Research On The Application Of Adaptive Fuzzy Clustering In Network Operation Situation Awareness

Nowadays,the Internet has a high penetration rate,and people’s lives are highly dependent on various network applications.Therefore,it is particularly important to control the state of network operation and establish a real-time network operation situation awareness system.Network operation situation awareness is the process of collecting,evaluating,and predicting characteristic data related to network operation,and finally making subsequent decisions based on the results,such as adjusting network resource configuration,etc.,to ensure that the network can operate normally and efficiently.The network data set has the characteristics of non-a priori,massiveness,etc.The network state assessment is fuzzy and the situation prediction needs to be real-time.Fuzzy C-means(FCM)algorithm can discover the inherent information of non-a priori data sets to make meaningful divisions.The introduction of fuzzy theory into the algorithm is suitable for dealing with ambiguity problems,and the algorithm has high execution efficiency.Therefore,the fuzzy C-means algorithm is applied to the situational awareness problem in the thesis.The main research contents of this thesis are as follows:1.A perception model suitable for network operation situation is proposed,and the perception problem is decomposed into various modules.Based on the principle of completeness and the network performance index system proposed by international organizations,the index selection work has been improved.For the collected index data,based on the network business,the impact of the index on the performance of the business is studied,and the data transformation of the corresponding index can eliminate the index dimension and weight,and make the clustering effect more real and effective.Based on the fuzzy clustering algorithm,a situation assessment process and a real-time updated situation prediction model are proposed,and the effectiveness of the prediction model is verified.2.Optimize the fuzzy clustering algorithm.In the thesis,FCM algorithm is selected as the algorithm of situation assessment and prediction,and the algorithm is improved according to its own shortcomings,so that it can adaptively find the optimal partition according to the data set.1)A new effectiveness index is proposed function based on sample similarity.It is verified by experiments that the new index can effectively identify and judge the validity of the current division,so that the algorithm can adaptively determine the number of clusters and identify whether the situation prediction result is valid.2)The network data is noisy and the algorithm is greatly affected by the selection of the initial clustering center.This thesis uses the SRFCM algorithm to enhance the correct division ability of the FCM algorithm,and then combines the particle swarm optimization(PSO)with the SRFCM algorithm to propose an improved PSO-SRFCM algorithm,using PSO The global search capability makes the algorithm self-adaptively converge to the optimal solution.It is verified through experiments that the PSOSRFCM algorithm has better partitioning ability and stability.3.Build a network topology,run network simulation software on the network topology map to obtain network simulation data,and perform corresponding data transformations on the collected data.Perform fuzzy clustering on the transformed data,analyze the cluster centers,and attach corresponding labels to obtain the link operation status according to the cluster centers.Predict the situation of the subsequent collected data and evaluate the results of the prediction.The scheme in this thesis can effectively evaluate and predict the network operation situation,and provide support for the subsequent decision-making of network managers.