[Cryptography] TV set power correlates to TV channel?

Jerry Leichter leichter at lrw.com
Sat Dec 3 07:47:37 EST 2016


> "LED" TVs really use the LEDs just for the backlight (older TVs used
> fluorescent lights for backlighting).  The picture itself is formed by
> an LCD (liquid crystal display) panel that sits in front of the
> backlight.  LCD displays -- not their backlights -- do draw different
> amounts of power at each instant, based on what fraction of the color
> subpixels have their transistor driven to the voltage for black versus
> the voltage for illumination.  For example, a small PixelQi laptop LCD
> screen that I have the specs for, draws 199 mA at 3.3V when showing an
> entirely black screen, and 280 mA at 3.3V when showing an entirely
> white screen....
It's worth pointing out - since we're getting into the details of the technology - that OLED screens are very different:  Rather than filtering a backlight, each pixel is a small OLED, so the power drawn depends on the brightness of the pixels.  A black screen draws almost no power; a bright white screen, maximum power.  LCD's should only draw significant power while switching (I'm not sure what exactly the quoted specs are for), while OLED's draw power continuously.

I would guess - given tons of published work on similar attacks - that it would be possible to correlate the power drawn by an OLED screen with a database of "power signatures" and pretty easily determine the show that was being watched.  Doing this from the DC side of the power supply is likely just as easy, since not much else draws varying amounts of power - and what does (the audio output comes to mind) is just as correlated with the show being watched.

If you move to the AC side of the power supply, your signal is going to get distorted.  You'd be looking at the input of a digital converter that likely has a significant amount of capacitive filtering which will level out the power draw.  At the least, this will serve as a low-pass filter on the signal.  I'd guess you could still do it, but it would take a longer sample.

Now move out to the point where you're seeing the power drawn by the whole house.  Yes, the signal will still be there - but whether you can pull it out from other junk on the line, and the distortions introduced by the house circuitry, is a difficult question to answer without trying it.  Note that at this point we can deliver gigabit-speed streams of data over house wiring; but doing so requires substantial intelligence and cooperation between both sender and receiver.

Of course, you do get the question:  Why would anyone bother?  The information is already available to whoever is supplying you with the signal.  Various slightly more intrusive mechanisms - how many TV's are in rooms without outside windows? - would be simpler, cheaper, and more reliable.
                                                        -- Jerry



More information about the cryptography mailing list