Research On Moving Target Defense Decision-making Method Based On Dynamic Attack-defense Game Model

With the rapid development of network technology and the applyment of all kinds of network infrastructures,network system has become the key support for the effective running of different industries in society.However,network security accidents happening frequently has proved that,cyber threats has caused great damage to people's living and the network defense ability has to be improved.Especially,the network attack technology progresses more rapidly,the existing passive defense technology is hard to deal with the complex and changeable network attacks effectively and the defense part is stuck in a terrible situation.To change this situation,we need to call for a new active defense technology to ensure and enhance the security of the network system,which should be active and controllable.As an active defense technology,Moving Target Defense(MTD)can effectively improve the ability and effectiveness of defense.In the application field of Moving Target Defense technology,how to choose the optimal defense strategy under different attack-defense scenarios is the key to the defenders,and it has also been focused by researchers in recent years.As the attack-defense confrontation process of Moving Target Defense has the same characteristics as game theory,such as goal opposition,relationship non-cooperation and strategy dependence,we combine Moving Target Defense with non-cooperative game theory to research on the decision-making problems.We propose the corresponding game model under different attack-defense scenarios and design the corresponding optimal defense strategy selection algorithm.The main work includes the following aspects:1.As the network attack-defense scenarios generally have the game-theoretical characteristics,we use the signal game theory to describe the characteristics of Moving Target Defense technology,and explore the defense decision-making method based on attack-defense game analysis.Firstly,we formally define the Moving Target Defense strategy from the perspective of attack surface transferring and detection surface changing.Secondly,we construct an interactive structure with the defender as the signal sender and the attacker as the signal receiver.Meanwile,we propose the attack-defense signal game model and design the optimal defense strategy selection algorithm based on Bayesian equilibrium solution method.Finally,we summarize the characteristics of the active defense combined with signal game and Moving Target Defense technology.2.In the actual Moving Target Defense confrontation process,the change of attack-defense strategy and the change of attack-defens environment will lead to random changes of the stable state.Furthermore,the attack-defense confrontation process shows the multi-stage characteristic,and the single-stage signal game model can not describe the multi-stage confrontation process.In order to solve the above problems,we construct a MTD model based on multi-stage Markov signal game.Meanwhile,we take the total discounted payoff of as the objective function,to solve the problem of payoff attenuation in the multi-stage process,we use a discount factor to deal with the total payoff.On the basis of solving the multi-stage equilibrium,we design the optimal strategy selection algorithm.Finally,the simulation verifies the effectiveness and feasibility of the model and method.3.With the increasingly fierce network security confrontation,the network attack-defense process is becoming real-time and high-frequency,and the running state of the network system can not avoid the influence of random factors.To solve the above problem,we propose a Markov differential game model to deduce the attack-defense process of Moving Target Defense,and take the discounted total payoff function of attack and defense game as the objective function.Then we analyze the saddle point equilibrium path,and design the optimal defense strategy selection algorithm.Finally,we verify the effectiveness of the model and method by simulation experiments.