Temporal Point Process Based Information Dissemination Over Cyber-Physical Interacting Networks

Information dissemination generally occurs simultaneously over both physical and cyber networks. However, it is highly challenging to design information dissemination mechanisms due to the heterogeneity and interactive characteristics of cyber-physical interacting networks (CPINs). In addition, the stochasticity of information forwarding and forgetting over these networks has further aggravated this situation. In this paper, we propose a temporal point process based method to address information dissemination over CPINs. We first construct a CPIN model to capture the heterogeneity and interactive characteristics of physical and cyber networks. We then build a temporal point process based information dissemination model (TPP-IDM) to characterize the stochastic evolution of information dissemination over CPINs. This model captures the dynamic stochastic nature of the information dissemination accurately over time by computing the probability distribution of the sizes of information dissemination over CPINs, which greatly reduces the bias caused by the traditional deterministic solutions. Furthermore, we develop an intensity learning based on maximum likelihood estimation (MLE-IL) method to improve the prediction accuracy of TPP-IDM by learning the most appropriate intensities. Finally, experimental results show that our method can predict the size of information dissemination over CPINs more accurately.