Lab Home | Phone | Search
Center for Nonlinear Studies  Center for Nonlinear Studies
 Colloquia Archive 
 Postdoc Seminars Archive 
 Quantum Lunch 
 CMS Colloquia 
 Q-Mat Seminars 
 Q-Mat Seminars Archive 
 Kac Lectures 
 Dist. Quant. Lecture 
 Ulam Scholar 
 Summer Research 
 Past Visitors 
 History of CNLS 
 Maps, Directions 
 CNLS Office 
Thursday, September 10, 2009
10:00 AM - 11:00 AM
CNLS Conference Room (TA-3, Bldg 1690)


An Adventure in Computer and Computational Epidemiology

Joseph Kong

Given a computer network, how does a trojan help the spread of other malwares (malicious softwares)? On sexual contact networks, how do sexually transmitted diseases (STDs) help the Human Immunodeficiency Virus (HIV) spread? Moreover, reducing HIV incidences through STD treatment is part of HIV prevention programs since the 1990ís as advocated by the World Health Organization (WHO). In fact, understanding the synergistic spread of STD and HIV is explicitly listed as an open problem in a Nature Reviews article recently. Faced with such threats in computer security and public health, we are in need of mathematical models to better understand these complex epidemic dynamics. In this talk, I will show that these problems, which are from seemingly completely different domains of computer security and public health, can be modeled under a common framework as the spread of two synergistic pathogens on a network. The synergy comes from the local interaction, where if a node is infected with the trojan (or an STD), the node becomes more susceptible to other malware (or HIV) by an amplification factor. We develop an analytical methodology for modeling the complex dual-pathogen epidemic spread on networks. Our analytical results show how both the increased susceptibility engendered by trojans, as well as, the heterogeneous structure of the underlying network work together to significantly reduce the epidemic thresholds for synergistic virus and ignite epidemics. In addition, we show that employing the strategy to fight viral epidemics indirectly by combating the trojan (or the STD) is likely to be ineffective, if only marginal reduction in trojan (or STD) prevalence can be achieved. We demonstrate the validity of our results on large-scale synthesized graphs and real-world networks that include email, P2P, and online social networks, where malwares are known to spread on a recurrent basis.

Host: Luis Bettencourt, T-5