Predictive analytics for cyber-attack timing in power Internet of Things: A FlipIt game-theoretic approach

Abstract

The Power Internet of Things is increasingly challenged by complex security threats, which highlights the importance of developing predictive models for attack behaviors and implementing targeted defensive measures. In response to the challenges of traditional network attack predictors, which fail to forecast attack times, and the lack of integration of attack–defense confrontation processes in these methods, this paper proposes a novel method based on FlipIt games. This method not only aims to predict the timing of attacks but also considers the dynamic nature of attack–defense confrontations, adaptively suggesting defense strategies. The paper constructs a strategy evolution probability model for both attackers and defenders based on the infectious disease model and analyzes the interactive process of confrontation between them. It then establishes a network attack time prediction model based on FlipIt games and proposes an attack time strategy prediction algorithm that incorporates exponential probability distribution into the revenue calculation process to predict the timing of network attacks. Finally, by setting up simulation environments and conducting simulations using MATLAB, the proposed model is verified to effectively analyze the attack time of the virus and provide corresponding defense strategies in real-time. In the process of attack and defense, to achieve optimal attack benefits, attackers should prioritize selecting low-frequency and low-cost attack strategies.