סמינר המחלקה להנדסה ביו רפואית 06.11.16

Speaker: Tsvika Dagan

Ph.D. student under the supervision of Prof. Avishai Wool

Sunday, November 6^th, 2016 at 15:30
Room 011, Kitot Bldg., Faculty of Engineering

Parrot, a software-only anti-spoofing system for the CAN bus

Background

Modern cars have multiple dedicated computers under the hood called ``electronic control units'' (ECUs). These ECUs control all aspects of the car's operation: from the engine, breaking and steering controls to the car's entertainment systems. The ECUs are connected to each other in a network that typically uses the CAN bus protocol. CAN bus is a simple serial protocol, with absolutely no security components: it was designed under the assumption that all ECUs are legitimate, trustworthy, and operating according to their specifications. However, over the last few years researchers have shown that many ECUs are vulnerable to attack. Since CAN bus in itself is so naive, any attack on one ECU can immediately allow lateral movement, to attack other, more critical, ECUs; the subverted ECU can trivially spoof (masquerade as) any other ECU and cause significant damage. Replacing CAN bus with a more robust technology is probably a good idea. However, due to the huge investment made by manufacturers, and the decades it takes until old cars are scrapped, it is an important goal to improve the security stance of cars within the limitations of CAN bus. Thus finding methods to block the lateral movement, from the originally compromised ECU to others, is a major step in this direction.

Abstract

In this talk I will describes a novel anti-spoofing system for in-car CAN bus networks. If an attacker compromises one of the car's electronic control units (ECUs), and from there tries to attack another, more critical, ECU, the Parrot system blocks this lateral movement. Unlike previous firewall-based solutions or cryptography-based solutions, the attack messages are identified and destroyed by the legitimate message ID's owner. Our method does not merely drop messages that are non-conforming with policy: the Parrot defense typically disconnects the compromised ECU from the bus. And unlike previous solutions, that require a modified controller (since they violate the CAN bus protocol), our method is able to shut down the attacker while obeying the protocol rules. Hence, the Parrot defense can be added as a software-only patch to any standard ECU. We implemented the Parrot system and tested its behavior in detailed experiments. With CAN controllers that are able to transmit fast enough we were able to disable the attacking ECU in most experiments. For slower controllers, we showed a successful alternative.

(The talk will be given in English)

Speaker:     Prof. V.S. Borkar
                      Dept. of EE, IIT, Powai, Mumbai, India

Monday, November 21^st, 2016
15:00 - 16:00

Room 011, Kitot Bldg., Faculty of Engineering

Nonlinear gossip

Abstract

A model for distributed algorithms due to Tsitsiklis can be viewed as a regular perturbation of the classical gossip iteration, perturbed by a Robbins-Monro type stochastic approximation iterate. We consider variants wherein the gossip part is replaced by a suitable nonlinear operation and characterize the asymptotic behavior of the iterates. The general results can be specialized to particular situations to draw more specific conclusions.‏

(The talk will be given in English)

Speaker:     Prof. Michael Bronstein
                      USI / TAU / Intel

Wednesday, November 2^nd, 2016
15:00 - 16:00

Room 011, Kitot Bldg., Faculty of Engineering

Partial deformable shape correspondence

Abstract

Finding correspondence between 3D shapes is one of the prototypical problems in computer graphics, geometric processing, and vision. Different flavors of this problem arise in applications ranging from texture and animation to marker-less motion capture . The recent progress in the development of commercial real-time 3D scanning technology has brought the need for fast, accurate, and reliable correspondence methods capable of dealing with real-world noise and artifacts. Particularly hard settings of the correspondence problem include non-rigid correspondence (where the shapes are allowed to undergo deformations), partial correspondence (where a subset of the shape has to be matched to its deformed full version), and geometric and topological noise (the latter arising, for example, due to occluded parts in the acquisition process).

In this talk, I will show some recent results on computing partial correspondence between deformable shapes expressed in the spectral domain. Our main theoretical result is a perturbation analysis of the Laplacian operator, giving a bound on the change in its eigenvectors as a result of part removal. Based on this observation, we develop a generalization of functional maps capable of dealing with partial correspondence, clutter, and topological noise in very challenging settings.

(based on joint works with E. Rodola', L. Cosmo, J. Masci, A. Torsello, D. Cremers, O. Litany, A. Bronstein)

Bio

Michael Bronstein is an associate professor of Informatics at USI Lugano (Switzerland) and associate professor of Applied Mathematics at TAU. He is also a Principal Engineer at the Intel Perceptual Computing group. Michael got his Ph.D. in Computer Science (2007) from the Technion. His main research interests are theoretical and computational methods in spectral and metric geometry and their application to problems in computer vision, pattern recognition, computer graphics, and machine learning. His research appeared in international media and was recognized by numerous awards. In 2012, Michael received the highly competitive ERC starting grant. In 2014, he was invited as a Young Scientist to the World Economic Forum, an honor bestowed on forty world's leading scientists under the age of 40. Besides academic work, Michael is actively involved in the industry. He was the co-founder of the Silicon Valley start-up company Novafora, where he served as VP of technology (2006-2009), responsible for the development of algorithms for large-scale video analysis. He was one of the principal inventors and technologists at Invision, an Israeli startup developing 3D sensing technology acquired by Intel in 2012 and released under the RealSense brand. This technology can now be found in new generation computers from all the major brands.