Research On MPLS-TE Routing Algorithm Of Supporting Multi QoS Traffic

In recent years, Internet traffic increased seriously and happened suddenly and unpredictably, which raised new demand about how to use bandwidth effectively.At the same time, various kinds of new businesses appeared constantly so that customers require various kinds of QoS(Quality of Service)also. Internet must optimize performance on one hand to provide various kinds of services and give QoS guarantee, on the other to use effective bandwidth management tools to distribute limited network resources.MPLS(Multi-protocol Label Switching) has unique technical advantages in supporting QoS and TE(Traffic Engineering)and solving network expansibility. The key technique of MPLS supporting QoS and TE is constrain based routing(CBR).The existing CBR algorithms can not satisfy different requirements of different customers because they choose route mainly according to bandwidth parameter,ignoring the other factors such as delay and packet loss ratio.They also aim at different optimization objects and have different efficiencies,computing methods, computation patterns and accuracies.In order to guarantee the QoS requirement of traffic, increase the full throughput of the network and decrease the amount of refused request, this paper synthesizes the existing routing algorithms and raises a MPLS-TE routing algorithm supporting multi QoS traffic.This algorithm includes two stages:off-line optimization and on-line routing. In off-line stage,it forecasts the traffic requirement between source and destination node and computes the approximate optimal paths between source and destination at the aim of minimizing resources consumption.In on-line stage,for the K kinds of requests,it chooses a optimal path for best-effort traffic, bindwidth-sensitive QoS traffic and delay-sensitive QoS traffic from the usable paths on considering bindwidth and delay factors.In order to avoid congestion,there is admission control for the traffic entering into the network when the network load is heavy.By expanding this algorithm the network can support more business types.The simulation results got by the MNS simulaition software prove this algorithm achieves the goal of supporting different QoS and optimizing the network's overall performance.