Using reinforcement learning for agent-based network fault diagnosis system
Jingang Cao · 2011
In the network, it is important that faults can be diagnosed at early stage before they result in serious fault. However, the situation is not optimistic, which depends on what network management software is used. Aiming to this problem, a mobile agent-based network fault diagnosis model is proposed. In the model, agent can learn by reinforcement learning (RL), which can improve fault diagnosis performance. The structure and function of model, especially the architecture and learning algorithm of diagnostic agent, is depicted. At last, compared the system performance through simulation and experiment, and results show that the model has greater advantage.