ansichart

Eureka! Wifi security idea.

14 posts in this topic

I thought of this neat idea, I don't know how practical it really is.. but an interesting concept none the less.

Imagine a Wireless AP that has two directional antennas spaced a significant distance apart (each having a small motor so the antennas can pivot, and point in different directions), and 1 omni-directional antennas. The two directional antennas used to triangulate the client NIC. When a client NIC wants to associate to the WAP, a pseudo-random key is generated by the WAP. Half of the key is sent using 1 directional antenna to the client NIC, and the other half is sent using the other directional antenna. That way, only the client NIC will receive the full key. Then the rest of the communication can be carried out using the omni-directional antenna, encrypted with that key. That way, other clients on the wireless network cannot decrypt other peoples data, because everyone is using different keys that only the WAP knows, and the individual clients know.

You could further expand on this, by having a GPS implemented on the WAP and the client's NIC. The GPS coordinates can be used as extra security parameters.

Anyway, what are your thoughts about this?

Edited by ansichart
0

Share this post


Link to post
Share on other sites
. Half of the key is sent using 1 directional antenna to the client NIC, and the other half is sent using the other directional antenna. That way, only the client NIC will receive the full key.

If someone is listening, they could grab the two keys out of the air and put them together.

Edited by SchippStrich
0

Share this post


Link to post
Share on other sites
. Half of the key is sent using 1 directional antenna to the client NIC, and the other half is sent using the other directional antenna. That way, only the client NIC will receive the full key.

If someone is listening, they could grab the two keys out of the air and put them together.

The two directional antennas are spaced apart, so the signal would intersect at 1 point, where the client is.


* *
\ /
\ /
\ /
/\
/ \
/ \

If there was two computers with wifi NICs along the two paths were they don't intersect, they could put the two key halves together and get the key.

I realize this isn't that great, but it's still an interesting concept.

It would be better if both the client and the WAP generated a private/public key-pair and then using their private keys to encrypt a pseudo-random-generated chunk of data (of some prespecified size), and then digitally sign it. Then the WAP and Client would XOR together the two chunks of data to produce a key that will be used to encrypt/decrypt (symmetric key encryption) all further communication throughout the duration of their session. I know this procedure (or something close to it) is done in several different protocols, since symmetric key is faster. It uses the best of asymmetric encryption and symmetric encryption... virtually no down side to it as far as I am concerned.

0

Share this post


Link to post
Share on other sites

So are you planning on making this a project? or just sharing a thought.

Just curious...

0

Share this post


Link to post
Share on other sites

So are you planning on making this a project? or just sharing a thought.

Just curious...

Just sharing a thought. When I first posted it, I thought it was brilliant. Then I thought about it some more, and I realized how impractical it would be.

0

Share this post


Link to post
Share on other sites

I thought of this neat idea, I don't know how practical it really is.. but an interesting concept none the less.

Imagine a Wireless AP that has two directional antennas spaced a significant distance apart (each having a small motor so the antennas can pivot, and point in different directions), and 1 omni-directional antennas. The two directional antennas used to triangulate the client NIC. When a client NIC wants to associate to the WAP, a pseudo-random key is generated by the WAP. Half of the key is sent using 1 directional antenna to the client NIC, and the other half is sent using the other directional antenna. That way, only the client NIC will receive the full key. Then the rest of the communication can be carried out using the omni-directional antenna, encrypted with that key. That way, other clients on the wireless network cannot decrypt other peoples data, because everyone is using different keys that only the WAP knows, and the individual clients know.

You could further expand on this, by having a GPS implemented on the WAP and the client's NIC. The GPS coordinates can be used as extra security parameters.

Anyway, what are your thoughts about this?

That's unnecessarily complex. MACsec variant with preshared keys or public-key crypto would work. It's a recent standard that secures the MAC layer of Ethernet. It could be applied to many wireless devices using only firmware and client software upgrades, plus a tiny modification to the protocol for wireless environment. There just aren't any existing standards or official support for wireless application of this technology by major vendors. If you are interested in MACsec, I think Safenet supplies products (and info) for it. In any case, numerous existing protocols like SSL and IPSEC could be lightweighted and retargeted to protect Layer 2 wireless communications. The alternatives are simple, heavily analyzed, field proven, often layer agnostic, and had hardware acceleration (AES, RSA, DSA, SHA, etc.). They just decided to make some complicated ass, proprietary cryptosystems that didn't quite do the job. IT history is littered with committees and standards bodies making bad decisions like these.

0

Share this post


Link to post
Share on other sites

...MACsec variant with preshared keys or public-key crypto would work...

Thanks for the info army_of_one. I was intrigued by this and did some google searching. I found this article which has some good info regarding layer2 encryption (including MACsec):

http://www.technewsw...?wlc=1286821396

Thanks for the link. It's a good summary of Layer 2 encryption technology. It didn't mention MACsec, though. The advantage of MACsec is that it's a standard, either official or in the making. If it takes off, we'll see numerous vendors making MACsec-compatible products. This further brings the benefits of competition. Right now, most Layer 2 encryption systems are incompatible. The current model is that each vendor has their own crypto products, along with management software. Standards like IPsec and MACsec make things more vendor neutral. Always good for us.

0

Share this post


Link to post
Share on other sites

I'm not sure I completely follow everything being discussed here, but am gonna throw a random thought out there, correct me if I say something stupid :-). Army_of_one, while I agree that the idea does seem a tad impractical, when, like you said, it's possible now to secure the MAC layer using things like MACsec, I believe that ansichart's idea was focused on a different layer, in order to provide security. I'm talking about the physical layer. If there is literally only one possible location where one can retrieve the full key (due to the directional antennae), then unless someone were to have 2 machines, at 2 separate locations, attempting to grab both parts of the key out of the air, then it is physically impossible for the key to be nabbed. This of course provides security at the lowest possible level.

Here's the part where I'm even less knowledgeable, so again, correct me when you read something foolish. Although I don't know how MACsec works, I know that it's possible to spoof a MAC address. So while layer two encryption does allow for full network security, with nearly no overhead, it more than likely still has its own vulnerabilities.

Single good point: Encryption can always be overcome but the laws of physics are insurmountable. If it is physically impossible to retrieve the data, then the data is safe. The idea isn't necessarily practical for day to day use, but for extremely confidential/volatile data transfers, you have to admit that it makes a good deal of sense, and has probably already been implemented in that manner.

0

Share this post


Link to post
Share on other sites

I'm not sure I completely follow everything being discussed here, but am gonna throw a random thought out there, correct me if I say something stupid :-). Army_of_one, while I agree that the idea does seem a tad impractical, when, like you said, it's possible now to secure the MAC layer using things like MACsec, I believe that ansichart's idea was focused on a different layer, in order to provide security. I'm talking about the physical layer. If there is literally only one possible location where one can retrieve the full key (due to the directional antennae), then unless someone were to have 2 machines, at 2 separate locations, attempting to grab both parts of the key out of the air, then it is physically impossible for the key to be nabbed. This of course provides security at the lowest possible level.

Here's the part where I'm even less knowledgeable, so again, correct me when you read something foolish. Although I don't know how MACsec works, I know that it's possible to spoof a MAC address. So while layer two encryption does allow for full network security, with nearly no overhead, it more than likely still has its own vulnerabilities.

Single good point: Encryption can always be overcome but the laws of physics are insurmountable. If it is physically impossible to retrieve the data, then the data is safe. The idea isn't necessarily practical for day to day use, but for extremely confidential/volatile data transfers, you have to admit that it makes a good deal of sense, and has probably already been implemented in that manner.

"the laws of physics are insurmountable"

You really need to be careful with that kind of thinking. We aren't discussing the laws of physics: we are discussing a system, a security claim, and a method to implement the security claim. Physics doens't change that much over time, but security engineering efforts fail all the time, even those made by experts in simple fields. So, the best way to look at it is through risk assessment, design analysis, implementation analysis and pen testing (attacker's view).

So, we begin by looking at the goal: setting up a secure session between a WAP and a computer. This is essentially a solved problem in every area except wireless and the newest wireless protocols are actually very good. But, let's look at the new proposal.

He's got two directional antennas each broadcasting half the key. They must both be pointed in the same direction or hitting the same spot of the laptop will receive both keys. If anyone gets a single wire near that line of sight, they grab both keys. Also, it's hard to contain EMF waves: the directional wave will bounce around. If it's got some decent range and the user is close, the wave will have enough energy to possibly land somewhere else. Even if the waves or frequencies hop, spread-spectrum style, an attacker could pick this up with the right equipment. The new system will also require additional hardware and firmware at the WAP and maybe client NIC that are currently unnecessary. Hence, it's basically security by obscurity and there are tons of cheap tools for attackers.

Now, let's look at it from the attackers or risk mgmt perspective. Risk management here say you need to look at the value of your asset to the attacker to determine how much effort or money they'd expend to get it. Then, the value to you to figure out how much your willing to spend to protect it (or if you don't care). If your machine isn't worth crap, then you just have to worry about war drivers and existing protocols with decent passwords stop them. If the asset is worth spending a few grand to get, then I'd probably just do a physical attack on your machine to install a rootkit while you take a piss at Starbucks. Do you have firewire? Game over. Maybe get your truecrypt key, then steal your hard drive and decrypt its contents.

Do you see how the security scheme doesn't make a difference if your system matters to attackers? This is why it's not worthwhile. You can't just look at security by cryptography. Security is a system-level thing: every entity that's involved in operation must be considered, from the user to the PC's to the environment. If you just want to encrypt your wireless session, then existing schemes are good enough and a token-supported MACsec-like protocol with hardware-assisted crypto would be ideal. It still only protects your network and data from casual attackers, though. Don't sweat it though: 99% of all things "security" or "secure" are vulnerable to sophisticated attackers. Have to be really thorough and consistent to deal with them.

Note: I'm not knocking that he came up with the idea and proposed it. Coming up with new ideas or new ways of looking at things is great. I'm just doing a security and practicality assessment. Not really intending my posts as a rebuke the the O.P.

0

Share this post


Link to post
Share on other sites

Army_of_one, I definitely agree with a lot of what you said, and you made many good points. Just from your response I would venture to say that you've had a job in the field of risk management before.

However, I stand by what I said about the physics behind the idea, and why they're important. Disregarding all other factors, if the idea were to be tested and eventually properly implemented it could quite possibly make obtaining a key a near physical impossibility, and that is a fairly worthwhile security claim. This is, of course assuming that the computer and the WAP are tucked away somewhere safe as well (e.g. somewhere where you can't use the firewire exploit, or steal the hard drive while I'm takin' a piss at Starbucks...like in a locked metal cabinet bolted to the floor :P) To get to the point, after re-evaluating it, I think the idea would be good for higher up government officials and the owners of certain Fortune 500 companies. As you explained with your analysis of the idea, it would require an immense amount of extra work to be done properly. Custom firmware, possibly even a new protocol designed specifically for such technology, and then there's no guarantee that it would work. The way I look at it though, when the assets in question are worth severely large amounts of money (well over a few thousand dollars) the addition of this form of physical security, seems to me like it could have a few interesting purposes.

To give a simple example, if one day Eric Schmidt decided he wanted his super secret bat cave (C'mon, 5.5 billion dollar net worth? He HAS to have a bat cave) or one of Google's more important test labs, to suddenly have Wifi access, and the WAP was a good distance away, then this type of idea would undoubtedly come in handy.

To get back to the point though, I completely see where you're coming from, when you say that there are many other things to recognize when securing a system, and didn't mean to make it seem like you were rebuking the O.P. Now that I read over what I posted earlier it does seem like I was putting words in your mouth, and for that I apologize.

0

Share this post


Link to post
Share on other sites

The biggest problem I can see is using a powerful WiFi adapter (like a 2Mw Alfa), high-gain omni directional antenna, and bi-directional RF amplifiers, it might be possible to pick the most directional RF signal at a distance from awkward angles. If the RF signal is directional, it will spill even further, but be narrower. Perhaps adding reflectors behind the antenna of the client or on the inner side of surrounding walls might degrade the RF signal's strength. I don't know how effective that would be, because my home has aluminum siding, and still gets signals from my neighbors' WLAN, and they can pickup the signal from my WLAN. It would be a fun concept to play around with, or add another layer of security when combined with other layers to secure the payload.

EDIT: perhaps I don't understand the original concept. Was the original idea to transfer the key securely, by means of making the RF signal unobtainable by directing it in a secure manner, where no outside attacker would be able to detect it?

Edited by tekio
0

Share this post


Link to post
Share on other sites

Army_of_one, I definitely agree with a lot of what you said, and you made many good points. Just from your response I would venture to say that you've had a job in the field of risk management before.

However, I stand by what I said about the physics behind the idea, and why they're important. Disregarding all other factors, if the idea were to be tested and eventually properly implemented it could quite possibly make obtaining a key a near physical impossibility, and that is a fairly worthwhile security claim. This is, of course assuming that the computer and the WAP are tucked away somewhere safe as well (e.g. somewhere where you can't use the firewire exploit, or steal the hard drive while I'm takin' a piss at Starbucks...like in a locked metal cabinet bolted to the floor :P) To get to the point, after re-evaluating it, I think the idea would be good for higher up government officials and the owners of certain Fortune 500 companies. As you explained with your analysis of the idea, it would require an immense amount of extra work to be done properly. Custom firmware, possibly even a new protocol designed specifically for such technology, and then there's no guarantee that it would work. The way I look at it though, when the assets in question are worth severely large amounts of money (well over a few thousand dollars) the addition of this form of physical security, seems to me like it could have a few interesting purposes.

To give a simple example, if one day Eric Schmidt decided he wanted his super secret bat cave (C'mon, 5.5 billion dollar net worth? He HAS to have a bat cave) or one of Google's more important test labs, to suddenly have Wifi access, and the WAP was a good distance away, then this type of idea would undoubtedly come in handy.

To get back to the point though, I completely see where you're coming from, when you say that there are many other things to recognize when securing a system, and didn't mean to make it seem like you were rebuking the O.P. Now that I read over what I posted earlier it does seem like I was putting words in your mouth, and for that I apologize.

Army_of_one, I definitely agree with a lot of what you said, and you made many good points. Just from your response I would venture to say that you've had a job in the field of risk management before.

However, I stand by what I said about the physics behind the idea, and why they're important. Disregarding all other factors, if the idea were to be tested and eventually properly implemented it could quite possibly make obtaining a key a near physical impossibility, and that is a fairly worthwhile security claim. This is, of course assuming that the computer and the WAP are tucked away somewhere safe as well (e.g. somewhere where you can't use the firewire exploit, or steal the hard drive while I'm takin' a piss at Starbucks...like in a locked metal cabinet bolted to the floor :P) To get to the point, after re-evaluating it, I think the idea would be good for higher up government officials and the owners of certain Fortune 500 companies. As you explained with your analysis of the idea, it would require an immense amount of extra work to be done properly. Custom firmware, possibly even a new protocol designed specifically for such technology, and then there's no guarantee that it would work. The way I look at it though, when the assets in question are worth severely large amounts of money (well over a few thousand dollars) the addition of this form of physical security, seems to me like it could have a few interesting purposes.

To give a simple example, if one day Eric Schmidt decided he wanted his super secret bat cave (C'mon, 5.5 billion dollar net worth? He HAS to have a bat cave) or one of Google's more important test labs, to suddenly have Wifi access, and the WAP was a good distance away, then this type of idea would undoubtedly come in handy.

To get back to the point though, I completely see where you're coming from, when you say that there are many other things to recognize when securing a system, and didn't mean to make it seem like you were rebuking the O.P. Now that I read over what I posted earlier it does seem like I was putting words in your mouth, and for that I apologize.

I appreciate your reply. I think You're correct that high value assets need higher protection and that using physics to our advantage is nice when we can. After all, that's what the so-called air gap strategy is all about: two systems separated by at least half a foot with no shared cabling will prevent a virus from spreading with high confidence. I certainly have done some risk assessments, but that's not my background. My specialty is a niche market that requires high assurance: utter confidence that the system has no serious flaws and will always perform safely. There's never a guarantee, but certain engineering techniques can drastically reduce defects & stop certain threats outright. We look at every kind of attack imaginable, classes of problems, and every aspect of design and operation. Have to deal with everything and provide convincing arguments that the solutions will work. I'm mainly doing it as a hobby right now, but occasionally do production systems. If you're justifying a scheme for high value assets, then I use this kind of mindset when evaluating it. The government standard for this level of security is "High Robustness." Nearly every product on the market doesn't exceed "Low," with an occasional medium (which is low with improvements). So, with my background, I'm evaluating your physics claim like this:

Security claim: 1) Attacker must intercept beam to get keys. 2) It's impossible to intercept the beams due to directional antennas. 3) Attacker can't get the keys and system is secure.

The weak assumption, ignoring all other threats, is No 2. Tekio made some of the same objections I've made. Directional beams of any useful strength continue to radiate from the surfaces they hit. They might even go through windows and walls easily if they are in certain frequency ranges. A sensitive antenna and digital signal processing algorithms would likely collect these stray waves. Because it's hard to reliably control EMF fields, it's hard to say with any confidence the wave can't be intercepted. The only way to reliably ensure this is to use very low power waves and focus them across one known path in a very narrow way, ensuring the majority of probable reflections can't send the wave to attackers. This means the user will have to be in a specific place very close to the access point to use the network. Guess what? An ethernet jack with IPSec is cheaper and has the same level of mobility (read: none). So, the nature of EM waves makes "can't intercept" unbelievable from a physics, technical or usability standpoint. With the limitations to secure it, I don't think the market for this network would be large enough to provide a significant ROI. And it's likely insecure.

There's another avenue of attack, though. This class is called an emanation attack or emanation security (EMSEC). Your computer and nearly all electronics emanate electromagnetic fields that, with the right equipment and expertise, can be used to recreate some or all of whats going on in them. The distance is 3 yards at best and around 100 at worst. Leaks also occur through cabling, from networked cabling to power cords. I include a few links demonstrating attacks like this, although they originally began as far back as WW2. Even if your hardware doesn't leak much, it can be made to leak via active RF injection attacks. In these, they basically fire a beam at your computer and then catch it when it comes back. The EMF fields in your computer alter their beam in a way that conveys information about what was going on. These techniques have been used by researchers and intel agencies to get PIN's from ATM's, recreate whats on a monitor, read your key strokes as you type them (slow delay actually), and pull encryption keys out of RAM. Organized crime has been known to do it for high value assets, just paying some smart college guys or ex-spooks to do it. The government's EMSEC protection program is called TEMPEST and many details are classified, as are the testing procedures. TEMPEST Level 1 computers are safe from most passive attacks and sometimes (!) active attacks. NSA has a list of certified products on the web and the vendors post their product specs and pics (ugly). Unless you have a system like this, you are vulnerable to an EM attack. They cost as little as a few thousand to pull off depending on who you hire. I hope that bat cave has data of phenomenally low value. ;)

So, we have directional antennas bouncing secret data all over a room or area that attackers might bug. We also have both the PC and the access point emanating the key as background EMF over at least 10-20 feet. Catching the bouncy wave costs $50 to a few hundred depending on attacker's technical skills. TEMPEST attack costs a few thousand or a few tens of thousands. The target is Eric Schmidt's fortune. I'm left with a feeling of deep sympathy for Mr. Schmidt's impending losses. :wink: He would have been better off buying NSA-certified Type 1 networking devices for about $80k a year (devices one-time fee + salary of COMSEC custodian required by law) and keeping the main networking equipment in a safe place (WAP can be untrusted in a good design, meaning you can buy Linksys WRT).

NSA spills its guts on TEMPEST attacks

http://www.tomsguide...,news-1207.html

TEMPEST information page (risk & attack capabilities have increased since publication)

http://www.eskimo.co...mpestintro.html

Sniffing Keystrokes with Laser and power lines (btw, I came up with the laser idea independently in late 90's. not that original.)

http://news.cnet.com...0200631-83.html

Making a monitor leak play Beethoven over AM (only time Beethoven feels creepy)

Even a cellphone causes STU-III "secure" fone to leak secrets (active attack)

http://cryptome.org/nsa-tempest.htm

James Atkinson claiming to exploit STU-III leak w/ blackberry

http://groups.google...fe105d515?fwc=1

Great Seal Bug (my favorite EMSEC attack)

http://www.pimall.co...eatsealbug.html

Does this all open your eyes? Protecting high value assets is hard. I usually recommend people avoid wireless access for data requiring high confidentiality. They are better off using a comfortable office or room with a PC wired to the network, with filters on the outward facing main power lines and networking cables if they need EMSEC protection. The facility must also be monitored and access controlled to prevent tampering. An example of tampering is putting some mesh metal near where this network would be accessed so the directional wave becomes omnidirectional. The Soviets used that exact attack against our embassies.

0

Share this post


Link to post
Share on other sites

All I have to say now, is that, that is probably the creepiest/ most invasive type of attack I've ever heard of...I LOVE IT. That said, I downloaded "Tempest for Eliza" on Friday, read the readme file, and haven't stopped playing with it since. So thank you; guess Eric Schmidt is S.O.L. if he ever wants wireless access. :)

0

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!


Register a new account

Sign in

Already have an account? Sign in here.


Sign In Now