Selection Of Network Defense Strategies Based On Evolutionary Game Model

With the development of the Internet and the ever-increasing level of informatization,the cyberspace is becoming increasingly complex and the situation of security threat is becoming increasingly severe.For different network security environments and defense requirements,different defense strategies and technical methods are needed.Evolutionary games have the advantage of improving their offensive and defensive capabilities based on learning from others.Therefore,this paper combines the evolutionary game theory with the reality of network attack and defense,to construct a network attack-defense evolutionary game mathematical model,and to study the network defense decision-making method under different scenarios of attack and defense,which can solve the problem of network attack behavior prediction and optimal defense strategy selection.The main contents are as follows:1.For the problem that the traditional game theory based on completely rational condition is inconsistent with the actual attack and defense,we introduce the evolutionary game theory with bounded rationality,to construct the basic attack-defense evolutionary game model based on non-cooperative evolutionary game theory.The attack and defense strategy set in the model is expanded from simple 2?2 to m?n,which improves the universality of the model and method.The method of solving the evolutionary stable equilibrium of this model is given.The evolution law of attack and defense behaviors under limit of bounded rationality is analyzed and summarized,which can provide basic model and method for network defense decision-making.2.Since the network attack and defense is a random process of multi-stage state transition,we combine the basic attack-defense evolutionary game model with Markov decision theory to construct a multi-stage Markov attack-defense evolutionary game model and design an optimal defense strategy selection algorithm.Taking the total discounted return of the game as an objective function,a discount factor ? is introduced to describe the reducing of returns in the multi-stage attack-defense process,and the game equilibrium is analyzed and solved,which can be adapted to select the optimal defense strategy for multi-stage attack-defense game process.3.For the real-time randomness of attack-defense process,based on the basic attack-defense evolutionary game model,we introduces the nonlinear Itó stochastic differential equations by using Gaussian white noise,constructs a stochastic evolutionary game model of network attack and defense and designs an algorithm of optimal defense strategy selection with real-time stochastic characteristics.The Taylor expansion is used to solve the attack-defense stochastic differential equations,and the stability of attack-defense strategies is analyzed according to the stability discriminant theorem of stochastic differential equations.The influence of different intensity random disturbances on the evolution process of attack and defense is analyzed by the experiment,which improves the accuracy of the defense decision-making method.4.For the strategic incentive relationship between the individuals in the network attack and defense,we use the incentive coefficient to improve the replicator dynamic equation based on the basic attack-defense evolutionary game theory,and to optimize the learning mechanism of attack-defense evolutionary game models.By solving the model equilibrium,we analyze the stability of the attack-defense evolutionary process by using the local stability analysis method of the Jacobian matrix.The optimal defense strategies under different conditions are obtained.