innismir.net

It has finally happened. I can finally write a blog post about my two favorite subjects: Information Security and Ham Radio.

Chris Paget made some news this weekend at yearly DEFCON hacker conference in Las Vegas. Paget demonstrated the flaws of the GSM cell phone protocol by creating a simple device to intercept every GSM call in a small area. Chris did a lot of work making sure that he wasn’t violating anyone’s privacy by intercepting these phone calls, up to enlisting the help of the Electronic Frontier Foundation. When reading some of Chris’ preparations, I was impressed, but the first thing that popped into my head was “Wait, that’s nice and all, but what about FCC regulations?”

To take a quick detour into FCC regulations, most unlicensed devices fall under Part 15 of the FCC rules. They have to be tested and certified by the FCC before they are marketed and sold in the United States. Whenever you see your favorite technology blog talking about how some new device is being tested by the FCC, they’re talking about this testing.

So, when Chris was explaining his presentation I figured he was going to go one of two ways: Either he was going to unveil some kind of new FCC certified cell phone interceptor (unlikely), or he was going to put on an eye patch, raise the Jolly Roger and go full pirate. However, after the presentation was done, I was reading the coverage of the presentation and saw that he did it a way that I hadn’t considered: Chris, unbeknownst to me, had an amateur radio license, so he tried to classify his transmissions under Part 97.

Part 97 is the Amateur Radio section of the FCC rules. Amateur radio is classified as an “experimental” service. As I’ve stated in my “Why you should be an Amateur” presentation, amateur radio is “radio hacking.” Chris saw that part of the European GSM band overlaps with the 33cm amateur radio band, so he (and I!) have rights to transmit there. Seems like a perfect fit, right?

Unfortunately, no. While Chris did seem to catch the “no encryption” part of the rules, he didn’t realize that his transmissions were not legal under Part 97 either for other reasons. Part 97.111 and Part 97.113 establish but “authorized” and “unauthorized” transmissions of amateur radio stations, of which Chris, by my count, violated the rules 2 or 3 different ways:

1. Chris was using his Part 97 transmitter to communicate with Part 15 devices, not other Part 97 devices. (Violates Part 97.111(a))
2. Chris’ GSM cell site was beaconing to cell phones to let them know it’s there. That counts as a way transmission. (Violates Part 97.113(b))
3. Chris was impersonating a AT&T cell phone site. You can’t impersonate people on amateur radio. (Might Violate 97.113(a)(4))

Chris does get props for establishing a Morse Code beacon to ID himself every 10 minutes as defined by the rules, however, that is like a restaurant owner trying to convince the health inspector that his restaurant is OK despite the rats and roaches because his employees wash their hands after they go to the bathroom. Too little, too late.

I’m not trying to string up Chris here, I’m honestly worried for him. He’s admitted that he’s had conversations with the FCC regarding this presentation which he classified as “unproductive”. This, combined with the fact that the FCC enforcement bureau loves to hand out documents with “Notice of Apparent Liability” at the top and five figure fines on the bottom leads me to wonder if Chris isn’t headed toward a protracted legal battle with the Feds. Chris’ presentation shows a major shortcoming with the current FCC rules dealing with research. Chris should not have tried to find a loophole within the FCC regulations to do his research, it should have been legal for him to establish a low powered signal to do and demonstrate his research. We, as researchers, are running into another version of the same ostrich syndrome that prohibited users from listening to cell phone and pager traffic that were transmitted in-the-clear back in the early 1990s, and to a lesser extent, still are. With the expansion of data networks to mobile devices, it’s become even worse, as Chris’ presentation demonstrated. By not allowing research into these fields the FCC is keeping the sunlight out of the dark corners of our mobile networks and allowing the mobile phone companies to convince us that everything is OK when in reality someone with $1500 worth of equipment can intercept local mobile phone traffic is negligent at best, and criminal at worst.

While I disagree with Chris’ characterization of his transmissions being “cool” because he’s licensed as an amateur radio operator, I fully support his research and his efforts to do this research in a controlled environment. I also hope that the FCC will realize that this type of research only helps people and all the laws in the world won’t help bad people from doing this same type of activity in a malicious manner, as they already are.

While going through my backlog of RSS entries that have piled up over the past week, I came across this story from Byron Achohido (via Threatpost, which I highly recommend) who talks about the moral ambiguity of the release of tools that can allow rainbow tables made for cracking the A5/1 GSM encryption cipher. First, let’s get this out of the way: Attacks like this against A5/1 have been around sine at least October 2007. The big deal with these new tools is that they provide the basis of taking the computation time down from days or hours to seconds. These tools are rainbow table generators. They do not do any kind of sniffing or cracking, just a boat load of computations.

This aside, I find the story interesting for a number of reasons: First I like how the iPhone is specifically mentioned. Byron mentions that:

Hackers could go after sensitive information exchanged while using Web apps for phone banking and stock trading; or they could eavesdrop on sensitive conversations, discussion about medical histories, for instance.

Actually, in cases where you are on a 3G network, you’re safe from this attack on the data side, as 3G networks use the A5/3 cipher. The problem is that, at least in AT&Ts case, even if you are on a 3G network, any voice calls are routed over regular GSM channels, which use the faulty A5/1 cipher. I believe T-Mobile is in the same boat. Fixing this is rather simple from a technical standpoint, just flip the voice side over to 3G as well. Of course, we know that in real life it’s almost never that simple. Both carriers’ 3G network is nowhere near the size of their GSM networks, and who knows what kind of capacity they have on the 3G side. However, the decision here is completely on the carrier: What do they value more, their customers security and privacy or their profit margin?

Plus, I think the larger question here is when did mobile phones become secure? I think any person with a background in Information Security or Radio that was around in the early 1990s either monitored cell phones or knew of someone that did. While with the introduction of digital phones the monitoring became more difficult by your simple geek, given a sizable sum of money, it is still possible. The creation of devices such as Cryptophone proves this. Even before these tools were released, there are attacks on GSM in the wild which are “active” attacks, such as spoofing cell towers and then telling the phone to go sans encryption.

Next, regarding the question of releasing these tools; Byron calls the release taking the “morally debatable high ground.” I think his logic is really flawed, and he shows why in his article:

As this timeline depicting the emergence of the Conficker worm shows, the bad guys pay big bucks to black hat researchers adept at finding vulnerabilities, which can be immediately exploited for profit — before anyone issues a patch.

And now grey hat researchers,  like Moore and Nohl,  build careers out of concocting campaigns to embarrass vendors under the banner of compelling vendors to resolve security flaws in popular products – usually highly profitable cash cows — in a timely manner.

It’s been shown that attackers pay large sums of money for attacks that aren’t patched, making a market for enterprising attackers with questionable morals to develop them. With the existence of this market, why are we assuming that the bad guys don’t have rainbow tables for A5/1 already computed and are actively recording calls from high value targets? Cons It’s silly. Releasing these tools essentially destroys the already tattered blanket of ignorance people have been wrapping themselves up in since people started shouting that A5/1 was insecure and once again shows us that mobile phones are, by their very definition, insecure devices.