Slashdot Mirror
Chair of IEEE 802.11 Responds to WEP Security Flaws
Subject: WLAN/ Response of WEP Security
Importance: High
Response from the IEEE 802.11 Chair on WEP Security
Recent reports in the press have described the results of certain research efforts directed towards determining the level of security achievable with the Wired Equivalent Privacy algorithm in the IEEE 802.11 Wireless LAN standard. While much of the reporting has been accurate, there have been some misconceptions on this topic that are now spreading through the media. Befitting the importance of the issue, I am inclined to make a response from the Chair to clarify these issues with the following points:
1. Contrary to certain reports in the press, the development of WEP as an integral part of the IEEE 802.11 standard was accomplished through a completely open process. Like all IEEE 802 standards activities, participation is open to all interested parties, and indeed the IEEE 802.11 committee has had a large and active membership.
2. The acronym WEP stands for Wired Equivalent Privacy, and from the outset the goals for WEP have been clear, namely to provide an equivalent level of privacy as is ordinarily present with a wired LAN. Wired LANs such as IEEE 802.3 (Ethernet) are ordinarily protected by the physical security mechanisms within a facility (such as controlled entrances to a building), and the IEEE wired LAN standards do not incorporate encryption. Wireless LANs are not necessarily protected by physical security, and consequently to provide an equivalent level of privacy it was decided to incorporate WEP encryption into the IEEE 802.11 standard. However, recognizing that the level of privacy afforded by physical security in the wired LAN case is limited, the goals of WEP are similarly limited. WEP is not intended to be a complete security solution, but, just as with physical security in the wired LAN case, should be supplemented with additional security mechanisms such as access control, end-to-end encryption, password protections, authentication, virtual private networks, and firewalls, whenever the value of the data being protected justifies such concern.
3. Given the goals for Wired Equivalent Privacy, WEP has been, and continues to be, a very effective deterrent against the vast majority of attackers that might attempt to compromise the privacy of a wireless LAN, ranging from casual snoopers to sophisticated hackers armed with substantial money and resources.
4. The active attacks on WEP reported recently in the press are not simple to mount. They are attacks, which could conceivably be mounted given enough time and money. The attacks in fact appear to require considerable development resources and computer power. It is not clear at all whether the payoff to the attacker after marshalling the resources to mount such an attack would necessarily justify the expense of the attack, particularly given the presence of cheaper and simpler alternative attacks on the physical security of a facility. Key management systems also reduce the window of these attacks succeeding.
5. In an enterprise or other large installation, the complete set of security mechanisms typically employed in addition to WEP would make even a successful attack on WEP of marginal value to the attacker.
6. In a home environment, the likelihood of such an attack being mounted is probably negligible, given the cost of the attack versus the typical value of the stolen data.
7. IEEE 802.11 is currently working on extensions to WEP for incorporation within a future version of the standard. This work was initiated in July 1999 as Task Group E, with the specific goal of strengthening the security mechanisms so as to provide a level of security beyond the initial requirements for Wired Equivalent Privacy. The enhancements currently proposed are intended to counter extremely sophisticated attacks, including those that have been recently reported on in the press. In addition it needs to be noted that the choice of encryption algorithms by IEEE 802.11 are not purely technical decisions but they are limited by government export law restrictions as well.
8. Certain reports in the press have implied that frequency hopping wireless LAN systems would be less vulnerable to security attacks than other wireless LANs. This is not true given that in such frequency hopping systems the hopping codes and timings are unencrypted and consequently are easily available to an attacker.
9. By far the biggest threat to the security of any wireless LAN is the failure to use the protection mechanisms that are available, including WEP. Any IEEE 802.11 installation where data privacy is a concern should use WEP.
I would like to thank the following long serving members of the IEEE 802.11 Working Group, and those Wireless Ethernet Compatibility Alliance members, for their efforts in assisting me in drafting this response from the Chair to this important issue:.
- Vic Hayes, IEEE 802.11 member & ex-IEEE 802.11 Chair
- Al Petrick, IEEE 802.11 WG Vice-Chair
- Harry Worstell, IEEE 802.11 WG Vice Chair
- John Fakatselis, IEEE 802.11 Task Group E Chair & TGE QoS Sub-Group Chair
- Dave Halasz, IEEE 802.11 TGE Security Sub-Group Chair
- Matthew Shoemake, IEEE 802.11 Task Group G Chair
- Phil Belanger, WECA Chairman & IEEE 802.11 member
- Greg Ennis, WECA Technical Director & IEEE 802.11 member.
Chair, IEEE 802.11 , Standards Working Group for Wireless Local Area Networks.
http://www.ieee802.org/11
You basically need a laptop, a 2.4GHz receiver, and an antenna.
You forgot the most expensive resource: tons of time. In order to do anything meaningful without being attached to the wired portion of the network as well, you need to wait for the IV (salt) bits to be repeated. It will take few hours on a very busy LAN, and months on your home installation. Even after this IV collision, all you get is a XOR of two frames - not nearly enough to recover even one of them. Essentially, few hours of work yield you one bit of data (not key!). Of course, if you have enough patience to sit (literally!) for years in the close proximity (or direct line of sight with directional antenna) from the target, you may finally be able to fake few packets on the air.
Please note, that this does not even take into the account the cost of tools, since the tools are indeed a one-time investment.
If you have access to the wired network behind the firewall, breaking into the wireless portion becomes much easier. However, in this case the easiest thing to do would be just to use the wired connection for hacking anyway, and not even bother about wireless. That's what WEP was designed to do: make the wireless piece of the LAN to be about as hard to hack as the wired one.
Seriously... if they try, they'll break it. If they have a signal amplifier, they can sniff their way in from the comfort of their home, not "sitting in a car outside the company gates for 24 hours" as we've been led to believe.
Connectionless-oriented networks will ALWAYS be more susceptible to attack for this reason; POE. One point of entry from any connection-oriented LAN (router) means that hackers have only one way to touch your network, assuming a decent NAT is setup, or a correctly configured firewall... and also assuming each user on the network doesn't have a static external IP address.
But if you're using a wireless LAN, none of those precautions matter. Once the encryption is broken, you've lost the benefits that a single POE can provide. Now they don't have to pass the NAT and get through the firewall... they can just slink in from some poweruser's account who decided "password" would be sufficient to authenticate him to the server. Hey, we shouldn't be too rough on the guy, at least he didn't leave it blank like some other users I've seen.
Of course, if you don't know how to setup a wireless LAN and don't even bother installing encryption, and decide DHCP will make everything so much easier, then all the hacker has to do is set himself to grab an available IP off the network, and we're back to guessing any user's password.
No, they still can't gain root access this way, but they can still do a ton of damage to company data that the user had access to.
Any company that values their data will keep building network infrastructures and pulling drops of CAT5 through the ceilings and walls.
I can't wait to see how Intel's internal wireless LAN works. It should be a good test.
Protector of Capitalist views,
Protector of Capitalist views,
Meorah
Even 1024-bit RSA encryption is a trade-off, in terms of assuming that the cost in time and computing cycles to factor the key is not worth the data being encrypted.
So true... I know I routinely encrypt data which is worth far more than the planet's computing resources. Nope, 1024-bit RSA might be good enough for military secrets but my e-mail needs real security.
WEP provides minimal security, hopefully equivalent to the difficulty of plugging in to a wired network (hence the name).
The really annoying thing about the WEP weaknesses is that it didn't have to be so lousy. RC4 is a perfectly good encryption algorithm and the correct ways to use it are very well known. The WEP weaknesses weren't a result of deliberate security/performance tradeoffs, they were a result of cluelessness. Two tiny changes (using a larger IV and using a secure hash algorithm instead of CRC-32) fix all the identified problems!
I obtained my first network experience on my college network, and let me tell you, I'd feel a lot better about 802.11 with WEP than I used to about ethernet in my residence hall!
Although this statement is funny, it misses the point. The ethernet in your residence hall was a wide-open book... IF you could plug into it. Someone standing on the lawn with a laptop was out of luck, unlike on a WEP network.
Note to ACs: I usually delete AC replies without reading them. If you want to talk to me, log in.
The real cost of an attack is the cost of spending days to months monitoring a network, and then the time analyzing the data.
Yes, but as the Berkeley vulnerability paper states, the 'dictionary attack' method will gather the full set of cipher possibilities in a matter of about 48 hours, recording them to a 12 gigabyte disk. from then its' simple search and replace on a packet-by-packet basis. No brainwork or months of computational power required.
Kevin Fox
Kevin Fox
Does anyone know if that would work?
I'm not sure if you meant whether or not the white noise generators would stop tempest snooping or if "The Man" can get a warrant based only on suspicion. So I'll take a stab the computer part since I'm lazy.
The FCC says that all electronic equipment licensed for use in the home can't interfere with any other equipment. Any equipment capable of interfering electronically with anything else needs to be licensed, like your TV remote or Nomad Jukebox. My computer doesn't interfere with the TV that I own, or the cordless telephone that I don't own but my neighbor just bought. Tempest snooping works by monitoring/interpreting EM emissions from your computer. Computers output a poop load of EMR(r for radiation) and the only way to sheild it is by putting enough insulation between it and the rest of the world.
A friend of mine was in the military and one of his biggest complaints (other then working on 386's only 2 years ago) was that they were HEAVY as hell. When I asked why he said, "They're sheilded in case of a directed EMP."
btw, directed emp really means 'In case we get nuked."
"Me Ted"
BOSTON SUCKS!
"Bad security is better than no security. With bad security, you *think* you're safe. With no security, you know you aren't."
A system using 128 Bit WEP thinks its secure. A system using 0 bit WEP uses IPSec* and actually is.
Yours Truly,
Dan Kaminsky, CISSP
DoxPara Research
http://www.doxpara.com
* Yes, IPSec is an utter horror to configure. It manages to work, which is more than I can say for WEP!
Nope. Keyboards and flat panels are safe. Point is that monitors use a high voltage to scan the screen, which produces mucho electromagnetic signals to pick up - quite possibly from enough distance that you don't have to be in the same building. This is at a fixed frequency which is fairly easy to pick up, and there's nothing much else of the same kind of emissions level to interfere with it.
Keyboards and screens use 5V or less to (respectively) send data down the cable and to scan the screen. This is happening at fixed frequencies, but it's such a low voltage that you'd need an extraordinarily sensitive receiver placed pretty damn close to the equipment. Point is that signal strength is proportional to the SQUARE of the voltage. A 5V signal doesn't produce 20 times less EM noise than a 100V signal, it produces 400 times less!
And if you did have this extraordinariny sensitive receiver, it'd be swamped by the broad-band noise generated by the processors, RAM, etc which are also giving off the same kind of emissions. Never mind the shielding which every bit of kit has to try and prevent it giving off EM emissions.
Grab.
Y'know, it's starting to burn my butt.
This guy spent a couple of kilobytes using valuation judgments to justify the failure to put adequate security in his network specification.
But how the hell does he know what I'm going to be sending over the network?
Even if I have crypto in every other layer of my stack, I would still want good security from the airlink box. But here it's not even a standardized option.
He had a chance to do something immutable and wise, and he blew it.
--Blair
First of all, I dispute that a lot of resources are required to crack these systems. You basically need a laptop, a 2.4GHz receiver, and an antenna. The laptop (assuming it's new) should provide more than enough computing power for this job. Sure you need some software, but really, how much longer do you think it'll be before script kiddies will have access to software to do this for them?
I also dispute that your average enterprise network provides adequate security to protect it's data even in the event that WEP is compromised. Most corporate security systems I know essentially assume that unwanted parties cannot join the network and listen in on traffic. If I break the WEP key for a network, this assumption becomes invalid. The attacker can then watch as proprietary documents get passed over the network (normally documents move unencrypted) and read them.
As for the "wired-equivalent" aspect of this. I can't remember a wired network that I could compromise without having to get physical access to a network jack. Corporations could increase their security based on how hard they made it to get to those jacks. Indeed, in some cases there are no controls, but in most companies, there are at least some controls and companies have the option of significantly increasing the physical access security. WEP does not provide an equivalent to this (indeed, I'm not sure there's a clear standard meant by "wired-equivalent").
Finally, after reading over the material on this, I have to say that the WEP group just went about this fundamentally the wrong way. They tried to use RC4 for authorization, which it's not particularly well suited for. They also effectively tried to use a single key for both authorization and encryption. They used 24-bits of variability to protect against RC4 compromise, even though that's significantly less than one would need. All of these are no-no's, and the WEP group should have been aware of these problems (and made some effort to address them).
It is possible, today, to do wireless communications securely. WEP could have employed existing proven approaches, but instead they rolled their own. Guess what happens when you do this boys and girls?
sigs are a waste of space
The "Reply from the Chair" states several times that the prohibitive cost of an attack is a rationale for saying there is a low risk of such an attack, especially in a residential or other low-security setting. The logic used to reach this conclusion is flawed.
One thing to be clear on here is that the efforts used by the Berkeley group to find these shortcomings, while significant, were not only not extraordinarily expensive, but they were also one-time costs. The legwork has been done, and replicating this work to sniff new networks would not be expensive, beyond using the proper software, a slightly modified Orinoco card, and a 12 gig hard drive. This is hardly beyond the means of someone even casually interested in another's data.
Kevin Fox
Kevin Fox
802.11 describes ethernet over RF at 2Mbps, freq. hopping, and DS/SS[1]. 802.11b describes ethernet over RF at 11, 5.5, and 2 Mbps over DS/SS. Freq. Hopping may have been recently added (I'm not certain), but I don't know if it was incorporated into 802.11b or if it is its own standard -- it was not, however, part of the original 802.11b.
I (and my Netadmin and RF Engineer) are of the belief that WEP was part of 802.11b, but not 802.11. We haven't read the specs in awhile, so don't quote us.
-Mith
[1] Direct Sequence Spread Spectrum
--
For quite some time the technolgy has been around to avoid a "middle of the night" break-in by Big Brother. A Dutch scientist by the name of Wim van Eck published a paper entitled "Electromagnetic Radiation from Video Display Units: An Eavesdropping risk?" which was published in the "Computers & Security" journal. In it he talks about electromagnetic emmissions from CRT's which can be intercepted, reconstructed, and viewed from a remote location.
I understand why wireless tools increase risk, but we the people cannot forget that what we read, what we write, whatever appears on our computer monitors *can* be intercepted.
Again, I'm no infosec wizard, and I'm still learning a lot, but I believe TEMPEST (Transient ElectroMagnetic Pulse STandard) standards(reduntant?) detailed in NACSIM 5100A (a document classified by the NSA) tell the amount of shielding necessary. But...if we knew how to protect ourselves -- we couldn't be watched.
What I'm trying to say is -- although new standards present new troubles, don't forget about the dangers out there now.
--Joey
Forget that, I want one of those talking chairs.
The Berkeley group presented a mathematical explanation of how WEP could be attacked, but as far as I could tell they never actually went out and did it against a real live network.
Their description said it would require sitting there and recording all packets sent for an entire day or so. Then you still have to play cryptographic games and make guesses about what plaintext corresponds to what data. Are there downloadable kits to do this?
Wait a minute. Because people won't take proper precautions to protect their own damn data, it's the Chair of this standards commities fault? He simply presented options, the same options one has with any network. I fail to see how the irresponsability of others is in anyway his fault. If you want to protect your data, do so. If you're a sysadmin and are tasked with protecting your company's data, do so. But blaming this guy for preserving the choice is crap.
--Jimmy has fancy plans; and pants to match.
Who cares about security in the Windows Entertainment Pack? I just wanted to play Mah-Jongg...
---- Just another spud server.
What concerns me about the push towards "wireless" protocols such as Bluetooth is that it increases the vulnerability of internet traffic, no matter what measures you take to keep it secure. Well, I suppose you could use narrow-beam lasers to transmit the information, but in reality there is going to be far more opportunity to listen in with no chance of such listening being detected.
I'm sure that certain government agencies would love for people to be using this. Rather than actually having to go into peoples houses with all the Fourth Amendment difficulties that entails (or at least should; the Fourth has been gutted by successive governments in the last eighty years), they can simply snoop on you from the comfort of the house across the street if their instruments are good enough.
How is anyone going to keep anything private if this happens? It has been shown time and time again that governments are interested in nothing more than maintaining their own power at the expense of citizens rights. And most people just don't care enough for this to be a problem.
Hopefully our new Attorney General will hold to the Constitution better than the last one did. Because otherwise travesties like Waco will happen again, where our Constitutional rights are taken away by stormtroopers masquerading as law enforcement.
What is the point of proving equivilent privacy. I agree with the point that existing LANs are not particularly secure (espeically from internal actors). However creating a security system for wireless lans (which of course lack physical protection) that only offers weak protection seems pointless.
Let's look at in the way my 8th grade logic teacher would.
A. If the protection (encryption, procedure implementation, etc) is weak it will get broken.
B. When it is broken the exploit will be avaliable to everyone. (not necessarily, but usually true).
C. Once the exploit is widely avaliable the protection system will provide no protection whatsoever.
No protection is less than some (physical) protection.
Therefore the protection is not equivilent.
Now of course there is one other thing. The DMCA makes it illegal to break even bad encrytion. So I guess it is just as illegal to view WEP protected data as it is to break into a building.
Thoughts?
H: Dude, what's it say on my back?
CN: d00d! What's it say on mine???
H: l33t! What's it say on mine, dude???
CN: d00d! Come on, what's it say on mine!?!?!???
H: L33T!!!