- talks - lists.cs.princeton.edu

Talk on Thursday -12/11/2008
by Ginny Hogan 09 Dec '08

09 Dec '08

Eva Tardos, Cornell University will present her Talk. "Games in Networks: the price of anarchy and learning"- at 4:15pm Computer Science Room 105.

1 0

H Ringberg preFPO - correction
by Melissa Lawson 08 Dec '08

08 Dec '08

The committee should be; Jennifer Rexford, Mike Freedman, Nicholas Duffield (AT&T), Rob Calderbank, and Brian Kernighan. --Melissa

1 0

H Ringberg preFPO
by Melissa Lawson 08 Dec '08

08 Dec '08

Haakon Ringberg will present his preFPO on Friday December 12 at 10AM in Room 302. The members of his committee are: Jennifer Rexford, advisor; Nicholas Duffield (AT&T), Mike Freedman, Vivek Pai, and Rob Calderbank. Everyone is invited to attend his talk. His abstract follows below. ---------------------------------------------- Unwanted traffic is a major concern in the Internet today. The algorithms that attempt to identify this unwanted traffic are called network anomaly detectors, and their task is challenging in part because many network attacks mimic the behavior of normal traffic. The victims of these attacks should therefore collaborate by correlating their attack data, under the assumption that a given malicious host is likely to affect more than one victim. The major obstacle to such collaboration is that the corporations who manage these networks are hesitant to openly share information about the hosts (customers) that use their services. In this thesis, we propose and evaluate the feasibility of an architecture for privacy-preserving collaborative anomaly detection. First, we study the potential gain from collaboration on traces from operational networks. We do this by calculating the fraction of detected network anomalies (viz., IP scans, port scans, and DoS attacks) that could have been mitigated in these traces if some subset of the victim hosts collaborated to block attackers. Second, we propose and evaluate the efficiency of a novel cryptographic protocol that allows victims to collaborate to identify the hosts sending unwanted traffic. The protocol allows each participant to submit a set of IP addresses that they suspect might be engaging in said activity. The protocol preserves privacy because it never reveals who suspected whom, and a submitted IP address is only revealed when more than n participating networks suspect it. Finally, we close the loop by evaluating the feasibility of making a commonly-used single vantage-point anomaly detector faster and more scalable. Specifically, we propose and evaluate a system that leverages machine learning to bring the classification intelligence of the signature- and packet-based intrusion detection system (IDS) Snort to the IP flow realm. We demonstrate that our system can effectively learn to classify many Snort alarms on IP flows. In sum, the thesis presents a complete collaborative anomaly detection architecture, from the high-confidence detection of unwanted traffic at single networks to a privacy-preserving protocol that allows these networks to collectively increase their confidence in their detections.

1 0

Talk Today 4:15 - COS 105
by Ginny Hogan 03 Dec '08

03 Dec '08

Vladlen Koltun, Stanford University will present his talk, "Computer Graphics as a Telecommunication Medium " ,at 4:15pm today Computer Science Room 105 - Tea will be at 3:30 -

1 0

Colloquium: Vladlen Koltun on Wednesday, Dec 3, 4:00-5:00pm
by Michael J. Freedman 01 Dec '08

01 Dec '08

Hi, This Wednesday at 4:00pm, Vladlen Koltun from Stanford University will be giving the department colloquium. The talk is located in the small auditorium (CS Room 105). I looking forward to seeing you there. Thanks, --mike Abstract -------- I will argue that the primary contribution of computer graphics in the next decade will be to enable richer social interaction at a distance. The integration of real-time computer graphics and large-scale distributed systems will give rise to a rich telecommunication medium, currently referred to as virtual worlds. The medium provides open-ended face-to-face communication among ad-hoc groups of people in custom environments and decouples shared spatial experiences from geographic constraints. I will outline a research agenda for enabling and advancing the medium. The three driving themes are system architectures, content creation, and interaction. System architectures aim to support the medium at planetary scale. Content creation aims to enable untrained participants to create high-quality three-dimensional content. Interaction aims to make virtual world communication seamless and natural. I will demonstrate preliminary results in each area. Bio: Vladlen Koltun is an Assistant Professor of Computer Science at Stanford University. He directs the Virtual Worlds Group, which explores how scalable virtual world systems can be built, populated, and used. His prior work in computational geometry and theoretical computer science was recognized with the NSF CAREER Award, the Alfred P. Sloan Fellowship, and the Machtey Award.

1 0

1:30-2:30pm Fri Dec 5 talk on Effective Diagnosis of Routing Disruptions from End Systems
by Jennifer Rexford 30 Nov '08

30 Nov '08

Invited talk, note the unusual room (CS 102). Please let me know if you would like to meet with Morley Mao while she is here. Speaker: Professor Morley Mao, U. Michigan Title: Effective Diagnosis of Routing Disruptions from End Systems Date/time: 1:30-2:30pm Friday December 5 Room: 102 in the CS building Abstract: Internet routing events are known to introduce severe disruption to application performance. So far effective diagnosis of routing events has relied on proprietary ISP data feeds, resulting in limited ISP-centric views not easily accessible by customers or other ISPs. In this work, we propose a novel approach to diagnosing significant routing events associated with any large networks from the perspective of end systems. Our approach is based on scalable, collaborative probing launched from end systems and does not require proprietary data from ISPs. Using a greedy scheme for event correlation and root cause inference, we can diagnose both interdomain and intradomain routing events. Unlike existing methods based on passive route monitoring, our approach is able to measure the impact of routing events on end-to-end network performance as well. We demonstrate the effectiveness of our approach by studying five large ISPs over more than three and half months. We validate the accuracy of our approach by comparing with the existing ISP-centric method and with events reported on NANOG mailing lists. Our work is the first to scalably and accurately diagnose routing events associated with large networks purely from end systems. Bio: Z. Morley Mao is an assistant professor of Computer Science and Engineering at the University of Michigan. Morley received her Ph.D., M.S., and B.S. from the University of California at Berkeley. She is a recipient of the NSF CAREER award in 2007, an IBM Faculty Award in 2008, and was recently named the Morris Wellman Faculty Development Professor. Her research interests include network measurements, routing protocols, distributed systems, and network security.

1 1

Yi Wang preFPO
by Melissa Lawson 26 Nov '08

26 Nov '08

Yi Wang will present his preFPO on Wednesday December 3 at 1:30 PM in Room 401. The members of his committee are: Jennifer Rexford, advisor; Michael Freedman and Kobus van der Merwe (AT&T), readers; Larry Peterson and Ed Felten, nonreaders. Everyone is invited to attend his talk. His abstract follows below. ------------------------------------ Title: A Principled Approach to Managing Routing in Large ISP Networks Abstract: Today's Internet Service Providers (ISPs) face a number of challenges in introducing changes to their networks at different frequencies. Firstly and most frequently, ISPs need to perform daily planned maintenance to keep their networks running healthily. However, despite of the importance and frequency of the planned maintenance operation, there is no existing method that can handle such operations without causing packet losses and routing protocol re-convergence. Secondly, once in a while, ISPs have to adjust their routing policies to meet changing needs and make flexible trade-offs between different policy objectives (such as business relationships, performance, security, etc.). However, today's network architecture forces administrators to realize network- wide routing policies through router-level configurations and makes it very difficult (if not impossible) to make flexible trade-offs. Thirdly, ISPs also have the incentive to make a one-time change (transition) from the current restrictive route-selection model, in which each router selects a single best route for all neighbors, to a neighbor- specific route-selection model, in which each router can provide value-added customized route- selection services to different neighbors. However, it is currently unknown if such transition to a more flexible route-selection model may cause the Internet routing to oscillate. In this thesis, we take a principled approach to solving these "change-related" challenges and propose three abstractions that guide the design and implementation of our system solutions. First, we propose the abstraction of a "neighbor-specific route selection problem" and a corresponding "Neighbor-Specific BGP" (NS- BGP) model that capture the requirement of customized route selection for different neighbors. We prove that the Internet is guaranteed to remain stable even if individual ISPs make transition to this more flexible route-selection model incrementally. Second, we propose the abstraction of "policy configuration as a decision problem of reconciling multiple objectives". Guided by this abstraction and the Analytic Hierarchy Process, a decision-theoretic technique for balancing conflicting objectives, we designed and implemented a prototype of an extensible routing control platform (Morpheus) that enables an ISP to select routes for different neighbors individually and make flexible trade-offs among policy objectives through a simple and intuitive configuration interface. Finally, we propose the abstraction of the separation between "physical" and "logical" configurations of routers, which leads us to the design and prototype implementation of "virtual router migration" (VROOM), a new, generic technique to simplify/enable a broad range of tasks that require the network to adapt in certain ways, from planned maintenance to reducing power consumption. Collectively, the contributions of the thesis provide simple and effective solutions for an ISP to autonomously introduce changes to its network at various frequencies without sacrificing global routing stability.

1 0

Bill Press speaking at PICASso today, 12:30
by Elena Zaslavsky 24 Nov '08

24 Nov '08

Bill Press of "Numerical Recipes in C" speaking at PICASso today. From: picasso-bounces(a)lists.cs.princeton.edu [mailto:picasso-bounces@lists.cs.princeton.edu] On Behalf Of Cecile Huttenhower Sent: Monday, November 24, 2008 9:12 AM To: picasso-faculty(a)lists.cs.princeton.edu; picasso(a)lists.cs.princeton.edu Subject: [PICASso] TODAY - What Is a Statistical Model and When Do I Need One? ** PICASso: ** Program in Integrative Information, Computer and Application Sciences ** www.cs.princeton.edu/picasso ** ** TALK TODAY -- Monday, November 24, 2008 Interdisciplinary Computational Seminars http://www.cs.princeton.edu/picasso/computational_lunch.php TITLE: What Is a Statistical Model and When Do I Need One? SPEAKER: Bill Press, CS, U. Texas TIME: Monday, November 24, 2008 Seminar begins at 12:30 p.m. (lunch provided ~12:20) LOCATION: Room 104, Computer Science, Princeton University ABSTRACT: Whether by differential equations (ODEs or PDEs) or algebraic systems of equations, computational science traffics mainly in deterministic models of physical systems. While we sometimes make the mistake of thinking that the model "is" the physical system, it is always only an abstract, and approximate, representation. Statistical models are a different kind of abstraction, not for the deterministic part of a system, but for how it interacts with physical stochasticity and/or with Bayesian priors expressing imperfect knowledge. This talk compares statistical and deterministic models, using examples drawn from cosmology and bioinformatics. SIGN UP FOR THE PICASSO MAILING LIST: ===================================== If you would like to be kept informed of computationally-oriented events in (and around) Princeton, please SUBSCRIBE to the PICASso mailing list by visiting https://lists.cs.princeton.edu/mailman/listinfo/picasso. This page also contains information on how to UNSUBSCRIBE. PLEASE FORWARD THIS MESSAGE TO OTHER COMPUTATIONALLY-ORIENTED RESEARCHERS WHO MAY BE INTERESTED IN THIS EVENT, OR FUTURE PROGRAMS. THANKS!

1 0

D Xiao preFPO
by Melissa Lawson 14 Nov '08

14 Nov '08

David Xiao will present his preFPO on Friday November 21 at 1PM in Room 402. The members of his committee are; Boaz Barak, advisor; Avi Wigderson (IAS) and Russell Impagliazzo (IAS), readers; Sanjeev Arora and Rob Schapire, non-readers. Everyone is invited to attend his talk. His abstract follows below. New Connections Between Cryptography and Learning Cryptography and learning are intuitively inverse tasks: the first deals with obscuring information while the second aims to extract hidden information. Not surprisingly, many relationships were discovered early on between the two: a secure encryption system implies that learning how to decipher encrypted messages is hard, while on the other hand extremely powerful learning algorithms can potentially be used to break the security of cryptographic protocols. In this thesis, we deepen the understanding of the relationship between the complexity of these two tasks and use their relationship to inform the design of secure cryptographic protocols. In our first set of results, we investigate under what assumptions learning can be proven to be hard for polynomial-time algorithms. First we show that the assumption ZK != BPP (and in particular the worst-case hardness of problems such as Graph Isomorphism and Quadratic Residuosity) suffices to imply that learning is hard. We then show that on the other hand the weaker assumption P != NP is unlikely to imply that learning is hard via standard proof techniques. In our second result, we study secure fault localization in Internet routing. In this problem, a sending router and a receiving router wish to discover if there are any adversaries along the path between them causing faults, i.e. that are dropping or modifying packets. We build a formal model of security for this problem and show that in order to achieve security, the intermediate routers must share keys and perform cryptographic operations. This suggests that this security goal inherently requires expensive computation at all nodes in the network, not just the sender and receiver. Our result follows from applying a specific learning algorithm to breaking protocols that do not satisfy the requirements of keys and/or cryptographic operations at each intermediate router.

1 0

Syslunch presentation by Marc Shapiro (NOW!)
by Sapan Bhatia 11 Nov '08

11 Nov '08

Marc Shapiro is a director of research with INRIA and the chairmain of Eurosys, the ACM SIGOPS chapter of Europe. He is going to give a talk in today's systems lunch: Speaker: Marc Shapiro, INRIA and LIP6 Title: Telex: A Principled Platform for Peer-to-Peer Collaboration Abstract: Developers of co-operative applications encounter difficult issues of distributed state, disconnections, and conflicts. Our approach is to factor these system issues out from application logic, but we recognise that they cannot be entirely divorced. The Telex platform that we designed is application-independent, but parameterised by application-specific information. An application provides Telex with actions (operations) and constraints (invariants). In return, Telex takes care of replication and persistence, drives application progress while verifying its invariants, and ensures that replicas eventually agree on a correct, common state. Time permitting, I will present a number of contributions. I show by example how application design proceeds from high-level invariants to actions and constraints, and how the platform ensures application safety and liveness. The main data structure of Telex is a large, replicated, very dynamic graph; I discuss the engineering trade-offs for such a graph and our solutions. I describe our how Telex ensures, in the background, that replicas converge to a safe state. Finally, I describe experimental evaluation of the Telex platform. Joint work with Lamia Benmouffok, Jean-Michel Busca, Joan Manuel Marquhs, Marc Shapiro, Pierre Sutra, Georgios Tsoukalas

1 0