Optical Burst Switching (OBS) is a promising solution for the future Internet and has attracted many attentions. In OBS network a key issue is channel scheduling. The objective of existing channel scheduling schemes is to maximize the number of bursts to be transmitted. Unfortunately it does not mean the maximum channel utilization. To maximize channel utilization we introduce an optimization scheduling scheme, in which the problem of channel scheduling is mapped to a problem of finding maximum weight independent set (MWIS). That is, the channel scheduling is equivalent to find a set of no-overlapping bursts such that the sum of bursts sizes is maximal amongst all sets of no-overlapping bursts. Our proposed optimization scheduling scheme can be solved as an integer programming problem in polynomial time, and is analyzed by a GX/GY/1 queue system. Simulation results show outperforming in terms of channel utilization and bit loss probability compared with existing schemes. Discrimination to smaller burst and the alleviation with adjusted weight are discussed. The improvement with adjusted weight is showed as well.
Optical Burst-switched (OBS) is a promising switching technology and expected to support the future Internet backbone with dramatically increasing bandwidth demand. In an OBS network, burst contention causes burst loss due to bufferless nature of OBS core network. This kind of burst loss will interact with the above TCP layer. In this paper, we study the impact of this interaction on TCP fairness. We find significant unfairness among TCP flows that share the OBS core
network, i.e. one flow obtains higher throughput while any others with much lower throughputs. The cause is the phenomenon called "the bigger eats the smaller (BES)", in which a TCP flow with higher rate occasionally will "see" less burst contentions and increase its rate further, while a TCP with lower rate will see more burst contentions and decrease its rate continually. Discuss a simple model to explain BES and verify that a continuous sequence of bursts will enhance BES. Then observe that offset time will be a good choice to control TCP fairness by a curve of unfairness control with offset time adjustment. Finally an adaptive offset time scheduling (AOS) algorithm is proposed. AOS assigns burst offset time value adaptive to the rate of TCP flow. The simulation results show that the fairness can be significantly improved by our AOS scheme.
In this paper, we consider congestion control in input queued crossbar switch environment where each input port with finite buffer space while TCP protocol is employed for end-to-end congestion control. We find that it is impossible to achieve efficiency and fairness among TCP flows at the same time only by queue management. Then we propose a scheme of hFS&rEDF, which combine heuristic fair switch arbitration (hFS) and queue management policy of early drop front randomly (rEDF). In our proposed scheme, switch arbitration strategy of hFS unevenly allows input ports to transfer packets to output ports while packets at head of any other input ports involved in conflicts have to be dropped by the policy of rEDF with a probability. Simulation results prove that our proposed scheme can achieve better tradeoff between throughput and fairness.