Signing video... (fwd)

Jim Choate ravage at ssz.com
Mon Jan 26 21:19:42 PST 1998



Forwarded message:

> Date: Mon, 26 Jan 1998 19:43:17 -0800
> From: Steve Schear <schear at lvdi.net>
> Subject: Re: Signing video...

> >I was exchangin idea with a couple of members off-list and an idea came to
> >me about how to time stamp the frames. It would require a minor modification
> >to the WWV transmission but it wouldn't be much. Have the WWV transmitters
> >sign each 'tick' of the clock and have a WWV receiver in the vcr. This would
> >allow time stamping the frame with a signed reference that could be checked.
> 
> Although this could reduce the cost and power consumption in the camera how
> does this guarantee that the signature information wasn't inserted later.
> With today's excellent digital postprocessing it could be very hard to
> detect.

The fact that it's hard is what makes it worth doing in some environments.

The goal is to provide a reasonable doubt as to the validity of the picture
on the tape.

Don't know about cost and power consumption...but this is where I'm at
on this issue now, by no means is it fully worked out:

The actual signal on the tape can be mapped to a specific head mechanism
using the same sorts of technology that is used to identify specific cell
phones from the rf signature. So we can do an analysis of the signal and
verify if a given camera striped the tape. Hysterisis, amplifier slew rates,
power spectrum analysis, etc.

The pattern of wow and flutter of any motor and drive mechanism is unique
and can also be mapped to a specific mechanism. This wow and flutter creates
regular and repeatable irregularities in the pattern on the tape. This
allows us to match a drive mechanism to a specific program on a tape.
Irregularities in the drive wheel will cause specific pattern in the way the
stripes are laid down (ie wiggles in the same place each time that part of
the wheel comes around).

Irregularities in the syncronizing clocks in the tape mechanism can be
traced to particular oscillators (usualy slew rate as a function of
frequency stability).

By signing the frames with a signed time reference we can verify which clock
was used, though we can't verify that it wasn't recorded and played back at
a later time. We can at least verify the tape wasn't stiped prior to the
claimed time. For security applications this could be quite useful.

So if we have a fully analog system there are several means to test the
various parts of the systems to verify source veracity. We can also test the
tape recording mechanism even on a digital or hybrid system.

So let's look at hybrid systems where the video signal is source digital or
is digitized in the camera (for special effects, etc.).

If a frame isn't signed we don't trust it.

So we have a camera with a ROM that has a signature. We could replace the
code in the rom and keep the signature. If the owner doesn't replace the ROM
prior to testing then we can easily verify the ROM and it's signature don't
match. We don't even need the signature to do this, simple comparison
between the rom that should be there versus the one that is present is
enough.

If they do replace the ROM then we have a frame that is signed but potentialy
modified prior to the d/a conversion process. The question is can we in fact
identify such non-linear editing? If the answer is no then there is a clear
reason to suspect the validity of any video source that went through any
sort of digital processing.

I suspect the answer is no since this question is equivalent to signing a
document that was modified prior to the signature being placed on it. The
signature process only signs what was fed into the algorithm and can't
detect changes to the document prior to the signing.



    ____________________________________________________________________
   |                                                                    |
   |       The most powerful passion in life is not love or hate,       |
   |       but the desire to edit somebody elses words.                 |
   |                                                                    |
   |                                  Sign in Ed Barsis' office         |
   |                                                                    | 
   |            _____                             The Armadillo Group   |
   |         ,::////;::-.                           Austin, Tx. USA     |
   |        /:'///// ``::>/|/                     http://www.ssz.com/   |
   |      .',  ||||    `/( e\                                           |
   |  -====~~mm-'`-```-mm --'-                         Jim Choate       |
   |                                                 ravage at ssz.com     |
   |                                                  512-451-7087      |
   |____________________________________________________________________|







More information about the cypherpunks-legacy mailing list