Slashdot Mirror
NSA Approves First 802.11b Product for Secret Data
joehoya writes "I realize this is a couple of days old, but the National Security Agency recently certified the Harris Corp's Secnet-11 as the first 802.11b system permitted to carry US SECRET level data. See press release. The system integrates NSA crypto with commercial chipset based 802.11b PCMCIA cards and access points to create a secure wireless LAN. Unfortunately, you and I won't be able to buy them, as they are only available to organizations with an NSA COMSEC account."
Or, in English (and not marketdroidspeak) you can have perfectly secure communications over existing 802.11 as long as you encrypt at the protocol level rather than the hardware (link? I need to study my OSI seven layer network burrito) level. So why do we care about this anyway?
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
When will someone take one apart and find out its a and figures out how rip the firmware out of it for use in standard cards?
My potato gun was confiscated by the United Nations. They said I wasn't allowed to have weapons of mash destruction.
I wasn't able to find this in the press release. Does anyone know if the encryption algorithm would be public key based, or would it be DEC or something like that?
Atheism is a religion to the same extent that not collecting stamps is a hobby.
I hope that at some point technology like this makes it out to the hands of the average consumer. It's good to see that at least someone is trying to make wireless access more secure. It would be nice to be able to pick up a secure wireless product at some point, and use it out of the box without worries of it being insecure.
But until then, there's always VPN or SSH tunnels. And as an added bonus, you can impliment SSH tunnels for free. (even for web and other traffic... not just SSH data)
-Through the server, over the router, off the firewall... Nothing but 'Net!
who is fooling who here? None of the OSes (only Windows versions) it works with are certified for TOP SECRET data.... guess its pretty useless till someone does the linux port eh?
-jon
In a recent article we discussed the futility of implementing a detector detector in a network. This seems that this would be one use that would actually help as an extra layer of defense.
This is my digital signature. 10011011001
impressive stuff... from what the datasheet says this all looks to be implemented hardware on the card - but given the low-level facilities of the chipsets on consumer-grade 802.11 cards is there any reason why some bright coder can't do a similar thing in driverspace?
If they have good reason at all to be that paranoid about a wired LAN, I think it won't take long for this "secure" wireless thing to come back and bite the NSA.
This space intentionally left blank.
The main NSA COMSEC Account is 880099, and its address follows:
"The lesson to be learned is not to take the comments on slashdot too literally." --Vinnie Falco, BearShare
My unit sets up networks in the field (I'm a Marine) and most of the work involves running a fiber backbone and running CAT5 to each and every computer in the fieldHQ (tents). We had looked into running wireless but of course the security was non-existent. Maybe now, we can spend more time training the junior Marines on real networking, not running a CAT5 drops to some officer who "has" to check his email. My platoon will be looking into this tomorrow, I can assure you.
This guy is way out there
The academic/civilian crypto community is far behind the NSA. We don't really have any experience deducing the structure of cryptosystems with only ciphertext and bits of plaintext. The academic community today would have had a hard time breaking Enigma (assuming the wirings were not known) even with the computing power available to us. Bear in mind that almost all of the work done to cryptoanalyse rotor systems is still classified. It is safe to say that there are entire categories of cryptanalytic and cipher design techniques that we are ignorant about. Look at skipjack-- the use of a stepped LFSR in the round function adds tremendous variability to the cipher system. Skipjack appears to be exactly as strong as it needs to be (I believe the best attack we know about breaks 31 of 32 rounds with essentially all of the text available). It is also incredibly fragile, a peak of strength in a broad valley of weakness: practically any change one makes to it makes it much easier to break.
I work in the defense industry, so I have to deal with security issues on occasion. Even though they got someone to sign off on the security of the wireless transmission, it will be interesting to see how they actually implement this technology as part of a larger accreditable system.
In my experience with security organizations, they tend to overemphasize the role of physical safeguards in designating a system as "secure," especially when it comes to COMSEC. How will they feel about accrediting a system in which multiple COMSEC units can be moved outside of a secured perimeter?
"she says i'm lousy conversation. as if that's supposed to help."
This is a non-event. And secret is not a life-or-death classification level, as anything that is considered remotely important will automagically get tagged with TS + keyword.
Pedro
----
The Insomniac Coder
To get something approved for processing at the SECRET level is a moderately big deal for those who work with such data. For the outside world, it's not the last word on the quality of the system.
You can't, for example, get a Linux box approved to process SECRET information (at least, last I checked). Windows is approved, however. Yet, for the commercial user, I would say that Linux is more secure than Windows. What matters is how the system is set up. I'm kind of surprised that there's any demand for wireless networking at the SECRET level. With few exceptions, a classified box has to be physically disconnected from all other machines and operate only from hard drives with no communications software on them. There was an article on cnn.com today about a hacker who got access to sensitive but not classified information on military networks. The reason he didn't get access to classified information is because of the way it's protected.
And forget about anything at the TOP SECRET level or above. We have a room at the office that does work at the TS level. If you bring a disk in there, you can't leave with it. If you bring a hard drive in there, it can't leave the room. Once a computer goes in there, it can't leave either. Well, that's not entirely true...security chops them up into little tiny pieces, waves magnets over them, and does some other magic to make them completely clean before they can leave. They're certainly never useable again. They even destroy the monitors before removing them from the room, in case an image might be burned into them.
Anyway. People who deal with SECRET information will probably be interested in this article, and I'm sure life will go on with no change for those who don't.
One of the biggest aspects of military security lies in not revealing what technology they use. Any information given gives a potential attacker a clue where to start...a lack of such information greatly increases the time to even ascertain whether an encryption is even worth spending time on. As anyone who has hacked or tried to find security vulnerabilites knows, one of the nicest things you can hope for when trying to gain access to a server is what software and version its running. If you don't know that, it kind of makes it hard to know where to start.
Some highlights:
- The card sticks out of the computer with two antennas poking up.
- It uses an NSA encryption algorithm called BATON (from various
stuff on the Web, I get the impression that BATON is a 64-bit block
cipher with 128-bit keys that is designed for very fast operation)
- the message address is encrypted to prevent traffic analysis (this
is a big selling point against VPN technology)
- Each packet has an 80-bit IV (it's rare to learn even that much
about a Type 1 encryption system)
- Cards cost over $2500 each. That's 30 times the price of a commercial WiFi card, but cheaper
than traditional NSA encryption data products which seem to run
around $5K per node.
- "Red keys" are loaded via a special cable that connects to a data
transfer device such as the CYZ-10.
I wonder how much work it would be for someone to implement a commercial version of this using Rijndael, or AES, or something unclassified. With a larger market than the government, maybe it could be cheaper, and the development costs made up on volume...Let's face it, it's a pain to set up IPSEC on all your boxes...