Optimal end-to-end sliding window flow control in high-speed networks

Abstract

The performance of the end-to-end sliding window flow control mechanism on a single virtual circuit in the high-speed, wide-area network environment is investigated. The propagation delay is taken into consideration as the dominant delay in the analysis of the sliding window flow control mechanism. A closed queuing network is used to model a single virtual circuit in high-speed networks. Three different approaches are adopted which give similar results. By the influence of propagation delay, the optimal window size which maximizes the network power no longer depends on the number of nodes only, but also depends on the ratio of propagation delay to switch processing delay. The results show that the optimal window size still can be represented as a simple function of network parameters. These results are validated by simulation. >