We present a flow optimization model for IP networks where end-to-end paths are identified for each ingress-egress pair in the MPLS-capable part of the network. The model computes link- and path flows by incrementally distributing the offered traffic among the paths. Fault-tolerance and congestion control are achieved by spreading the offered traffic over a set of active paths. A set of back-up paths is reserved for carrying the traffic of failed or congested paths. The routing performance of this model is evaluated through two algorithms: weighted traffic distribution (WTD) and flow deviation (FLD). WTD distributes the offered traffic among the available paths using weights allocated to these paths. FLD is a variant of the well-known flow deviation algorithm.
展开▼
