Network Node Resource Risk Assessment Based On ANFIS

Computer networks have become an indispensable part of human life.Along with the gradual deepening of Internet applications,more and more malicious attacks are also hidden in the network.Network attacks or excessive usage can all lead to network node resources becoming tense,network node performance declining,and even the risk of crashes.Once the network node will cause great economic losses,the security of the node host can be analyzed and evaluated by studying the risk value of the network node resources to ensure the normal operation of the node and the user's good experience.Therefore,research on the network node resource risk assessment has certain the actual significance.When the network node host is attacked or used excessively,the main resource indicators such as network card traffic,CPU utilization,disk occupancy,and memory usage of the network node host are bound to fluctuate abnormally.Their fluctuations are not Independent,but there is a certain degree of relevance.Therefore,when establishing a network node resource risk assessment model,the nonlinear,time-varying,and uncertainties of the network node resource sequence need to be considered.Traditional risk assessment methods based on mathematical formulas are difficult to deal with nonlinearity and time-variation.The adaptive fuzzy inference system(ANFIS)can analyze the precise mathematical formulas of the objects to be processed.The model parameters will be based on the training data.Constant self-adjustment to achieve a stable state.Based on this,this paper proposes a network node resource risk assessment model constructed using an adaptive fuzzy inference system.The dissertation first describes the research on building network node resource risk assessment model based on Grid Partition(GP)ANFIS model.Based on the data of main performance indicators of node hosts caused by various types of network behaviors,a risk assessment model based on grid-based ANFIS network node resources was constructed,and the risk value of node resources was evaluated.Secondly,the paper studies the network node resource risk assessment model based on subtraction clustering method(SCM)ANFIS model,and also uses the node host main performance indicators to construct a network node resource based on subtractive clustering for risk assessment model.The risk of node resources was evaluated.The subtraction clustering effect is obvious,and the network structure of the network node resource risk assessment model is greatly optimized while the final evaluation effect is not affected,making the network structure clear and clear.Finally,the thesis describes the process comparison between the classical BP neural network node risk assessment model and the ANFIS evaluation model based on grid partitioning and subtraction clustering.The results show that: the adaptive fuzzy inference system model based on subtractive clustering can be compared.Objectively and truly reflect the current risk situation of the network node host.This assessment model is a good network node resource risk assessment model,which provides a new idea for the study of network node resource risk assessment.