Design And Implementation Of User Fine-Grained Behavior Simulation System For Cyber Range

In the current Internet era,more and more network security problems are troubling people.As the demand for conducting network offensive and defensive drills,security tests and other experiments under real Internet scenarios has grown,Cyber Range platforms emerge as the times require.In the Cyber Range,user fine-grained behavior simulation is an important part of foreground behavior simulation.In the past research,user fine-grained behavior simulation was often based on the way of script writing to generate the behavior in a specific scene or based on user behavior habits to simulate in a probabilistic way.However,both of them have common shortcomings: they can't achieve the "simulation" and universality in the simulation.Therefore,this paper will focus on the fine-grained behavior of network users,that is,individual user behavior.First of all,this paper analyzes the fine-grained behavior of users,summarizes the current research status and research methods,determines the needs of this behavior simulation system.Based on two research routes of user software behavior simulation and time behavior simulation,the technology and system flow required in this paper are established.Then,this paper simulates the behavior sequence between multiple applications and the operation sequence within a single software and gives different machine learning algorithm models according to the two simulation methods.Behavior simulation between multiple applications uses Leak GAN model,but when the operation behavior is simulated in software,not all event click behaviors generated by Leak GAN network are realizable click sequences.Therefore,this paper introduces a reinforcement learning model and creatively combines Q-Learning with a generational confrontation network with information leakage to achieve user software behavior simulation.Through experiments in this paper,it can be concluded that model adopted in this paper can fit the real user behavior on the Internet with a probability of data distribution correlation coefficient of 99%.In the simulation of time behavior,ARMA algorithm model is used to fit the user's time behavior based on time series.After training,the model can predict user's future time behavior.Experiment effect will be verified by many ways,and the results show that this model can provide a high degree of fitting model with real time data.Finally,based on the research results of this pap er,a user fine-grained behavior simulation system platform is designed and implemented.According to the user's needs,the system can carry out targeted behavior simulation process and show the fitting results to the user in a visual way.