With the popularization of emerging services such as 5G, cloud computing, the backbone network traffic has grown rapidly and the network running state will change frequently. Unfortunately, the unrelated development of the IP layer and the optical layer in the existing network leads to lack of flexibility in services and makes the hardness and cost of the overall operation and maintenance of the backbone network higher and higher. although the IP layer and the optical layer are converged, due to the uncertainty and unpredictability of the IP business itself, the lack of dynamic interaction mechanism between the two layers becomes increasingly prominent. Therefore, this paper proposes a new NSE-RFC (a network operation state evaluation algorithm based on random forest classification) algorithm, which provides a scientific and comprehensive evaluation of network running state. The results show that the correct rate of the algorithm is 97.5%, which can accurately evaluate the network running status. Finally, the validity of the model is verified by the consistency of the evaluation results of the model and the variation of the time distribution characteristic map of each evaluation index.
|