Brownian Motion Based Queuing Analysis and Bandwidth Prediction for Aggregated Domain in Backbone Networks

It is widely accepted that to realize QoS guarantee in global networks, the negotiation of QoS parameters is no longer at the level of individual routers but at the level of domains. To abstract the QoS capability of a domain, detailed information should be hidden and the overall QoS estimation is needed. This paper uses a Dijikstra-like algorithm to calculate the maximum bottleneck path of a domain. Furthermore this paper demonstrates that the proper model to character the traffic in the backbone networks is using fractional Brownian motion, and with this model, this paper makes a correlated study on bandwidth, delay and loss rate. By ascribing the three QoS parameters to the single parameter of bandwidth, the abstract of QoS capability of a domain can be elegantly be advertised and negotiated. Numerical study shows this method is feasible and practical.