Congestion window based adaptive burst assembly for TCP traffic in optical burst switching networks

Abstract

Burst assembly is one of the key factors affecting the TCP performance in Optical Burst Switching (OBS) networks. Timer based burst assembly algorithm generates bursts independent of the rate of TCP flows. When TCP congestion window is small, the fixed-delay burst assembler waits unnecessarily long, which increases the end-to-end delay and decreases the TCP goodput. On the other hand, when TCP congestion window becomes larger, the fixed-delay burst assembler may unnecessarily generate a large number of small-sized bursts, which increases the overhead and decreases the correlation gain, resulting in a reduction in the TCP goodput. Using simulations, we show that the usage of the congestion window (cwnd) size of TCP flows in the burst assembly algorithm consistently improves the TCP goodput (by up to 38.4%) compared with the fixed-delay timer based assembly even when the timer based assembler uses the optimum assembly period threshold value. One limitation of this proposed method is the assumption that the exact value of the congestion window is available at the burst assembler. We then extend the adaptive burstification algorithm such that the burst assembler uses estimated values of the congestion winpassive measurements at the ingress node. It is shown through simulations that even when estimated values are used, TCP goodput can achieve values close to the results obtained by using exact values of the congestion window. dow that are obtained via