Quantifying the effect of extended offsets in optical burst switching networks

Optical burst switching (OBS) has been proposed as a candidate architecture for next generation optical networks. In OBS networks, multiple classes can be implemented by assigning a different-sized time offset between the control packets and bursts of each class. However, these systems are difficult to model. To date, the only proposed models of these systems make assumptions that significantly restrict the scope of their applicability. In this paper, we present a general analytical model for multi-class, OBS systems. The model is applicable to systems with an arbitrary number of classes where each class has an arbitrary burst-arrival rate, burst-length distribution, and offset size. We use the model to derive an accurate expression for the blocking probability of the highest-priority class in a multi-class OBS network and demonstrate its accuracy using simulation.