Monitoring Your Monitor

bje2 writes "Rememeber this story from a couple months ago about reconstructing data from the blinking LEDs of modems...well, CNet is running a story about reconstructing the display of a computer by using special hardware and the reflected glow of the monitor." Kuhn's paper (400k PDF) is available.

Comments

Usefulness?

[supercytro]

If you're in a situation in which you can take advantage of this effect, why not just look at the monitor itself?

Practical in the "REAL" world?

[RobertAG]

"Until that's resolved, the safest solution is to compute with the lights on. "

Or just close the window shades.

It seems like you can read the contents of a monitor under optimal conditions, but how often do you get optimal conditions? More often that not, a person sits in front of a monitor as he or she works. At best, then, you'd only be able to get bits and pieces of what's on the screen. You also have to contend with different grades of wall paint and/or wallpaper (not to mention furniture behind you) which might make this endeavor fruitless in most cases.

It's a nice trick in a lab, and probably worth publishing. But I think there are too many uncontrollable variables to make this practical.

Guess what....

[geewiz45]

My boss is an old spook who spent time at NASA and some other defense contractors. While there (about 10 years ago) he worked on this project. From what he tells me, they were able to monitor displays from a good distance without any troubles. Not only did they do this ten years ago, they also developed a coating for the monitors that would reduce the radio emissions. That way no one could monitor them...

I don't know what this guy patented but it's already been deemed useless by 10 year old US Government research.

This is not news.

[Ketnar]

Jeeze, this is going around your E to your A, people.

Anybody remember the tempest device? Able to lock on to a RF signal produced by hardware and reconstruct it, get displays, and rummored to be able to even spy on a CPU's activity if finely tuned enough. I read a rather lenghty article on how to build one years ago, but I'll be damned if I remember where I found it. I suggest a goodle hunt. The frightening thing is, the people who built it, were able to pick out a single display in a large office building -- eight blocks away.

RF signals are easyer to get to than the reflection of a monitors glow, I would think.

A guess as to how it works...

[markmoss]

I'm not any sort of expert in this, however from what I know of video the process has to be something like this:

The phosphors in the CRT do not emit only when hit by the electron beam. They have a certain persistence, so a dot keeps on glowing while the beam moves on through other dots. If you get a perfect recording of the signal, then reconstructing the picture requires merely syncing onto the video scan by means of the long and short black intervals (vertical and horizontal retrace), calculating each pixel's actual output by subtracting the fading output of previous pixels, and feeding the resulting video and sync into your own monitor.

However in using this in a normal "spying" situation, you get room lights and other "noise" in the signal. You've got to guess at the average ambient level and compensate (subtract it out) so the picture isn't washed out. Then, you are probably working with such a low level of signal per pixel that quantum fluctuations add significant noise. Subtracting signals accentuates the noise, so you'll wind up with a pretty grainy picture -- after lots of trial and error adjustments to find the best background level compensation, pixel fade rate, etc. But most data on computers is presented in quite high contrast, and stays on the screen for quite a while, so you can improve the picture by averaging frames. So it does sound possible to get a good enough picture for most espionage purposes (extracting text and diagrams, or sometimes just finding out what the guy is reading).

What it probably won't do unless you get really close:
-Spy on your Quake rivals; (I assume, not being a
Quake player myself) the picture changes too fast for frame-averaging to help much, and in general it's a detailed, lower contrast picture so graininess would have a greater impact.
-Pirate the Playboy channel from your rich neighbor, unless you are so hard up that just staring at a screen of approximately fleshtoned grains and imagining there's a nekkid woman somewhere in there is enough...
-Steal passwords protected by the "*" character, unless the login was incompetently programmed and it shows the actual character for a frame before covering it up. And probably not even then, because frame-averaging will often be needed for legibility...

Just handwaving here, but I expect that if someone can get a camera where this process works for any of the above, they probably could have focused it right on the screen and also physically wire-tapped the machine.

I don't buy it.

[REALMAN]

It all sound pretty bogus to me. The claim that blinking LED's can be used to reconstruct what you do on the computer is laughable. To reconstruct from a LED what is being written you would have to be able to correlate each blink as one bit. a 0 or a 1. 8 bits per byte, 1024 bytes per kilobyte etc.. ad nauseum.

Let's use the first half of a ripped mpg version of Star Wars Attack of the Clones.

The first half is 701 MegaBytes or 5,883,382,624 bits (that's close to 6 billion bits)
I can write that on my hard drive in 2 minutes which gives us apporximately 49,028,188 bits per second.

Now can anyone tell me that an LED is capable of blinking at a rate of 49 million times per second? And if it can are we able to discern 49 million blinks with the technology we have? From a distance?

Please...