Recently I had the pleasure to attend the 24th USENIX Security Symposium and its co-located Workshop on Offensive Technologies (WOOT) in Washington, D.C. The workshop has received quite some attention this year, 57 submissions of which 19 have been accepted, so that the organizers decided to double its length from one to two days.
The first day of the workshop began with an excellent keynote by Adam Langley, in which he reflected on the current state of SSL/TLS and its vulnerabilities. As a side remark he mentioned that to tackle the everlasting problem of such vulnerabilities to occur, research should be performed with a much higher level of abstraction rather than focusing on the exact details of a “certain hash function of some specific CBC cipher”.
A number of interesting talks followed such as the one by Florian Adamsky about the possibility to exploit the BitTorrent protocol family for distributed denial-of-service attacks. One attack point is the two-way handshake of the UDP-based Micro Transport Protocol (uTP) used by BitTorrent. The handshake consists of the initiator sending a ST_SYN packet, which is answered by the receiver with a ST_STATE packet. After both packets have been exchanged, the connection is established. By spoofing the IP address, an attacker can exploit this two-way handshake to mislead an amplifier into sending its ST_STATE and ST_DATA packets to a victim, potentially leading to a (D)DoS attack.
The second day began with an inspiring introduction to the NSA Playset by Michael Ossmann. Make sure to check out the related website: http://www.nsaplayset.org/. Afterwards, a group of researchers from Radbound University presented their analysis on WPA2 password generating algorithms in a number of Dutch wireless routers. Spoiler: the algorithms suck! More specifically, the group discovered that all analyzed algorithms used “either the router’s MAC address or serial number, or both, as input” to create the default key. By using additional information that is either known (such as the ESSID or the channel number of the router) or guessable (such as the MAC address), it was then possible to calculate the key for the tested routers (usually within seconds or minutes).
Another wonderful talk was given by a research group from Georgetown University, Washington, D.C. Imagine that somebody says “Cocaine Noodles” to your phone. What could happen is that your phone interprets these words as “Ok Google” and is ready to search for whatever is said next. At first, this may seem quite funny, but in a world of voice recognition devices (phones, smart watches, …) this could become a real threat. In their talk they discussed how to distort speech in such a way that a human does not recognize it as something meaningful but a voice recognition device does. Functionality of the devices could then be triggered remotely, for example, via ‘malicious’ sound bites in a YouTube video.
Of course, apart from the small selection presented above, there were many more awesome talks. All of the details (incl. the research papers) can be found here.
The actual Usenix Security Symposium began on Wednesday morning and lasted till Friday. There were a number of great talks but if I had to choose my favorite one it would be “GSMem: Data Exfiltration from Air-Gapped Computers over GSM Frequencies” from a research group at the Ben-Gurion University of the Negev. They succeeded in exfiltrating data from an infected target computer through an air-gap over cellular frequencies. The most interesting part is how they created the transmission signal. In order to do this, they used the memory bus of a standard computer as a transmitter. By exchanging certain data packets between the CPU and the RAM, the electric currents running through the bus created electromagnetic signals with a well-defined spectrum. These signals could then be used to transmit information to a cellular phone nearby (~1.5 meters away). Larger distances (~30 meters) could be achieved with specialized hardware receivers.
The research papers for this and the other Usenix talks can be found here. Additionally, the Usenix ’13 paper “Dismantling Megamos Crypto: Wirelessly Lockpicking a Vehicle Immobilzer” can be found there. Due to legal restrictions regarding this research project, it was only now possible for the researchers to give their presentation and publish their paper.
Finally, a personal anecdote: while writing this blog post, I was sitting at the Ronald Reagan airport in Washington, D.C. All flights were delayed due to a major malfunction in the system that routes planes over Washington airspace. At the airport it was announced that the third backup routing system has taken over, but since it is 20 years old, nobody really knew how to operate it. The second backup system also had a malfunction, else they would have used that. Besides not catching my flight connection, it would be interesting to know what could have caused such a major malfunction. Whatsoever, in the end I was assigned a new and direct connection. But with a crashed routing system, I do not know if I really should have considered myself lucky…
Greetings and see you at the next Usenix
Oliver