation studies of systems that involve temporal uncertainty. The method uses interval timestamps for events and explores different execution orders of overlapping events by branching the simulation. We first present a sequential version of interval branching, and then extend it to parallel simulation using the logical process paradigm. The parallel interval branching algorithm uses the logical-process cloning technique for efficient computation of branches. Our preliminary experiments show its potential as an efficient method for discrete-event simulation studies.