Search

Burst assembly at edge nodes is an important issue for the optical burst switching (OBS) networks because it affects the traffic characteristics and has an influence on the burst loss performance. By applying the fractional Brownian motion (FBM) model, we find that the timer-based and threshold-based assembly algorithms could not avoid increasing the traffic variability of the science information network (SINET), which will deteriorate the network performance. Therefore a novel traffic-smoothing burst assembly method named sliding window-based assembly algorithm is introduced to reduce the variance of the assembled traffic. It can smooth the traffic by transmitting a sequence of bursts at an average length to make traffic be sent out at an average rate in small timescales. The simulation results show that the sliding window-based assembly algorithm is better than general assembly algorithms in terms of the burst loss ratio