CSL Women in Engineering Seminar

Title:  Optimal Malware Attack and Defense in Mobile Wireless Networks

Speaker:  Professor Saswati Sarkar

Department of Electrical Engineering

University of Pennsylvania

Date:  Wednesday, April 25, 2012

Time:  2:00-3:00 p.m.

Location:  141 Coordinated Science Lab

Abstract:  Malicious self-replicating codes, known as malware, pose substantial threat to wireless networks, and the economic viability of the investments directed towards wireless infrastructure is contingent on the design of effective countermeasures. The first step towards this end is to anticipate malware hazards, and understand the threat before the attacks are actually launched. In this talk, we quantify the fundamental limits on the damage that the malware can inflict by optimally choosing its actions. Such limitations arise because the capabilities of the malware are limited by the resource constraints of wireless networks which the malware utilizes as well to propagate the contagion. Specifically, malware needs to decide how best to utilize the limited battery reserves of the host in scanning the media and transmitting packets that carry the contagion, as once the host's battery is depleted, the network incurs a cost, but so does the malware as  the host can no longer be utilized to propagate the contagion and also in fulfilling its subversive ends. We formulate the maximization of the overall damage inflicted on the network as an optimal control problem, and subsequently identify structural properties of the optimal actions of the malware using Pontryagin's Maximum Principle. We show that the malware can inflict the maximum damage by choosing simple bang-bang control functions with at most two jumps. The network can launch counter-measures through power control based quarantining strategies and also by fetching security patches that immunize the vulnerable and heal the infected hosts. Such power control strategies however deteriorate the quality of service in the network, and the transmission of patches consume valuable transmission resources such as spectrum and energy. We formulate the optimal countermeasure selection as optimal control problems and identify structural properties of the optimal solutions. The optimal strategies again turn out to be  bang-bang functions with at most two jumps, and should therefore be readily implementable in resource constrained wireless devices. We next formulate the interactions between the attack and the defense as a dynamic game and prove that the robust defense retains its simple structures. We finally show how the formulations capture the essence of a diverse set of problems as remote as advertising systems, health-care, peer-to-peer networks, delay-tolerant networks etc., and in turn provides elegant solutions in each of the above contexts.

Biography:  Saswati Sarkar received Master of Engineering from the Electrical Communication Engineering Department at the Indian Institute of Science, Bangalore in 1996 and PhD from  the Electrical and Computer Engineering Department at the University of Maryland, College Park, in 2000. She joined the Electrical and Systems Engineering Department at the University of Pennsylvania, Philadelphia as an Assistant Professor in 2000 where she is currently an Associate Professor.  She received the  Motorola gold medal for the best masters student in the division of electrical sciences at the Indian Institute of Science and  a National Science Foundation (NSF)  Faculty Early Career Development Award in 2003.  She was an  associate editor of IEEE Transaction on Wireless Communications from 2001 to 2006, and is currently an associate editor of IEEE/ACM Transactions on Networks. Her research interests are in stochastic control, optimal control, security,  resource allocation, dynamic games and economics of networks and sustainable development.