Cracking GSM phone crypto via distributed computing
If you are using a GSM phone (AT&T or T-Mobile in the U.S.), you likely have a few more months before it will be easy for practically anyone to spy on your communications.
Security researcher Karsten Nohl is launching an open-source, distributed computing project designed to crack the encryption used on GSM phones and compile it into a code book that can be used to decode conversations and any data that gets sent to and from the phone.
Karsten Nohl talks about his distributed computing, open-source AE/1 cracking project at the Hacking at Random conference.
(Credit: Hacking at Random)He hopes that by doing this it will spur cellular providers into improving the security of their services and fix a weakness that has been around for 15 years and affects about 3 billion mobile users.
"We're not creating a vulnerability but publicizing a flaw that's already being exploited very widely," he said in a phone interview Monday.
"Clearly we are making the attack more practical and much cheaper, and of course there's a moral question of whether we should do that," he said. "But more importantly, we are informing (people) about a longstanding vulnerability and hopefully preventing more systems from adopting this."
This weakness in the encryption used on the phones, A5/1, has been known about for years. There are at least four commercial tools that allow for decrypting GSM communications that range in price from $100,000 to $250,000 depending on how fast you want the software to work, said Nohl, who previously has publicized weaknesses with wireless smart card chips used in transit systems.
It will take 80 high-performance computers about three months to do a brute force attack on A5/1 and create a large look-up table that will serve as the code book, said Nohl, who announced the project at the Hacking at Random conference in the Netherlands 10 days ago.
Using the code book, anyone could get the encryption key for any GSM call, SMS message, or other communication encrypted with A5/1 and listen to the call or read the data in the clear. If 160 people donate their computing resources to the project, it should only take one and a half months to complete, he said.
Participants download the software and three months later they share the files created with others, via BitTorrent, for instance, Nohl said. "We have no connection to them," he added.
Once the look-up table is created it would be available for anyone to use.
Distributed computing, which has long been used for research and academic purposes, like SETI@home, and which companies have built businesses around, not only solves the technical hurdle to cracking the A5/1 code, but it could solve the legal ones too.
A few years ago a similar GSM cracking project was embarked upon but was halted before it was completed after researchers were intimidated, possibly by a cellular provider, Nohl said. By distributing the effort among participants and not having it centralized, the new effort will be less vulnerable to outside interference, he said.
Nohl wasn't certain of the legal ramifications of the project but said it's likely that using such a look-up table is illegal but possession is legal because of the companies that openly advertise their tables for sale.
A T-Mobile spokeswoman said the company had no comment on the matter.
AT&T spokesman Mark Siegel said, "We take extraordinary care to protect the privacy of our customers and use a variety of tools, many technical and some human approaches. I can't go into the details for security reasons." He declined to elaborate or comment further.
Taking precautions
Carriers should upgrade the encryption or move voice services to 3G, which has much stronger encryption, Nohl said.
In the meantime, people can use separate encryption products on the phone, like Cellcrypt, or handsets with their own encryption, Nohl said. Amnesty International and Greenpeace are using phones with stronger encryption, for example, but it only works if both parties to a conversation are using the same technology, he said.
For data encryption there is Pretty Good Privacy (PGP) for e-mail and virtual private network (VPN) software for connecting to a corporate network, he said.
The encryption problem is particularly serious for people doing online banking, where banks are using text messages as authentication tokens. Banks should instead offer RSA SecurID tokens or send one-time pass phrases through regular mail, Nohl said.
"I think, potentially, this could have as much impact as the breaking of WEP (Wired Equivalent Privacy) had a few years ago," said Stan Schatt, security practice director at ABI Research. "That shook up the industry quite a bit."
As a result of breaking that encryption, enterprises were reluctant to rely on wireless LANs so the Wi-Fi Alliance pushed through an interim standard that strengthened the encryption scheme, he said.
"Vendors will jump in with interim solutions, like Cellcrypt," Schatt said. "Mobile operators themselves will have to jump in and offer additional levels of encryption as part of a managed service offering for people who want a higher level of encryption."
However, consumers aren't likely to want to pay extra for the boosted encryption strength, he said.
To snoop on someone's phone, a would-be spy would need to be within eyesight of the target, Schatt said. Or, spies could point a recording device in the direction of a building and grab whatever conversations were nearby, he said.
"If you stand outside a building of a competitor you could get conversations between product managers and about sensitive corporation information, like acquisitions," he said. "Corporations put even more sensitive information over their phones, in general, than they do over their e-mail."
Update Wednesday August 26 8:01 a.m. PDT: The project web page is here and the the talk with slides is here.
Elinor Mills covers Internet security and privacy. She joined CNET News in 2005 after working as a foreign correspondent for Reuters in Portugal and writing for The Industry Standard, the IDG News Service, and the Associated Press. E-mail Elinor. 






Ha. What a joke. I suppose it was also "possibly" by fairy dust. Or more likely, they couldn't find anyone interested enough to sponsor them.
A5/1 is only 54 bits, so "cracking" it is not really an accomplishment. It's strong enough that it would take more effort than some casual script kiddie program, so eavesdropping isn't really feasible. Nor is tapping thousands of random phone calls from the comfort of home. You could spend months of distributed computing to make a table to be able to follow me around and try to run through your hashes before I noticed you or finished my call, but you'd get better results peering in my window while I type at my computer.
If I were someone who needed true protection, I could run a real encryption program. It wouldn't even need to be anything fancy like 256 bit--this entire project is made irrelevant with 128 bit.
You are already aware of the weakening in the implementation of A5/1. the industry needs to move to secure peer-reviewed encryption algorithms (and disclosure of how keys are held) otherwise we will keep having keys padded with 0s and 128bit keys with 64 bit entropy.
Disclaimer: I work for Cellcrypt
-Rodolfo
@jaguar717: "fairy dust"? I guess you've not heard of the many researchers that have been gagged by various entities when they tried to report on security weaknesses? Try doing a little research on the subject before issuing such a glib comment next time, k?
And as new-type mentioned, rainbow tables are very much a script-kiddie's toy these days...in regular use for war-driving against wifi installations as just one example. Just think of the juicy details a microsoft agent could gain by soaking up GSM calls outside the apple headquarters with a directional antenna and a laptop, or vice-versa. This is a huge deal...especially with the much smaller scale of hardware needed after the tables are created and thus are plausibly denied versus the specialized gear that was previously requried. And once the table is made, nobody has to make it again..it gets passed around via bitorrent or IRC/newsgroup channels.
And as for your comment "You could spend months of distributed computing to make a table to be able to follow me around and try to run through your hashes before I noticed you or finished my call, but you'd get better results peering in my window while I type at my computer."...it's pretty obvious that you're clueless to how these attacks work. The idea is to capture the raw data for decryption later...not necessarily real-time...and easily done with a high-gain directional antenna and a laptop. And the ramifications of getting a "lucky find" could be devastating to an individual or organization..see my above example of microsoft/apple.
Nearly *all* people need "true protection"..they just don't realize it. And good luck in finding people on both ends of all the calls you make who are willing to run the same 3rd-party crypto program as you (unless you work in a gov't agency or company that mandates such a standard). Once again, you don't seem to grasp the bigger ideas here...the phone companies need to step up to the plate and fix the built-in crypto before some Very Bad Things (tm) happen.
Maybe there is a reason why after 15-years, nobody really cares. The average person isn't important enough to make this worthwhile. Why would someone spend $100,000 to $250,000, 80 computers, and 10+ days just to overhear some random guy tell his wife that he'll be home late from work? Heck, with all the people using bluetooth headsets and talking openly in Malls or public areas overhearing some random conversation is easy.
Because the data set needs to be generated only one time, and then it can be used by anybody who wants to eavesdrop on ANY GSM conversation in the world. It's not just random people talking to their wives or listening to people babble at the mall...think about how many people conduct business on their cell phones. How many account numbers and personal identifying bits of data are disclosed over them in the course of a day in an average office building (just about everybody has a crackberry these days and a huge chunk of them use GSM). How many passcodes, PIN numbers, and credit card numbers are entered into various automated systems through phone connections each day? You're pretty short-sighted and naive if you don't see the major downside here.
A party line is where several houses shared a single phone line. It was cheaper that way. Never say anything on the phone or on the radio (cell phone) that you would not want the whole world to know.
It was about 1970 before the phone company was told to provide private lines to everyone who wanted one. Prior to that a party line was often the only service available. My parents phone line went to three other houses.
There is still a federal law on the books (Communications Act of 1934) that makes it illegal to repeat what you hear on the radio (or party line phone), unless it is commercial broadcasting.
Don't worry, there is always a way to get you in jail.
Ummm, what about Mitnick?
The same has been done by a lot of voice crypto providers, and unfortunately i have to say also by Cellcrypt that is referred in the article.
Most of the voice encryption systems out there does not have published source code, does not have published protocol with reference implementation and are self-referenced security systems.
On 3G there was a public selection of the cryptographic system, there was different participant and there was a winner subject to public scrutiny along with a reference implementation.
Only public and open standards, like the ZRTP (http://en.wikipedia.org/wiki/ZRTP) can be considered trustful.
All the other voice encryption technologies does not have the requisite to be considered "secure" because:
- are not open protocols
- are not subject to public evaluation and scrutiny
- does not have a public and available reference implementation
- does not have "multiple" implementation (this means that a vulnerability in a single implementation can make the whole system security to collapse)
Currently the only two voice security technologies that can be considered are ZRTP and DTLS made public and open by Cisco (http://en.wikipedia.org/wiki/DTLS) .
All the other have the same approach and risks of the GSM protocol.
Remember that security it's a matter of approach and methodology and not a matter of technology marketing.
Take care and make your consideration on voice encryption technologies and products.
Mike
Thanks!
KM
Elinor
- by sectechbiz September 18, 2009 12:10 AM PDT
- Dear Sir
- Like this Reply to this comment
-
(14 Comments)Sectechbiz is a UAE based Manufacturer company .
If you are Looking for Mobile Jammers, Spy Phone, Secure GSM Phone,Voice and Fax Logger,Voice Fax and Data Encryptions,UHF Transmitters, Biometric USB Drive and other Security and Surveillances Products.
Then You can Contact us ,we will provide you best products in market .
For more details you can visit our site http://www.sectechbiz.com/ .
You can mail me your exact requirement, and we will reply very soon.
Our mail id is
marketing@sectechbiz.com
Thanks & Regards
SectechBiz
Dubai,UAE