A Network Attack Evaluation Model Based On Variable Weight Theory

The Internet has been widely used.At the same time,network security has a greater and greater impact on individuals and society.With the increasing security threats faced by information systems,network attack effect evaluation technology has emerged.Through the analysis of the state changes of the target information system,the effect evaluation of network attack can accurately and effectively evaluate the attack,and then improve the ability of the information system to deal with various attacks in complex environment.The network attack effect evaluation system can evaluate the advantages and disadvantages of attack methods and improve the ability of professional attackers.At the same time,the evaluation data can also provide improvement basis for security defense measures such as threat detection.This paper studies and analyzes the vulnerability of network security and the means of network attack,investigates the key technologies in the evaluation process,and analyzes their advantages and disadvantages.On this basis,aiming at the shortcomings of the traditional network attack effect evaluation model,this paper establishes a network attack effect evaluation model based on variable weight theory and optimizes the evaluation method.Then,this paper introduces variable weight theory to solve the problem of the limitation of constant weight on the evaluation index in the evaluation process.This paper designs an improved entropy weight method to solve the problem of inaccurate determination of special entropy weight by traditional entropy weight method and the problem of strong subjectivity in the evaluation process of traditional fuzzy analytic hierarchy process.The main work and innovations are as follows:Firstly,based on the analysis of the characteristics of information system security threats and network attacks,according to the establishment standards of the evaluation index system,this paper proposes a set of network attack effect evaluation index system for the attack process.Taking the life cycle of network attack as the basic division of the index,the attack initiation stage,attack action stage and attack result stage are evaluated respectively,which weakens the pertinence of the traditional evaluation index system and increases the coverage of the index system.The evaluation indicators meet the criteria of "collectable,calculable and standardized",which improves the availability of the evaluation index system.Secondly,this paper focuses on the weight calculation method in the evaluation process.In order to make up for the strong subjectivity of the traditional evaluation weight,this paper introduces the entropy weight method into the determination process of the evaluation weight,optimizes the shortcomings of the traditional entropy weight method,and puts forward a comprehensive evaluation method based on fuzzy analytic hierarchy process and improved entropy weight method.The comprehensive evaluation weight obtained through the information entropy of subjective experience and objective data is more scientific and reasonable.Then,on the basis of determining the constant weight,this paper introduces the variable weight theory,optimizes the variable weight function,and solves the problem of unsatisfactory evaluation results of network attack effect caused by the compensation mechanism between indicators through a reasonable variable weight method.Combining the calculation and evaluation results of TOPSIS algorithm,this paper proposes a network attack effect evaluation model based on variable weight theory.Simulation experiments show that the evaluation model is accurate and reasonable.Finally,based on the above work,this paper designs and implements a network attack effect evaluation system based on variable weight theory,divides the system into four subsystems,and expounds the design principle and mechanism of the core functions of the subsystem respectively.Through test cases,the rationality of the system function and the effectiveness of the evaluation model are verified.