基于粒子群权值优化的网络可生存性增强方法

针对网络中发生频率最高的单链路瞬时故障，提出了一种应用粒子群算法优化链路权值来增强网络可生存性的方法。引入费用函数对利用率过高的链路赋以惩罚性的高费用来避免链路过载，以网络在无故障场景下最高链路费用与单链路故障场景下最高链路费用的加权和作为目标函数，建立了优化算法模型，并应用粒子群优化算法求解最优权值。实验结果表明，算法求得的权值可以使网络在故障条件下保持较低的链路利用率，避免了因流量转移而造成网络拥塞，增强了网络可生存性。

Approach of enhancing network survivability by optimizing weights based on particle swarm optimization

As most of network failures are transient single link failures, a new approach of using Particle Swarm Optimization (PSO) algorithm to optimize link weights for enhancing network survivability was proposed. A cost function was introduced to put high cost on links with high utilizations for avoiding link overloaded. The objective function was a weighted sum of two proportions: one is the maximum cost under normal state, and the other is the maximum link cost under all single link failures. Then the algorithm model was built and PSO algorithm was used to find the optimal weights. The experimental results show that the weight calculated by the proposed method can keep lower link utilization under failure states, and prevent the network from congestion due to traffic diversion. Therefore, the network survivability can be improved.