Traffic prediction based burst assembly mechanism for OBS

Abstract

Optical burst switching provides a feasible paradigm for the next IP over optical backbones. However its burst loss performance can be highly affected by burst contention. In this paper, in order to minimize contention and also to provide better proportional differentiated services, an efficient burst assembly mechanism with traffic prediction characteristics has been developed to reduce the variance of assembled traffic and improve the burst loss performance. Simulation result of the proposed assembly mechanism based on ns-2 network simulator has been presented. Result show that the proposed burst assembly scheme gives better network performance in terms of resource contention, delay and thus lower burst (packet) dropping probability.