Formulating mathematical models for accurate approximation of malicious propagation in a network is a difficult process because of our inherent lack of understanding of several underlying physical processes that intrinsically characterize the broader picture. The aim of this paper is to understand the impact of available information in the control of malicious network epidemics. A 1-n-n-1 type differential epidemic model is proposed, where the differentiality allows a symptom based classification. This is the first such attempt to add such a classification into the existing epidemic framework. The model is incorporated into a five class system called the DifEpGoss architecture. Analysis reveals an epidemic threshold, based on which the long-term behavior of the system is analyzed. In this work three real network datasets with 22002, 22469 and 22607 undirected edges respectively, are used. The datasets show that classification based prevention given in the model can have a good role in containing network epidemics. Further simulation based experiments are used with a three category classification of attack and defense strengths, which allows us to consider 27 different possibilities. These experiments further corroborate the utility of the proposed model. The paper concludes with several interesting results.
展开▼
