Research on Network Security Situation Awareness and Dynamic Game Based on Deep Q Learning Network

Abstract

In today's increasingly complex network environment, increasingly severe network attacks, and real-time dynamic changes in offensive and defensive scenarios, network security technology has also evolved from passive security to active security technology, and expanded from analyzing unilateral security elements to comprehensively analyzing overall network security. Aiming at the problem of inaccurate assessment results due to the lack of comprehensive analysis of threat information, protection information and environmental information in existing assessment methods, this paper proposes a defensive random game model. This model analyzes threat propagation and establishes a threat propagation access relationship network, and then establishes a random game model for the game process of threat action and protection strategy implementation to solve the mixed strategy Nash equilibrium. The model comprehensively analyzes the dynamic changes of the security situation elements, ignoring the short-term situation elements that will not change, such as topology structure, service information, etc., and mainly analyzes the dynamic changes of attack information, vulnerability information, and defense measures, and predicts from the host and network levels. The experimental verification shows that the prediction method in this paper can improve the prediction accuracy and is more in line with the offensive and defensive scenarios.