Copy protection schemes for copy protected digital material6473560Abstract Recent copy-protection schemes for digital video copy-protection rely on a pixel-watermark together with the presence of a `ticket` to implement the `copy-once` state of content. This ticket is designed to entitle the owner of a disk/tape to making one copy of digital content. We disclose a method to pass along this ticket with the video signal after content has been converted to the analog domain. This form of the ticket is called the analog ticket. The analog signal may be recorded on a legacy device that does not comply with the copy-protection scheme (e.g. ordinary VCRs). The analog ticket is designed such that it is automatically removed in these legacy devices. Next generation copies by compliant devices are made impossible. Three possible ways of embedding the ticket signal into the analog video signal are proposed. Claims What is claimed is: Description BACKGROUND OF THE INVENTION
+ + - -
VBI lines _Easy to integrate for a _Broadcaster needs to be
compliant broadcaster. aware: must be spliced into
existing TXT, CC etc. signals
_May be deleted in
International Broadcast chain
3% _Will survive an entire _Noise reduction/format
overscan broadcast chain conversion/filtering on active
_No extra hardware in DVD video may harm ticket in this
player necessary. region.
_Could be visible on some
monitors (e.g. underscanned
PC monitor)
_Does not get removed in
Digital VCR
A first proposal is to transmit during VBI or the 3% overscan lines an Amplitude Modulated Y-signal, with a carrier frequency of approx. 3-4 MHz, and high amplitude, with DC-level 50% (gray). For instance, with blanking level=0%, peak white=100%, synchronization level=-43%: binary "0" has amplitude of 80% and "1" an amplitude of 100%. This means that peak levels of these signals will be 129% resp. 150%. An example of a ticket, embedded in this way, is shown in FIG. 2. The example shows the embedding of a bit sequence `10110` of the ticket in a line of the video signal, by AM-modulating a high-frequency, high amplitude Y-signal (not to scale). During recording, high (spatial) frequencies are amplified in the VCR (the so-called pre-emphasis), to compensate for undesirable properties of magnetic tape. To limit the frequency excursions of the FM modulator that records the Y-signal, the pre-emphasized signal is clipped before recording. This clipping removes any high-amplitude information in Y. FIG. 3 shows an embodiment of an apparatus for embedding the analog-ticket as shown in FIG. 2 in the analog video signal. The embedder of FIG. 3 embeds the analog ticket in line n of the video signal. The data bits are transmitted at 614 kHz using 50% DC-offset and amplitudes 80% and 100% to represent "0" and "1". This allows for approximately 32 bits/line. The "Gate VBI line n", is a monstable multivibrator which emits a pulse, as soon as its input sees the number n. A second proposal is to shift the phase of the colour burst in a predetermined line over 180.degree.. At the beginning of every line, before the active part of the line, the broadcaster transmits the colour burst: 10 to 12 cycles of the colour subcarrier frequency, f.sub.SC which has been fixed to 4.433618.75 MHz (for PAL) or 3.579545 MHz (for NTCS). The receiver synchronizes an on-board oscillator in frequency and phase to this burst, and uses it to decode the ensuing U,V (also called R-Y,B-Y) (PAL), or I,Q (NTCS) colour-signals which have been quadrature amplitude modulated (PAL/NTSC) onto the colour subcarrier. While recording on a magnetic tape in a VCR, the colour signals which modulate the colour subcarrier, are shifted down from the upper part of the spectrum (around f.sub.SC) to the lower part of the spectrum and centred around f.sub.U.apprxeq.625 kHz (PAL), 690 kHz (NTSC). These figures are for VHS. For Beta, Hi-8 and S-VHS, the sub-carrier frequencies lie in the same range (between 550 and 750 kHz). To be precise: if the analog input of the VCR sees a colour signal with a burst-frequency f.sub.SC ' and horizontal line frequency f.sub.H ' (both not necessarily equal to those specified in the standard), the colour burst frequency on tape should be f.sub.U =N.f.sub.H ', N being equal to 40+1/8 for VHS and being equal to 47-1/8 for 8 mm. The burst phase of f.sub.U is locked to f.sub.SC '. The above described procedure is called `colour-under` recording. On playback the VCR colour-under electronics will translate the colour signal from tape back to its original range around f.sub.SC in the upper part of the spectrum by mixing it in a heterodyne mixer with the sum frequency f.sub..DELTA..apprxeq.f.sub.SC +f.sub.U. The playback f.sub.SC is generated by a crystal oscillator, and f.sub.u is generated by a VCO which is locked to the colour-burst coming from tape. Variations in the design of the so-called sub-converter and the main converter exist, but there are always two oscillators, one of which is a stable X-tal reference oscillator and the other is locked to a (multiple of) the burst frequency. Timing instabilities (jitter) in the tape transportation system have a disastrous influence on the subcarrier phase and therefore the colour quality. Two phase comparison loops, a coarse control loop and a fine control loop, quickly compensate for these instabilities, by adjusting the heterodyne mixer frequency to produce a burst that comes out to the standard 4.433618.75 or 3.579545 MHz. Note that some intentional modulation of the burst phase/frequency in the input signal to the VCR will be lost in the output by this locking of the output burst to a reference crystal. The fine control loop "slowly" adjusts the heterodyne mixing frequency f.sub..DELTA. by comparing the phase difference of the reference crystal oscillator and the up-converted burst from tape, averaged over a few lines. The coarse control loop responds "quickly" to large phase differences. To understand its working, one should know that in VHS and its descendants, the phase of the colour subcarrier, and thus the burst as recorded by head-A (one of the two recording heads present in the VCR) is advanced by 90.degree., every horizontal line; for PAL this happens in addition to the standard .+-.135.degree. burst-alternation. For head-B (the other recording head in the VCR) no extra phases are introduced. This is done to reduce chroma-crosstalk between adjacent tracks. It should be noted here, that in the 8-mm system, the roles of heads A and B are interchanged. Now on playback, when comparator in the coarse control loop in the colour synchronization circuitry notices a large phase difference it can respond in two ways: the phase difference is big because (a) a large error has occurred; in that case the coarse control loop shuts off the burst and colour-signal for that line. (b) the oscillator is in the wrong phase of the 0.degree..fwdarw.90.degree..fwdarw.180.degree..fwdarw.270.degree..fwdarw.0 .degree. set of jumps; in that case, the coarse control loop advances the onboard oscillator for f.sub.u in steps of 180.degree. (or sometimes 90.degree.) until locking occurs. This is called the 4.PHI.-selector. To transmit a "0", advance the burst by 180.degree. on a fixed line n (say n=18). To transmit a "1", advance the burst by 180.degree. on line n+1. This "0" or "1" is detected by measuring the phase of the burst on line n+1. There are two ways in which this information is distorted on playback: 1. The coarse control loop sees a 180.degree. f.sub.SC -phase error, and responds by advancing the heterodyne frequency f.sub.u by 90.degree. or 180.degree.. 2. Even if the coarse control loop does not respond at all, there is a colour comb-filter, present in every VCR, to reduce the cross-talk between adjacent tracks. This 2H comb-filter puts out the average of line n and n-2 (for PAL; for NTSC: 1H, average of lines n and n-1). If the burst on line n is 180.degree. advanced, that will average to 0 with the burst on line n-2. In this way, one or two 2 bits per frame can be transmitted. The number of bits is limited because it takes the coarse control loop a few lines (typically 5-8) to recover, and obviously this recovery needs to take place before video lines are transmitted which are viewable on the screen. The following table shows what the amplitude A and phase .phi. of the burst on line n+1 will be after playback from a VCR with and without a responding coarse control loop.
4.PHI.-selector 4.PHI.-selector
advances 180.degree. does not act
2H 1H No 2H 1H No
Transmitted Comb Comb Comb Comb Comb Comb
bit A .theta. A .theta. A .theta. A .theta. A .theta.
A .theta.
"0" 0 -- 0 -- A.sub.0 180.degree. 0 -- 0 --
A.sub.0 180.degree.
"1" 0 -- A.sub.0 180.degree. A.sub.0 180.degree. A.sub.0
0.degree. 0 -- A.sub.0 0.degree.
The above table shows the amplitude and the phase-shift of the burst on line n+1 after transmitting a "0" or a "1" bit, after recording and playing back the bits by an analog VCR with various PLL-configurations. In this table, A.sub.0 is the normal burst-amplitude (approx. 43%). The (pathological case of the) last column shows what is measured on line n+1 without going through a VCR (i.e. directly at the output of say a DVD player). Notice that in none of the VCR configurations both "0" and "1" can be detected reliably upon playback. An example: transmit a "0" to a VCR with a 1H comb-filter and a 4.PHI.-selector that advances in steps of 180.degree.. Because of the "0", the burst on line n is 180.degree. ahead and the burst in line n+1 is 0.degree. ahead. On playback, the 4.PHI.-selector advances the burst on line n+1 by 180.degree. upon detecting the "bad" burst on line n. Then the 1H comb filter averages this with line (n+1)-1=n, which was already 180.degree. ahead, so the result is a burst with normal amplitude and 180.degree. phase-shift. A third proposal is based on the relationship that normally exists between the line frequency f.sub.H and the subcarrier frequency f.sub.SC. In a particular line in the VBI or the 3% overscan region, a signal is transmitted with approximately the same frequency as the colour subcarrier, f.sub.SC, which is phase-modulated with the bits of the ticket. I.e. "0" and "1" correspond to two colours, say at 0.degree. and 90.degree. in U-V space. The frequency of the ticket signal should be very close to (284-1/4) f.sub.H +25 Hz(PAL) or (455/2).times.f.sub.H (NTSC). However the following rules should be taken into account: For PAL: (a) this ticket-signal is transmitted only during even frames, not during odd frames, and (b) the ticket signal is inverted every 2.sup.nd, 6.sup.th, 10.sup.th etc. frame. For NTSC (a) this ticket signal is transmitted during all frames, and (b) the ticket signal is inverted every 2.sup.nd, 4.sup.th, 6.sup.th etc. frame. The inversion of the ticket-signal corresponds to assigning two new colours to "0" and "1 " which are shifted 180.degree. ahead. As a result of the fact that the ticket signal is locked to f.sub.H, the ticket signal has a fixed phase with respect to the horizontal synchronization pulse, on that line: i.e. on an oscilloscope using this horizonal synch pulse as trigger, the ticket-signal is steady. If the inversion in step b were not performed, the ticket-signal would appear together with its 180.degree.-shifted version, because of the fractional relation between f.sub.SC and f.sub.H.(for PAL, after two frames, the colour subcarrier on the same line has picked up a phase-delay of 180.degree., whilst for NTSC, after one frame, the colour-subcarrier on the same line has picked up a phase-delay of 180.degree.). The ticket signal is detected by averaging this line over a number of frames. This ticket signal is removed upon recording as follows: the frequency on which the ticket is phase-modulated is basically the colour-subcarrier.The VCR will interpret the ticket-signal as a sequence of colours. During playback, the VCR will lock this "Colour-signal" to a reference crystal-oscillator at f.sub.SC. The horizontal line frequency f.sub.H however, is unstable, because of jitter in the mechanics of the tape-transportation system. Therefore, on playback, the ticket signal no longer bears any fixed phase-relationship to f.sub.H and when added up over several frames will average to 0. FIG. 4 shows oscilloscope traces of the ticket before recording on a VCR and after playback from that VCR. In FIG. 4, the modulation parameter is "1" is +45.degree., "0" is -45.degree.. At the top: before recording on a VCR; at the bottom: after playback from that VCR .The vertical dashed lines indicate bit-boundaries: the places where the phase of the carrier may be changed to encode bit. It is clear from the picture at the top that all traces fall on top of each other, whereas in the picture at the bottom the carrier jitters back and forth, and averages away. FIG. 5 shows an embodiment of an apparatus for embedding the ticket as per the third proposal into the analog video signal. With the embedder of FIG. 5, the data bits are transmitted at 614 kHz using 0.degree. and 180.degree. to represent "0" and "1". This allows for approximately 32 bits/line. The ".div.2" block in FIG. 5 is for NTSC, and should become a ".div.4" block for PAL. The "Gate VBI line n" block in FIG. 5, is a monstable multivibrator which emits a pulse, as soon as its input sees the number n. For the transmission of the ticket in accordance with the second and third proposal, it is recommended to set the luminance-signal Y on the ensuing line to zero. If the luminance signal in the line is not zero, this does not hinder the methods, but taking the luminance signal equal to zero will merely mask any potential artifacts. The tables below summarize and list the advantages and disadvantages of the three proposals described above.
Medium Aggressor Method
1 AM a carrier .apprxeq. 4 MHz in Y- Pre-emphasis + limiter- R-IN
space at high circuit
amplitude (e.g. "0" = 1.2 V;
"1" = 1.34 V)
2 Modulate Colour Burst phase
0.degree./90.degree./180.degree./270.degree. R/D(?)-
rotary 4.PHI.-selector OUT
and/or comb-filter.
3 PSK f.sub.sc + inversion on Jitter in mechanical tape- R/D(?)-
successive frames transport, i.e. f.sub.H OUT
where R means `removed` and D means `distorted`.
+ + - -
1 High freq. & high _Simple/cheap _Clipping may be
volume Luma _Works with SECAM device dependent.
_High bit-rate
2 PSK Burst _Does not survive _On replay phase
MPEG/D-VCR, even comp. may be too
when slow and leave trig-
in the 3% overscanned gerable version
region of active video.2 of ticket.
_Lower bit-rate
May cause colour
errors in older equip-
ment (w/o quick phase
restore).
_CH-1/CH-2 problem
_No colour burst in
SECAM
3 PSK f.sub.sc + _High bit-rate _May not work in
inversion on SECAM,
successive frames _Averaging over
frames may be expen-
sive (memories etc.).
_Survives a D-VCR
As regards security, the following comments could be made. The ticket could be spliced into the analog signal upon playback by pirates. This could be prevented by relating the ticket to specific properties of the active video of the frame in which it is embedded. Upon playback, the analog video is mutilated (eg. the high frequency colours are stripped, but also loss of Y-resolution would work), and the illegally inserted ticket would no longer correspond to the underlying video. The requirements on an analog ticket can now be summarized as follows. Tickets should not be recognizable by compliant DVD recorders after a recording and subsequent reproduction step on/from an analog VCR, such as disclosed in FIG. 1. The ticket may be rendered unreadable by VCR (IN) (during recording) or in VCR (OUT) (during playback). Furthermore the ticket may be completely removed, or merely distorted (e.g. shifted in frequency). In order of preference:
VCR (IN) VCR (OUT)
Removed (R) - ++
Distorted (D) -- +
Complete removal beats distortion, to make it more difficult for hackers to restore the ticket by hand. Deletion in VCR(OUT) rather than VCR(IN) is preferable because a compliant DVD-recorder or future VCR will not replay (otherwise legal) copy-once material without a ticket. The ticket consists of 64 bits which should be transmitted approximately once per 10 seconds. Because of the analog nature of the channel, ECC should be applied and will roughly double the number of bits to 128. It should be difficult (cryptographically/computationally) to reinsert the ticket into analog video. Transmission of the ticket in analog video should require as little hardware as possible beyond that already in place to convert from MPEG to CVBS.
|
Same subclass Same class Consider this |
||||||||||