Ip Network Traffic Engineering Optimization Algorithm

With the popularity of Internet and a variety of high bandwidth services continue to emerge, bandwidth resources of IP network become increasingly strained. In order to make better use of network resources and provide better quality of service, traffic engineering has become a hot issue in IP network. In this paper, we will focus on finding the optimal link weight and paths with multiple QoS constraints.First, we research traffic engineering with a certain traffic matrix in IP network, and propose two new weight optimization algorithms. The first weight optimization algorithm is based on Lagrangian relaxation techniques. The second weight optimization algorithm is based on the MIP model of calculating optimal link weights and use mathematical models solvers (e.g. CPLEX). Our simulation shows that the second weight optimization algorithm can significantly reduce the amount of congestion ratio.Second, we research traffic engineering with uncertain traffic matrices in IP network, and propose two new weight optimization algorithms for finite traffic matrices and infinite traffic matrices respectively. The first weight optimization algorithm with multiple traffic matrices is based on Lagrangian relaxation techniques. Simulation shows that link weight derived from the algorithm can perform well for multiple traffic matrices. The second weight optimization algorithm with infinite traffic matrices is based on the MIP model of calculating optimal link weights under uncertain traffic matrices in IP network and use mathematical models solvers (e.g. CPLEX). Our simulation results show that the proposed algorithm performs better than existing weight setting schemes in terms of avoiding network congestion.Third, we research traffic engineering with QoS business in IP network. A new multi-constrained routing algorithm based on Lagrangian relaxation is proposed. We compare the algorithm with other multi-constrained routing algorithm. Then we present two new multi-constrained link-disjoint routing algorithm based on Lagrangian relaxation. Under the Lagrangian relaxation framework, the first multi-constrained link-disjoint routing algorithm relaxes all the QoS constraints and link-disjoint constraints and reduces the original problem to two independent sub-problems. The second algorithm relaxes all the QoS constraints and reduces the original problem to an unconstrained link-disjoint routing problem, which is a well-known tractable problem and can be solved efficiently. Our simulations show that the second algorithm not only performs better than the first multi-constrained link-disjoint routing algorithms in terms of the rate of success in finding the feasible paths pair, but also can achieve solutions which are close to the optimal solutions solved by commercial ILP solvers.