On Sat, Nov 03, 2012 at 06:44:09PM +0100, Morten Reistad wrote:
> In article <
5511f19dd3...@gmail.com>,
>
> We don't have to even inform the makers of television content. They
> can keep churning out their game shows, talk hosts, reality series,
> slanted news coverage, and infomersials. And ads.
>
> The levels of bandwidth used in broadcast media are so staggeringly
> much higher than what would be useful for clandestine communications,
> that hiding it gives you thousands of alternatives.
>
> Even a super-compressed TV station uses about 2 megabits of sustained
> traffic. That is when you regularly see the pixelation effects.
> A lousy over-compressed radio uses 128 kilobits, but that is so low
> people will balk at the music and quality of the adverts; so feeds
> below 192 kilobits are rare.
I was primarily considering coding schemes that don't require anything
more complicated than the human ear and pre-arranged code-phrases. In
the eighties and nineties, any decoding machine would be large enough
to be discovered in a determined search of a premises by potentially
unfriendly authorities.
> This newsgroup has a sustained bandwith use of 15 baud; 6 if we
> drop the quotes. I would assert that having access to communications
> assets of a few scores of baud, sustained, would be more than
> enough for a pretty big spy network.
Messages would necessarily be short and infrequent in this scenario,
and you're correct. A few thousand American agents living in Canada
could be sent instructions with a few bits an hour embedded in the
carrier medium. We're not talking about paragraphs and paragraphs of
data, although they might need a significantly larger channel for
information returning from the field but that is a separate problem
entirely.
> The total telegram and telex capacity between the US and the UK
> during WW2 never exceeded 5000 baud. They coordinated the biggest
> war in history through that, in addition to doing business as
> usual.
Convoys crossing the Atlantic must have carried quite a bit of traffic
as well, but that isn't real-time. Is it possible they used radio?
We always hear of the Brits cracking Enigma, but wouldn't the Allies
have had or wanted a similar capability? The disclosure of wartime
records was always something of a war of omission; they don't really
like to talk about any of it. I suspect a lot of espionage *analysis*
involves extracting as much as possible from small bits of
information, which is the province of experts. The layperson is at a
disadvantage in the discussion of such matters primarily because the
day-to-day context of intelligence operations is as good as another
world to them. A show like "Person of Interest" distorts the reality
in the service of its entertainment narrative, and obviously the
writers won't themselves be trained in matters of spycraft. All they
can do is write stories that will entertain their target audience in
the hopes they will sit there for the advertisements.
> One thing that cable tv and internet companies have to contend
> with is that commercial contracts by "the system" absolutely
> and totally forbids any form of recoding of content. This means
> that the bitstream from the advertiser goes directly from the
> show encoder to the consumer box unaltered. It can be framed, e.g.
> in ATM or IP, but content bits are holy.
If the advertising company is in the loop, they can add your
hypothetical encoding during the editing process. They already use
nonlinear editing software so adding the steganography software to the
pipeline is trivial. My approach requires the copy-writers fashioning
the text the actors speak; yours only requires suborning the editing
staff.
> So, let us assume 100 baud, about 7 times what this newsgroup
> consumes. That is 4 bits out of ca 80000 for every
> TV frame. This is about one order of magnitude less than the
> unassigned bits in the current video encryption standards. Or
> four bit changes per frame. You can hide this very well inside
> a rigged dct encoder, subtly changing hues in a small part of the
> picture.
You've also got the audio stream to work with, closed captioning
streams, etc. The only trouble is that you need equipment to decode
the stream and extract your naughty bits. Mind you, HDTV converter
boxes that allow people to watch digital TV on their old analog sets
are fairly inexpensive. I suppose our hypothetical spies can be
assumed to have been supplied with that equipment in advance or might
order one from a web-retail front, but that is beginning to introduce
multiple points of failure.
> And that bit stream reaches the other end totally intact. This
> encoding has actually fostered a whole service industry that
> operate such encoders. You only need a few, select people in
> such an industry; and you are set.
>
> The US telephone network has done something like this since the
> inception of digital signals. One in 8 least significant bits
> from a 8kHz 8-bit signal is "stolen" to provide a 1 kilobit
> framing and signalling channel. This is audible, and is a large
> part of the "uLaw hiss" that all leftopondians are so familiar
> with.
>
> that is a 1:64 "bit theft" without any attempt at concealment.
> Running 100 bps in a tv signal is a 1:8000 "bit theft",and it
> can be concealed very well.
>
> Even the operators of the encoders does not have to know anything,
> this can me coded in the software of the encoders.
Throw in digital satellite radio as well as an alternate transport.
However your scheme doesn't work for pre-digital technologies. NTSC
(known in some broadcast circles as Never The Same Color) loses badly
as a carrier of digital information. All the broadcasters and cable
operators would have to have carefully calibrated equipment from the
tape decks through the amplifiers and transmitters all the way to the
cable head ends. The quality of the signal from tape to the cable
subscriber would need to be really good to encode much information,
but as you say, low bit rates are all that's required and I don't know
enough about analog signaling to quantify the constraints.
There's another theoretically interesting approach. Today's TV's go
all blue-screen when confronted with a non-assigned channel, but way
back in the Nineteen Eighties and earlier a weak or absent signal
produced random "snow" for a picture and the audio amplifier went nuts
trying to amplify the noise floor of the circuit. A crypto stream
resembles nothing so much as noise, so at least in so far as the
civilian population is concerned, a transmitter broadcasting on a
non-assigned channel is probably going to look exactly the same as if
there were no signal. Not going to fool the FCC or the military
because they're going to detect the radio energy if it's not carried
within a shielded cable, so limited utility if we're discussing
cross-border TV transmissions, cable broadcasts, or satellite
television.
Therefore, pre-digital schemes would almost invariably had to rely on
stacking the content with a covert signal. Before 1900, what then?
Newspaper, and perhaps books. An interesting thought. We tend to
forget that anything happened if it didn't happen in the last month,
let alone before we were born. Espionage has been around forever,
and I think I've read of early ciphers used by officials in ancient
Rome, so long before we got the magic of HDTV people have clearly been
working on clandestine communications.
English obviously is great for this sort of thing because words can
and do have multiple meanings. If "owl" is a codeword among your
spies to mean "explode" then the phrase "I saw the barn owl last
night" has an innocuous meaning to an uninformed listener, and an
obvious meaning to members of the group. Everyone's probably heard
about the infamous Masonic handshake, but a handshake among electronic
peripherals is a rigidly defined protocol sequence. They probably
mean an exchange of a conversational nature which would sound
innocuous to the layperson as in the previous example. You can
elaborate on these substitutions schemes arbitrarily. Low bit rates
can be achieved in a text by fooling around with the punctuation
according to a predetermined standard, and much more. With a little
trouble, text can be aligned on page boundaries so it may be taken out
of the context of the narrative if you know what you're looking for.
> >I suppose the need in this regard depends on how much spying and
> >espionage relies on these corner-cases of covert communications. One
> >of the primary objectives of network television is to induce people to
> >buy the products made by television advertisers, whose products are
> >increasingly featured within the television programs themselves.
> >There's only so much traffic that can ride on top of that primary
> >purpose before it is affected by a third-party agenda and people start
> >taking notice. Such covert messaging should be rare to be effective.
>
> You can run the show perfectly intact, ads and all. You just introduce
> some subtle "data loss" in the compression, and it can be so subtle
> that you would notice tens of other compression effects before noticing
> this one. Perfect for TV.
>
> And then you would have to find it, decode it and decrypt it.
That's a game for the big agencies. Non affiliated individuals won't
have the faintest idea of where to start. How do you look for
something that probably isn't there most of the time? Impossible.