Re: [Fonts] filtered Tempest fonts
On 03 Mar 2003 22:21:43 -0500 James H. Cloos Jr. [EMAIL PROTECTED] wrote: Essentially a steg technique, yes? The image transmitted with radio frequencies is different (how exactly?) to that transmitted optically, the image transmitted optically being the decoded text, with the coded image being derived from the message. Now there's a thought (I've just used the word 'message' instead of character). Markus is talking about varying the image for each usage of a character: On Mon, 03 Mar 2003 14:32:28 + Markus Kuhn [EMAIL PROTECTED] wrote: The X11 font system is in addition at present not capable to vary each useage of a glyph, as it would be essential to make sure the filtering does not actually leak more information on digital video links. Why not vary character usage by varying the transmitted image of a word (phrase, paragraph, entire message) as a whole. Might be easier to do and more effective than attempting to vary the usage of individual characters. Not sure if you could do this with Freetype2. Then again adding more to the coded message (radio frequency image) other than recoding an individual character might not be necessary. GSO __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com ___ Fonts mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/fonts
Re: [Fonts] filtered Tempest fonts
Juliusz Chroboczek wrote on 2003-03-03 22:56 UTC: MK Putting an anti-tempest filter into freetype2 has been on my todo list MK for a long time Could you guys be so kind as to tell us mere mortals what you're speaking about? Compromising emanations of video display systems. Low-cost software-only countermeasures. Modifying the font renderer so as to minimize information-carrying radio frequency signals generated by the graphics adapter, video cable and display. This involves primarily two techniques: - Lowpass filtering the glyphs in horizontal direction (removing everything above 70% of the Nyquist limit (half pixel clock) is in practice a good compromise between readability on the screen and unreadability with with radio receiver). - Replace the least few significant bits with pseudo-random bits for each usage of a glyph on the screen. This causes frame-rate correlated noise that eavesdroppers cannot eliminate by periodic averaging. This technique is in particular useful for jamming emissions of VGA cards and digital video cables such as NEC's FPD-Link (used in many laptops) and DVI (used in recent desktop flat-panel displays). Optionally you can also generate glyph variants by slightly offsetting the Bezier control points (hinting is switched off in this anyway), and then randomly pick one of the glyph variants to make radio character recognition slightly more challenging. Results, see pages 31-44 of http://www.cl.cam.ac.uk/~mgk25/em-eaves.pdf I have already written a new techreport with all the gory details on this, which I hope to get around to finalise and publish in a few weeks. [The UK patent office was even so kind to grant me GB2333883 on this stuff. I'd be happy of course to grant XFree86 a free licence.] It's got something to do with deploying XFree86 in the American embassy in Moscow, right? No, these folks put their computers inside 100 dB shielding instead. It is more for people who can neither afford the £5 for an NSA-94-106 certified shielded room (say, the embassy of Lower Slovbodia in Paris), nor the £500 that an intitial psychiatric treatment course for mild paranoia costs. CRT and LCD eavesdropping is for much target equipment quite feasible within around 10-30 meters, in rurals areas sometimes even in the 100 meter range. However, I don't think it is actually done frequently, because there aren't many situations where you know in advance, which screen is going to display something so interesting that it is worth setting up all the fancy equipment necessary to do it. People who should worry about such things seriously usually have already taken care of voice eavesdropping, a in comparison far more significant threat. And once you believe that you have taken care of RF shielding, you can start to worry about optical VHF leaks from CRTs: http://www.cl.cam.ac.uk/~mgk25/emsec/optical-faq.html The fun never ends. Markus -- Markus Kuhn, Computer Lab, Univ of Cambridge, GB http://www.cl.cam.ac.uk/~mgk25/ | __oo_O..O_oo__ ___ Fonts mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/fonts
Re: [Fonts] filtered Tempest fonts
Anti-aliasing, also the psychological principle that we can recognize a letter of the alphabet from a partial or distorted representation (given the context - i.e. if you know what your're looking for). EM radiation measures, but also radio Optical Character Recognition and varying each use of a character/glyth to counter this. __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com ___ Fonts mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/fonts
Re: [Fonts] filtered Tempest fonts
GSO PS A quick plug for the other item on my 'OpenSource the computing GSO environment for Legal work' wish list. A document processor that GSO numbers paragraphs That's a question for comp.text.tex. Juliusz ___ Fonts mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/fonts
Re: [Fonts] filtered Tempest fonts
Juliusz Chroboczek wrote on 2003-03-04 12:16 UTC: MK - Lowpass filtering the glyphs in horizontal direction Why the glyphs? Wouldn't you want to do that for everything that's displayed? Sure, if that's feasible to implement. Text is just the most interesting and most radio-readable part of the screen, because of its high-frequency content. You can also do the filtering in analog right after the DAC on the card of course. I connected a spectrum analyser to a couple of graphics cards and found significant differences in their harmonics output. VGA outputs of some Toshiba laptops go well above 1 GHz with their harmonics, whereas a Matrox cards that I tested seemed to filter quite properly. I personally believe that careful selection of the graphics cards adds far more protection than the font filtering along could ever provide. And if you want to make a PC easy to eavesdrop (sabotage), then just short-circuit the output filters after the DAC. The user might even thank you for sharper pixels ... at the cost of more radio/TV interference. MK - Replace the least few significant bits with pseudo-random bits for each MK usage of a glyph on the screen. Are you implying that what the eavesdroppers get is the derivative of the signal rather than the signal itself? Not quite, but something similar. For analog displays (CRTs), they get a high-pass filtered version of the video signal, replicated throughout the spectrum by the discrete sampling. If you have a 100 MHz pixel clock, then any arbitrary 50 MHz band in the upper VHF or lower UHF range will contain a complete copy of a high-pass filtered version of the full pixel information. Therefore, continuous-tone images (JPEG photos, etc.) are practically impossible to see for an eavesdropper, unless they are dithered significantly. Markus -- Markus Kuhn, Computer Lab, Univ of Cambridge, GB http://www.cl.cam.ac.uk/~mgk25/ | __oo_O..O_oo__ ___ Fonts mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/fonts
Re: [Fonts] filtered Tempest fonts [OT]
That's fun. MK Sure, if that's feasible to implement. Yes, it's easy. And it doesn't break the protocol. Shadowfb, you introduce noise upon blasting to the real framebuffer. Because, GetImage and friends work from the shadowfb, you're not breaking the protocol. You might actually end up with an implementation that is not measurably slower than stock shadowfb. (That's my interpretation of the protocol: an X server that produces a black screen does respect the protocol, as long as GetImage returns the right data. Your interpretation may differ.) Now of course you need to look out for your shadowfb RAM broadcasting ;-) Juliusz ___ Fonts mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/fonts
Re: [Fonts] filtered Tempest fonts
G S Osler wrote on 2003-03-04 13:23 UTC: The question is how feasible is it for an average electronics engineer to install the basics needed. Slashdot Nov 99: New Scientist has an interesting article about a new toy we will all want. It's a card that plugs in one of your PCI slots and allows you to scan the EMF spectrum and read your neighbours terminal. In about 5 years you might be able to get one for just under £1000. (Modern Tempest Hardware costs about £3) http://slashdot.org/yro/99/11/08/093250.shtml Slashdot is not very good in quoting, otherwise you would have noticed that your interlocutor made that quick remark when a New Scientist reporter rang him up sometimes in November 1999. But we are actually getting there, with Analog Devices now shipping an 8-bit 1.5 Gsample/s ADC for under 500 USD and companies such as Echotek starting to produce first affordable data-acquisition boards where that chip is surrounded by a couple of high-end FPGAs to do the interpolating and periodic averaging necessary to lift the compromising emanations out of the noise. Combine that with a decent analog RF preselector front-end and a set of broadband antennas, and you too could be in the Tempest business. Tedious in practice, but not unfeasible. For myself the paranoia set in when I began to use my computer for a £30,000+ legal claim. I probabaly wouldn't start to worry much about compromising emanations of a single device unless £1e6 were involved. Burglary risks might be a far more important concern (as a friend of mine learned painfully recently, when he finally realized the difference between a realdog and a robodog after burglars took his Aibo). Markus -- Markus Kuhn, Computer Lab, Univ of Cambridge, GB http://www.cl.cam.ac.uk/~mgk25/ | __oo_O..O_oo__ ___ Fonts mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/fonts
Re: [Fonts] filtered Tempest fonts
It won't do any harm to start looking at this now. Am not a C programmer as a matter of routine but can handle it if needed so will take a look at freetype2. Cheers, GSO opensource project: http://sourceforge.net/projects/money-go-round __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com ___ Fonts mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/fonts
Re: [Fonts] filtered Tempest fonts
G S Osler wrote on 2003-02-28 21:05 UTC: Do we have any filtered Tempest fonts available for X. BDF was not designed to carry gray-scale bitmap glyphs, as would be necessary to display low-pass filtered fonts. The X11 font system is in addition at present not capable to vary each useage of a glyph, as it would be essential to make sure the filtering does not actually leak more information on digital video links. Putting an anti-tempest filter into freetype2 has been on my todo list for a long time (and might remain there a bit longer given current workload). Markus -- Markus Kuhn, Computer Lab, Univ of Cambridge, GB http://www.cl.cam.ac.uk/~mgk25/ | __oo_O..O_oo__ ___ Fonts mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/fonts
Re: [Fonts] filtered Tempest fonts
MK Putting an anti-tempest filter into freetype2 has been on my todo list MK for a long time Could you guys be so kind as to tell us mere mortals what you're speaking about? It's got something to do with deploying XFree86 in the American embassy in Moscow, right? Juliusz ___ Fonts mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/fonts
Re: [Fonts] filtered Tempest fonts
Juliusz == Juliusz Chroboczek [EMAIL PROTECTED] writes: MK Putting an anti-tempest filter into freetype2 has been on my todo MK list for a long time Juliusz Could you guys be so kind as to tell us mere mortals what Juliusz you're speaking about? Presumably the idea is to manipulate the glyphs in such a way that, while the visual display is not impaired, the recovery of that data from monitoring the monitor's EM emmissions is impaired. Essentially a steg technique, yes? Sounds fun. -JimC ___ Fonts mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/fonts
[Fonts] filtered Tempest fonts
Do we have any filtered Tempest fonts available for X. If not how would I go about converting Markus Kuhn's (http://www.cl.cam.ac.uk/~mgk25/st-fonts.zip). Glyph Bitmap Distribution Format (BDF) Specification (http://partners.adobe.com/asn/developer/pdfs/tn/5005.BDF_Spec.pdf) is the X font spec? Do there exist any utils to make the conversion? thx GSO __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com ___ Fonts mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/fonts