Diffusion-type autonomous decentralized flow control for open networks

Abstract

In our previous studies, we have proposed diffusion-type flow control for high-speed networks. The reason for proposing this type of control is that decision-making at each node leads to high performance of the whole network although each individual node handles its local traffic flow only on the basis of the information it is aware of from its immediately adjacent nodes. Our control method exhibits the desired stability and adaptability of network performance even when the network has an heterogeneous configuration and the capacity of links is dynamically changed. However, in order to make a comparison with other flow control methods under the same conditions, our evaluation used a closed network model, in which the number of packets was unchanged. This paper investigates the performance of our flow control method using an open network model, in order to show that our flow control method is still effective in more realistic networks. We identify the key issues associated with our flow control in an open network model, and demonstrate a solution.