A Multistage Dynamic Defense Method for Evolutionary Games

This paper proposes a multistage dynamic defense method for evolutionary games to address the challenge of accurately sensing unknown and homeopathic attacks on each node in the network and effectively accomplishing dynamic security. The method combines the replicator dynamic equations and the characteristics of the attackdefense game adversarial process to establish a discrete multistage offense-defense game model, and then quantifies the gains and equilibrium solutions for the model, simulates the multistage offense-defense game process under the high selection rate of the attack and defense strategy, and calculates the maximum objective function values of both sides. By analyzing these function values, the security situational awareness of the whole network nodes is completed to predict future security situations and system maintenance. Experimental comparisons show that the model approach has high operational efficiency and better defensive performance to ensure system integrity and other advantages. © 2022, International Journal of Network Security. All Rights Reserved.