Research On The Evaluation Model Of Network Attack Effect Based On Hesitant Fuzzy Theory

Network attack effectiveness assessment is able to proactively identify the weaknesses of networks in network protection,which is of great significance for network security,and thus has received extensive attention and research in recent years.Among the commonly used assessment models,objective assessment algorithms ignore the dependency between assessment and actual situation and a priori knowledge.And subjective assessment algorithms,such as hierarchical analysis and fuzzy hierarchical analysis,suffer from the problem of missing in the conversion process of subjective linguistic information to numerical information and have a complicated computational process.Therefore,the model studied in this topic expects to optimize the fuzzy hierarchical analysis method,retain the advantages of fuzzy hierarchical analysis based on practical and a priori knowledge,and reduce the loss of information in order to improve the rationality of the final assessment results.The main work of this topic is as follows:1)Based on the investigation of the current stage of network attack effectiveness assessment index system,and the study and research of fuzzy mathematics,probabilistic hesitant fuzzy sets are introduced for the quantification of assessment indexes.2)Combining probabilistic hesitant fuzzy sets,the method of calculating index weights by fuzzy hierarchical analysis is improved.3)A fuzzy C-mean clustering algorithm is introduced to calibrate the assessment index weights and obtain more reasonable evaluation results.4)To compare and verify the above proposed methods,design experiments to prove the accuracy and validity of the evaluation model.5)Based on the evaluation model proposed in this topic,propose a reasonable network attack effect evaluation index system,design and implement a network attack effect evaluation system,and test the functionality of the system.The specific innovation points of this topic are as follows.1)Propose a qualitative index quantification method based on probabilistic hesitant fuzzy term set.After analyzing each network attack evaluation model,we propose a probabilistic hesitant fuzzy term set for the quantification of qualitative indicators in response to the lack of qualitative indicators in the evaluation model.The natural language descriptions of qualitative indicators are converted into mathematical representations in probabilistic hesitant fuzzy language.Using the probabilistic hesitant fuzzy language set can reduce the loss of information when converting language to numerical values to a greater extent.This project designs and implements the processing process and demonstrates the effectiveness of this quantification method through experiments.2)Propose a method for calculating index weights based on probabilistic hesitant fuzzy setsWhen calculating weights by fuzzy hierarchical analysis method,there are problems such as missing information in the process of constructing consistency discriminant matrix,complicated algorithm of testing consistency matrix,and lack of non-consistency matrix modification method.The use of probabilistic hesitant fuzzy sets to construct the judgment matrix expresses the relative importance between indicators more precisely and makes the original judgment matrix closer to consistency.Meanwhile,an automated modification of the non-consistency matrix is proposed to improve the efficiency of weight calculation.This topic designs and implements the weight calculation method based on probabilistic hesitant fuzzy sets,and proves the effectiveness and rationality of the algorithm through experiments and comparisons.3)Propose the calibration method of evaluation index weights based on fuzzy Cmeans clustering.The subjective weighting method has the advantage of following the actual situation and a priori knowledge,but there may be a situation of relying on human judgment and generating errors.To reduce the errors in the weight calculation results,fuzzy C-mean clustering is used to perform weight calibration.Fuzzy C-mean clustering introduces the affiliation matrix to classify the categories,which has higher flexibility compared with the traditional clustering methods.In this project,we use the fuzzy C-mean clustering method to classify the categories of historical weight data,and use the affiliation matrix and clustering center to delineate the reasonable range of each evaluation index weight for the purpose of calibration.