Location Algorithm Of Validity Change Region Based On Game Theory

With the development of Internet technology,in order to deal with various possible new security problems,network security technology and various computer software and hardware have a trend of shorter and shorter iteration.In order to help architects avoid the redesign of complex models,research on change mining technology has emerged.Many of the existing change mining methods often start from the perspective of model association,and use structural analysis among activities in the model to find out the possible change regions after changes.Such methods mostly focus on optimizing the accuracy of the mining effect,the complexity and generalization of the algorithm,and are suitable for updating changes in general business processes.However,they ignore the effectiveness of the change region under specific circumstances,that is,mining the change region from a single perspective of the model structure has low credibility,and such problems are particularly prominent in the network security model.At the same time,in the field of network security research,game theory has become a powerful tool to study network attack and defense games in recent years.Most of the attack and defense research is based on game research.It can effectively analyze the contradictions and conflicts in the network security attack defense confrontation model,and deduce the optimal selection strategy of both sides.However,although there are many algorithms supporting different research directions in this field,more attention is paid to the research of existing vulnerabilities and security threats,aiming to identify threats through higher accuracy and lower cost detection methods.There is less research on new attack utilization points and unknown vulnerabilities.Although the existing attack methods for unknown new vulnerabilities are mostly combined with machine learning related research,such research requires a large number of multi-faceted security logs as the basis for learning in order to better predict and analyze the next attack target.Firstly,based on the concepts of Petri nets and change mining,this paper proposes a technology that uses the security standards between model activities to build change associations through the occurrence of specific security events;Then,using the global profit quantification method of static game,a new effective change region location algorithm based on global game strategy is proposed;Finally,a new game Petri net model based on multi-agent reinforcement learning is proposed by using dynamic game theory.The main research contents are as follows:(1)In the aspect of the combination of information security attack and defense field and change mining field,in order to break the technical limitation of traditional change mining,which focuses on the association between model structures,a change mining method specifically applied to information security attack and defense is proposed,that is,through the occurrence of specific security events,using the security standards between model activities to construct associations rather than using a single model to control the flow structure.The purpose of this paper is to introduce information security specification into change mining,so that the change region of mining can meet the expected security indicators.(2)After ensuring the feasibility of change research under specific security events,this paper proposes a method to verify the effective change region of attack and defense model based on global game strategy.The global benefit quantification method of static game is used to extend the research dedicated to specific security events to all attack and defense processes,so as to ensure the wide applicability of the research and effective combination with change mining.(3)In terms of the accuracy of real game process simulation and effective change region mining,in order to make the game process more realistic,so as to accurately reflect the most real strategy selection results of attackers in the defense system to ensure the accuracy of the change region validity verification.A new game Petri net model based on multi-agent reinforcement learning is proposed,which is suitable for mining changes in existing game scenarios.This model accurately analyzes the two sides of the game in the current situation through dynamic game,and selects the best way for the next step according to the other side’s strategic thinking,that is,it realizes the transformation from static analysis to dynamic analysis,so that the validity verification of the change domain has more credibility.Figure [23] Table [24] Reference [74]...