A network self-healing algorithm was presented. The proposed algorithm reduced the probability of selecting the failure or congestion paths and achieved self-healing by selecting recovery path adaptively when the nodes failure or link congestion in the network, which was based on Q-learning feedback mechanism, multi-QoS constrained of the evaluation function and path selection strategy using Boltzmann-Gibbs distribution. Simulation results show that the proposed algorithm outperforms in the recovery rate, the diffserv-aware capability and network resource optimization.%提出了一种网络自愈算法,当网络中的节点发生故障或链路出现拥塞时,该算法利用Q学习的反馈机制、多QoS约束的评价函数和基于Boltzmann-Gibbs分布的路径选择策略,自适应地选择恢复路径,降低了选择发生故障和拥塞路径的概率,从而实现了自愈.仿真结果表明,该算法在恢复率、区分业务能力和网络资源优化等方面,表现出了良好的性能.
展开▼