Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
CA 02364433 2001-08-08
WO 00/54453 PCT/US00/06296
SIGNAL PROCESSING METHODS, DEVICES, AND APPLICATIONS FOR DIGITAL
RIGHTS MANAGEMENT
Field of the Invention
This invention relates to the field of signal processing, and more
particularly relates to
techniques useful in encoding audio, video, and other content for digital
rights management purposes.
Background of the Invention
With the recent explosive growth in the use of electronic information,
enforcement of copyright
laws has become more difficult. The cost of the equipment required to copy
digital data representing
music, art, and other valuable information has been decreasing, while the
capacity of readily available
data storage media has been increasing. Inexpensive devices can write enormous
amounts of data to
digital storage media such as writable compact disks (CD-R or CD-RWs), multi-
gigabyte hard disk
drives, high capacity removable magnetic disks, and soon to be available
digital versatile disks (DVDs).
Readily available high-resolution printers and scanners bring the digitization
and reproduction of graphic
information within the means of most consumers. In addition, readily available
high-resolution sound
cards, including analog-to-digital and digital-to-analog converters, bring the
digitization and reproduction
of audio information within the means of most consumers. Not only is copying
digital files simple and
inexpensive, the Internet facilitates unauthorized distribution of copyrighted
works.
Unlike analog copies, which are always inferior to the original, a copy of
digital information can
be identical to that of the original, with no degradation due to copying.
Millions of dollars are lost
annually due to illegal but exact duplications of digital media and near-exact
duplications of analog
media. Because copying equipment is readily available, catching persons making
unauthorized copies
can be difficult. Even if an unauthorized copier is apprehended, the creator
of the original document
must still prove that the allegedly unauthorized copy was in fact copied from
his original work and not
independently created.
In one aspect, the technology detailed below relates to digital watermarking,
or "data hiding,"
and its use in solving the problem of illegal copying (e.g., hiding
authentication information or copy
protection information within the original data). Hiding auxiliary information
in original data, also called
steganography, has been used for thousands of years. In steganography, a
message is hidden within
another object or media, so that the message is essentially imperceptible to a
human observer (or
listener). Steganography is related to, but different from, cryptography, in
which the existence of a
message is typically obvious, but its meaning is not ascertainable without
special knowledge.
Hidden data, also referred to as auxiliary or embedded data, can be used to
prevent unauthorized
copying by embedding in the original data commands that are readable by the
copying device and that
instruct the copying device not to make a usable copy. Hidden data can also be
used to authenticate data,
that is, to prove authorship. One such technique entails embedding auxiliary
information in an original
work in such a manner that special knowledge, such as a secret algorithm or
code, is required to detect
CA 02364433 2001-08-08
WO 00/54453 PCT/US00/06296
2
and/or remove the auxiliary information. The copier would not be able to
remove the authentication
information, and the original creator could prove his authorship by retrieving
the embedded information,
which would identify him as the author.
Data hiding has uses besides the prevention and detection of unauthorized
copying. One such
use is content enhancement, that is, adding information to the original data
to enhance the content. For
example, lyrics could be embedded in audio data on a CD. The lyrics could be
viewed in a special
karaoke machine, while the audio could be played on an existing CD player.
Hidden data could also be
used to associate different segments of video data with different viewer-
selectable versions of the video
on a DVD. For example, a viewer could select between a version edited for
children or an unabridged
version, and embedded auxiliary data would indicate to the DVD player which
video segments to skip
and which to include for the selected version.
The original data in which the auxiliary data is hidden may represent any type
of information that
is perceivable with the aid of a presenting device. For example, the data may
represent music which is
presented using a compact disk or audio DVD player, a video film that is
presented on a DVD player, or an
image that is presented on a computer screen or a printer.
When the combined data is presented to a user by a normal presentation device,
the auxiliary
data should not interfere with the use of the original data. Ideally, the user
should not be able to perceive
the auxiliary data at all. Unfortunately, increasing the amount of the
embedded auxiliary data or its
robustness, that is, its persistence to attack and data transformation, may
incidentally increase its
perceptibility. The degree to which the auxiliary data can be perceived
without having an adverse impact
on the user varies with the application. For example, in CD quality audio, a
minor change from the
original data might result in unacceptable audio artifacts. In video data, a
minor change in a presented
image may be acceptable, even though the change might be noticeable if the
original and combined works
are presented and compared side by side.
Several techniques are known for hiding auxiliary information in original
digital data. Data can
be hidden in original data as headers or trailers appended to the original
data. Such techniques are of
limited use in protection of copyrighted works, because the auxiliary data is
easily located and stripped
out of the copy, as when changing format. More sophisticated techniques
distribute the auxiliary data
through the original data, entwining the auxiliary and original data until the
auxiliary data is difficult, or
even statistically impossible, to identify and strip from the combined data.
Most data hiding techniques that distribute the auxiliary data through the
original data are
computationally intense and therefore expensive to implement. Many of these
techniques are based upon
adding or subtracting periods of pseudo-random noise (PN) sequences with the
signal to represent the
auxiliary information, and these sequences may require filtering (a.k.a.
shaping) in the frequency domain.
The rest are based upon adding the auxiliary information to the original data
after the original data has
been transformed into the frequency domain, such as by a Fourier transform.
Auxiliary information can
be added in the frequency domain so that the energy of the auxiliary data is
spread across many
frequencies in a manner similar to that of the PN sequence. In addition,
auxiliary information can be
CA 02364433 2001-08-08
WO 00/54453 PCT/US00/06296
3
added to the phase of the frequency components with and without spreading the
information across
frequencies. Unfortunately, transforming the data into the frequency domain
and/or shaping the energy of
the PN sequence so it is less perceivable requires intense calculations.
The ability of users to detect auxiliary data depends not only upon the data,
but also upon the
characteristics of the human sense organs and the interpretation of sensory
stimuli by the brain. Some
data hiding techniques transform the original data into the frequency domain
and embed auxiliary data in
a manner such that the frequency spectrum of the original data reduces the
perception of embedded data.
This psychophysical effect is known as masking. The frequency distribution of
the original data is used
to determine preferred frequencies at which the embedded auxiliary data will
be less perceptible, that is,
masked. Others use the fact that we don't perceive phase as accurately as
magnitude in the frequency
domain.
There are some data embedding techniques that are less computationally intense
and that still
distribute the auxiliary data in the original data. Such techniques include
amplitude modulation,
frequency band elimination, distinct quantization, and least-significant bit
(LSB) replacement. These
techniques embed data in predetermined locations without regard to the
original data and are, therefore,
more likely to produce perceptual side affects in the combined data. In
addition, the LSB replacement
technique is easily disturbed by low level noise.
The ease of retrieving embedded data varies with the technique used for
embedding. Some data
hiding and retrieving techniques retrieve the auxiliary data by comparing the
combined data with the
original data. Others retrieve the auxiliary information using databases of
the PN sequences that were
originally used to hide the data. Techniques that require that a copy of the
original data or a PN database
be used to extract the auxiliary data are of limited use in applications in
which the combined data is
distributed broadly. Such techniques are useful in some applications, such as
data authentication, in
which the auxiliary data is retrieved rarely and only by the copyright owner.
Embedded data techniques are susceptible to removal of auxiliary information
for either of the
following reasons. First, the very nature of embedded data is incompatible
with bit-rate reducing (a.k.a.
compression) schemes, which remove the non-perceivable aspects of the data
such as done with MPEG
compression. Since a key feature of any embedded data is the fact that it is
non-perceivable, compression
schemes will act to remove the embedded data. Even if the embedded data is
designed to survive the
current compression technology, the next generation technology will probably
remove it. Bit-rate
compression schemes are very important in the digital distribution of media,
and receiving much research.
Second, noise reduction techniques will be able to remove embedded data. Noise
reduction techniques
are a hot topic, and used to restore old recordings. Since most non-
perceivable embedded data is similar
to noise, it will be removed by these noise reduction techniques. Again, even
if the embedded data is
designed to survive the current restoration technology, the next generation
technology may probably
remove it.
In another aspect, the technology detailed herein relates to methods and
systems for making
embedded data more robust against corruption and attack.
CA 02364433 2001-08-08
WO 00/54453 PCT/US00/06296
4
Embedded data may be susceptible to various types of corruption and attack.
For example, the
very nature of embedded data technology is at odds with bit-rate reducing
(a.k.a. compression) schemes,
which remove the non-perceivable aspects of the data such as done with MPEG
compression. Since a
feature of most embedded data is that it is non-perceivable, compression
schemes will tend to remove the
embedded data. Even if the embedded data is designed to survive the current
compression technology,
the next generation technology may result in its removal. Bit-rate compression
schemes are very
important in the digital distribution of media, and receiving much research.
Likewise, noise reduction
techniques, e.g., as are used to restore old audio recordings, pose a threat
to embedded data. Since most
non-perceivable embedded data is similar to noise, it will be removed by these
noise reduction
techniques. Again, even if the embedded data is designed to survive the
current restoration technology,
the next generation technology will probably remove it.
In another aspect, the technology detailed herein relates to ID assignment and
binding. Content
providers may want to allow only the person who bought content to access (i.e.
play, copy or record) that
content. One way to do this is to provide content that contains an ID, and
lock the ID to the consumer,
the rendering device or the storage unit. However, these existing solutions of
how to use the ID produce
unreasonable burdens for consumers.
One existing solution, known as user-binding, requires a person to carry an ID-
card and/or
remember a personal identification number (PIN) to access the content, similar
to the way bank ATM
machines work. The consumer has accepted this solution in order to access
money in the bank, a
situation where security is an advantage to the consumer too. However, it is
doubtful that consumers will
accept this requirement to access content, for example, play audio on a car
stereo. In addition, when a
group of people are sharing content, such as music, the process of each person
having to scan a card
before listening to their music is obtrusive. Finally, this solution requires
data that links the ID to the
user, so PINs and/or ID-cards can be produced. This data means the user's
privacy has been
compromised.
Another existing solution restricts playing of the content to one device,
known as player-binding.
This solution means your friend's music will not play in your car stereo,
neither will your movie play at
his house. This solution is not only inconvenient to the consumer, but also
reduces the sale of content
since many people buy content after playing or viewing it with their friends.
A final solution links the content to the storage unit, known as media-
binding. The storage unit
includes but is not limited to a magnetic hard drive, optical disk or
electronic memory. This solution
becomes cumbersome when the content should be allowed to move between
different storage unit types.
For example, a user, Joe, may want to play his audio from his computer's hard
drive over his home stereo,
or have the audio in his car or on a jog as portable electronic memory.
However, with this media-binding
solution, this audio can only be played in one place. and to move it from
Joe's stereo to his car, he has to
remember to where it was "checked out", otherwise. piracy cannot be
controlled. Importantly, he can't
just listen to it from each place as desirable to the consumer.
CA 02364433 2001-08-08
WO 00/54453 PCT/US00/06296
Another aspect of the technology detailed herein involves the use of multiple
embedded data -
with different characteristics - to serve rights management functions.
Another aspect of the technology detailed herein relates to scrambling of
content to protect
same.
5 It is often desirable to degrade digital signals so as to restrict access.
For instance, pay-TV
broadcasts are degraded so those who haven't paid for the program cannot watch
it because the picture is
unclear, while those who have paid for the program see a clear picture because
their recovery apparatus
has been enabled. Most recently, as a result of the digital audio revolution,
it is desirable to restrict MP3
(a standard bit-rate compressed audio file format) access. It is also
desirable to produce inexpensive
portable MP3 players, which in turn require that recovery of the original
signal be simple.
There are numerous existing methods of degrading digital content, a.k.a.
scrambling. Some
methods require a key to de-scramble the content, whereas others do not. Most
scrambling or degrading
methods are based upon either adding an interference signal to the digital
content, or moving the bits
around. Other methods use encryption, but this is very computationally
intense.
It would be advantageous if the information about a movie on a scrambled
channel could be
displayed to a viewer without having to de-scramble the information.
More recently, as a result of the digital audio revolution, some people would
like to see MP3 (a
Motion Pictures Expert Group Layer III standard bit-rate reduced audio file
format) access restricted as it
is easy to make an exact digital copy of the content. This restriction can be
implemented via scrambling
techniques. However, it is desirable to retrieve information about the
scrambled song without de-
scrambling the information, as this would allow a user to learn about a song
before deciding whether or
not to play that song, thus improving speed of the system for the user. It
would also allow a user's player
to quickly read copyright information, which could then enable playing. It is
also desirable to produce
inexpensive portable MP3 players, which in turn require that recovery of the
original signal be simple.
The prior-art contains numerous scrambling and de-scrambling methods. However,
these
methods are not designed to leave the header information alone during the
scrambling and descrambling
process, thus they are unable to retrieve information about the scrambled
audio without de-scrambling all
of the information.
The following detailed description addresses various ofthe just-cited issues,
redressing some of
these problems and providing new functionality not heretofore contemplated.
Brief Description of the Drawings
FIG. I is a flowchart showing acts employed in an illustrative embedding
technique.
FIG. 2 is a block diagram showing an apparatus used to embed or retrieve data
using the method
of FIG. 1.
FIG. 3 is a flowchart showing acts employed in an illustrative decoding
technique.
FIG. 4 graphically displays operation of a first illustrative embodiment.
CA 02364433 2001-08-08
WO 00/54453 PCTIUSOO/06296
6
FIG. 5 is a flowchart showing the embedding of data in accordance with the
first illustrative
embodiment. The dashed lines show interaction with the auxiliary data.
FIG. 6 is a flowchart showing the decoding of data. The dashed lines show
interaction with the
auxiliary data.
FIG. 7 graphically displays operation of a second illustrative embodiment.
FIG. 8 is a flowchart showing the embedding of data in accordance with the
second illustrative
embodiment. The dashed lines show interaction with the auxiliary data.
FIG. 9 is a flowchart showing the decoding of data. The dashed lines show
interaction with the
auxiliary data.
FIG. 10 demonstrates aspects of an illustrative embodiment in conjunction with
digital
compression techniques.
FIG. I I A and B are two block diagrams showing an embedding and retrieving
apparatus in
accordance with an illustrative embodiment.
FIG. 12 shows an illustrative embodiment of the apparatus of FIG. 2 for
embedding data.
FIG. 13 is shows an illustrative embodiment of the apparatus of FIG. 2 for
retrieving the data.
FIG. 14 shows a block diagram for an enabling process referenced in the
discussion concerning
attack resistance.
FIG. 15 shows the block diagram for a registration process.
FIG. 16 demonstrates the way in which dynamic locking blocks the duplication
of the auxiliary
data.
FIG. 17 displays the input and output for an exclusive-or (XOR) function.
FIG. 18A displays an overview of a process of dynamic locking and embedding of
the auxiliary
data.
FIG. 18B displays an overview of a process of retrieving and dynamic unlocking
of the auxiliary
data.
FIG. 19A shows the modification step of dynamic locking for locally masked
embedded data.
FIG. 19B shows the modification step of dynamic locking for pulse width
modified (PWM)
embedded data.
FIG. 19C shows the modification step of dynamic locking for embodiments based
upon PN
sequences. (Auxiliary data is abbreviated as aux.)
FIG. 20A displays the pseudocode in the form of a flowchart for locking and
embedding the
auxiliary data using header blocks.
FIG. 20B displays the pseudocode in the form of a flowchart for retrieving and
unlocking the
auxiliary data using header blocks.
FIG. 21 shows the basic process behind the example utilizations. (The dotted
boxes are
optional. The dashed boxes group similar items. In addition, although three
key locations are shown,
usually only one key is used and its location depends upon the utilization
requirements. Finally, the
CA 02364433 2005-03-09
WO 00/54453 - PCT/US00/06296
7
abbreviation ID is used and many times refers to an identifier, but can also
refer to any auxiliary
information.)
FIG. 22 shows an apparatus that may be used for these robust data embedding
techniques.
FIG. 23 shows an embodiment of the apparatus of FIG. 22 for dynamic locking.
FIG. 24 is a block diagram showing an embodiment of the apparatus of FIG. 22
for dynamic
unlocking.
FIG. 25 is an overview of a process of automatic ID management.
FIG. 26 is the pseudo-code for implementing an exemplary automatic ID
management process.
FIG. 27 is an apparatus to implement automatic ID management.
FIG. 28 is a portable MP3 audio player containing the apparatus of Fig. 27.
FIG. 29 is an overview of a process employing two watermarks.
FIG. 30 displays the pseudocode for the embedding process of Fig. 29.
FIG. 31 displays the pseudocode for the retrieving process for Fig. 29.
FIG. 32 displays an apparatus that may be used in connection with the process
of Fig. 29.
FIG. 33a is an overview of a scrambling process, where dotted boxes are
optional.
FIG. 33b is an overview of a descrambling process, where dotted boxes are
optional.
FIG. 34a is the pseudo-code for an exemplary scrambling or descrambling
process.
FIG. 34b shows the input and output for the exclusive-or (XOR) function.
FIG. 35 shows an exemplary apparatus for performing the scrambling or de-
scrambling
processes.
FIG. 36 is an overview of a degradation and recovery process.
FIG. 37 is the pseudocode for the degradation and recovery process of FIG. 36.
FIG. 38 is a simple and efficient example of the degradation and recovery
process using a
threshold crossing and adjusting only the next point.
FIG. 39 is the pseudocode for the degradation and recovery process of FIG. 38.
FIG. 40 is an overview of an apparatus suitable for implementing the process
of FIGS. 36-39
Detailed Description
Introduction to Data Hiding Arrangements
The following discussion proceeds with reference to exemplary embodiments,
methods, and
operational features. Except as explicitly stated, or reasonably indicated,
the examples given should not
be taken as precluding other arrangements.
In accordance with one embodiment, a method and apparatus of data hiding and
retrieval is
provided that offers high efficiency, with an attendant reduction in cost. In
some embodiments,
psychophysic data hiding is used - without the need to modify or transform the
original data - in order to
identify the locations at which the data should be hidden. In certain
embodiments the encoding leads to
essentially no detectable change in the content statistics, making the hidden
signal still harder to identify
and remove.
CA 02364433 2001-08-08
WO 00/54453 PCT/US00/06296
8
The technology can be implemented so as to permit the user to set parameters
that vary the
perceptibility, robustness, and embedding rate, permitting the disclosed
technology to be used in a broad
variety of applications.
An exemplary apparatus includes a logic processor and storage unit, such as
those that come
with a standard personal computer or on DSP boards. These devices act as data
readers, comparer and
data writers, so that the user's desired watermark can be embedded and/or
retrieved.
An exeplary process involves embedding and retrieving auxiliary information
into original data
to produce combined data. One or more detection criteria can be used to
determine where in the original
data to locate and/or adjust data points so as to carry the auxiliary
information. The detection criteria can
be used to locate positions - referred to as local masking opportunities - in
the original data at which the
embedding of auxiliary data will produce less perception, as compared to other
simplistic processes.
When embedding the auxiliary data, the data points in the original data are
investigated in
accordance with the detection criteria to determine the existence of local
masking opportunities. The
detection criterion or criteria may involve, for example, comparing the data
point to a predetermined
value and examining the relationship of the data point to nearby points. If
the detection criteria are met,
one or more of the nearby points, or the data point being investigated, is
changed to indicate the value of
an embedded bit of auxiliary data.
Thus, although the search for local masking opportunities typically progresses
point by point
through the data, the investigation of each point may include not only the
value of that point, but also
values of one or more nearby points and/or one or more relationships among the
points. If the
investigation of a point shows the existence of a local masking opportunity,
data is embedded by setting
the value of one or more of the local points, i.e., either the point being
investigated or one or more of the
nearby points.
The value to which the nearby data points are set in the illustrative
embodiment is typically
dependent upon the data point being investigated, as well as on the value of
the auxiliary data bit. The
data point value can be set so that it has a specified relationship with the
neighboring data points. The
process is continued until the original data has been traversed or no
additional auxiliary data remains to
be embedded.
Retrieving the auxiliary data is the inverse of the embedding process. The
combined data is
traversed using the detection criteria to locate the local masking
opportunities. As each local masking
opportunity is located, the nearby data point or points that was/were set to
indicate the embedded bit
is/are read to extract the embedded data. The process is continued until the
combined data has been
traversed.
In the preferred embodiments, a data point or points are set to a value
relative to the nearby data
points, and not to an absolute value. Both setting data points at the local
masking opportunity and setting
the data point to a value related to the nearby points, rather than to a value
unrelated to the original data,
provide masking that reduces the perceptibility of the data. The data is
extracted by determining the
relationships or values of the point or points near the local masking
opportunity.
CA 02364433 2001-08-08
WO 00/54453 PCTIUSOO/06296
9
For the two preferred embodiments described in detail below, only points with
large values are
adjusted. and by a minimal amount; thus, these embodiments are based upon the
masking of a weak
stimulus by an intense stimulus. The process is applicable to analog and
digital data. However, both
embodiments are explained in terms of digital media due to current switch to
digital media and the ease
of understanding.
Specifically, the first preferred embodiment uses the difference between a
data point after a peak
and the peak level to carry auxiliary information, as long as the peak is
above a large threshold and the
original difference between the peak and next point is not too great. This
large threshold and minimal
differences produce the desired perceptual masking. The embedding process
adjusts the point after the
above-threshold peaks to hide the auxiliary data. Correspondingly, the
retrieving process measures the
difference between each above threshold peak level and the next data point to
retrieve the auxiliary data.
The second preferred embodiment uses the change in slope across a positive,
large, steep,
threshold crossing to hide the auxiliary information, as long as the original
change in slope is not too
great yet steep enough to accept the ensuing adjustment. Again, the large
threshold produces the desired
perceptual masking. In the implementation, the embedding process adjusts the
change in slope to embed
the data, whereas the retrieving process measures the change in slope to
obtain the auxiliary data.
Usually, the preferred embedding process spectrally implicitly spreads the
energy of the
auxiliary information throughout the original data. This broadband approach
produces data that is more
difficult to remove than sub-band approaches that place the data in an
inaudible frequency range. If
desired, parameters can be chosen so that the process produces protected data
that is statistically identical
to unmarked data. Importantly, the process can be adjusted to produce the
desired tradeoffs between
perception, coding rate and robustness to attack.
Such embodiments preferably - although not essentially - operate on the
original data without
requiring any complex data transformations, such as a Fourier transformation.
Thus, if the original data
represents information in the time domain, the data can remain in the time
domain as the auxiliary data is
embedded and retrieved. Of course, the technology can operate on original data
of all types, such as in
the frequency or time-frequency domain. For example, it can be applied to MPEG
data, including the
MPEG I and 2 specification, ISO 11 172-3 and ISO 13818-7 respectively, which
exists in the time-
frequency domain.
Finally, the problem of bit-rate reducing techniques, known as compression,
removing the
watermark can be bypassed by using separate, but possibly identical, watermark
procedures during the
compression (a.k.a. encoding) and decompression (a.k.a. decoding) process.
Exemplary Embodiments and Methods
A system according to one embodiment comprises a method and apparatus for
hiding auxiliary
information (or data) in original data and for retrieving the auxiliary
information.
FIG. I is an overview of the steps involved in carrying out an illustrative
method to embed data.
FIG. 2 shows a block diagram of an apparatus 10 that may be used to perform
the method of FIG. 1.
CA 02364433 2001-08-08
WO 00/54453 PCTIUSOO/06296
Apparatus 10 includes a logic processor 14, which can be a general purpose
microprocessor, such as an
Intel Pentium or DEC Alpha, of the type a personal computer or engineering
workstation, a digital signal
processor (DSP), such the the Texas Instruments TMS320 line, a specialized
CPU, such as a media
processor, or a custom processing circuit. Apparatus 10 also includes a
storage unit 18, which can
5 include random access memory (RAM) or delays. Because the algorithms used in
the illustrative
embodiment are not computationally intense, they require calculations on the
order of less than one
million instructions per second and can be performed by most modern personal
computers, and one many
less capable devices as well (e.g., personal digital assistants, dedicated
media players, etc.).
The original data mentioned below may represent sound that is recorded by
sampling its
10 amplitude periodically, with each sample using binary numbers to represent
the magnitude of the sound at
a particular time. Likewise, the samples may represent pixels of an image or
video. Still further, the
original data can be any series of binary data associated into groups.
Similarly, the illustrative auxiliary
information is any data that can be represented as "I "s and "0"s, but other
symbol alphabets can likewise
be used with corresponding adaptation of the disclosed arrangements.
FIG. I shows that in step 20, a portion of the original data is read into
storage unit 18 of FIG. 2.
Step 24 shows that the sample data is investigated sequentially by the logic
processor 14 to locate sample
points that meet predefined detection criteria. Such sample points indicate
the existence of "local
masking opportunities," because the detection criteria are such that a change
in the value of the sample or
a few samples at or near that point to embed auxiliary data will usually have
minimal perceivable by the
listener of the sound. The amount of masking will depend upon the data type
and settings chosen by the
user. For example, the masking will be great for uncompressed audio and less
for bit-rate reduced
(digitally compressed) audio such as MPEG. The same detection criteria will be
applied during data
retrieval to locate the hidden data.
Each point in the original data is preferably investigated to determine
whether it represents a
local masking opportunity. The criterion or criteria for determining local
masking opportunities may
entail not only the value of the point being investigated, but may also
include the value of at least one
nearby or neighboring point, or the relationship between the nearby point and
the point being
investigated. The detection criteria can require, for example, that the point
being investigated exceeds a
certain threshold value and/or that the point be a local maximum or peak,
and/or that the point is a point
of local maximum in a first- or higher-order derivative. The criteria may
include a requirement that a
point subsequent to the point being investigated have a value that differs
from the point being
investigated by less that a prescribed amount, or have some other relationship
to the point being
investigated.
The sample data points can be considered as plotted on a graph, for example
with time on the x-
axis and the magnitude of the sample on the y-axis. Thus, the series of data
points can be considered as
having a slope between any points, and the value of the slope can be part of
the detection criteria. The
criteria may specify, for example, that a slope defined by the point being
investigated and a preceding
point exceed a particular value, or that the change in slope before and after
the point not exceeds a
CA 02364433 2001-08-08
WO 00/54453 PCT/US00/06296
11
particular value. The criteria could include any combination of requirements;
the detailed examples are
not essential or restrictive of the scope of the detailed technology.
In the illustrated cases, no complex data transformation is required to mask
the auxiliary data, so
comparing a point to the detection criteria is relatively quick and
inexpensive. Unlike many prior art
methods, which use distant points to convert the original data into the
frequency domain to determine
how to mask embedded data, the illustrative embodiment can determine masking
opportunities using only
nearby or neighboring points, i.e., points that are too close to use to
determine useful frequency data.
Nearby points include points that are next to the point being investigated or
within a relatively small
number of points, preferably less than 50 and more preferably less than 20.
The criterion can be as
simple as determining whether the point exceeds a threshold.
Step 26 shows that when a point meeting the detection criteria is located, the
value of a specified
sample point or sample points near the local masking opportunity is changed to
reflect the value of the
auxiliary information to be embedded. Although the changed sample may be
simply set to a particular
value to signify the value of the embedded bit, the new value typically
depends upon the value of both the
auxiliary data and the neighboring point or points that were investigated to
detect the local masking
opportunity. For example, the point may be set so that the change in value or
slope signifies whether the
embedded bit is a "I" or a "0" (or other symbol).
When a point is set to its new value, it is preferable that either the change
does not prevent the
original sample point from continuing to meet the detection criteria, or that
this local masking opportunity
is skipped and not detected in the retrieval process. Otherwise, the embedded
auxiliary data may not be
retrievable.
Alternatively, it is possible to merely embed the auxiliary bit as the least
significant bit, or other,
preferably low order, bit. The embedded bit is still masked because the
location of the embedded bit was
chosen to represent a local masking opportunity, such as when the data is
larger than a prescribed
threshold.
Step 30 shows that the process is ended at step 32 if no additional auxiliary
data needs to be
embedded. Otherwise, step 34 shows that if there is additional data in memory,
the search for local
masking opportunities continues. Step 36 shows that if all data in memory has
not yet been searched,
additional data is read into memory. Skilled persons will recognize that some
overlap of the data in
memory may be required to prevent missing local masking opportunities that
occur at the beginning or
end points of the data in memory.
FIG. 3 broadly shows the steps involved in carrying out a decoding method.
Because the same
processor and memory that were used to embed the data can be used to retrieve
the data, although not
necessary, the steps of FIG. 3 will describe extracting data using the
hardware components of FIG. 2.
Step 50 shows that a portion of the original data is read into storage unit
18. Step 52 shows that logic
processor 14 investigates each data point to determine the existence of a
local masking opportunity. If a
sample point meets the local masking opportunity criteria, step 54 shows that
the embedded "I " or "0" bit
of auxiliary data is extracted using the inverse relationship of how the
auxiliary data was embedded. Step
CA 02364433 2001-08-08
WO 00/54453 PCTIUSOO/06296
12
56 shows that if additional combined data is in the memory, the logic
processor continues to investigate
the remaining points with step 52. Step 58 shows that if all the data in
memory has been investigated, but
there is uninvestigated combined data in the data file, additional data is
read into memory in step 50.
Step 60 shows that the process is ended when all the combined data has been
investigated.
Two particular embodiments are described briefly here, and in detail below, to
demonstrate the
flavor of this methodology. As shown in FIG. 4, the first embodiment uses
large, positive peaks as the
detection criteria 120 and the auxiliary information is stored in the
difference 130 between the peak and
the next point. As shown in FIG. 7, the second embodiment uses large, steep
threshold crossings with
minimal change in slope as the detection criteria 140, and the auxiliary
information 150 is carried in the
change in slope.
The methodology is applicable to analog or digital data, even though the
preferred embodiments
use digital data. For example, analog data can be sampled at the Nyquist rate
to produce digital data in
which additional information is hidden. Then, the combined digital data can be
returned to the analog
domain by any existing method known in digital signal processing (DSP). The
analog data now contains
the embedded data, which can be decoded by using sampling. This is just one
possible method to encode
analog data with the above methodology.
The methodology is also applicable to audio, speech, images, video or any
other perceivable
signal. With audio and speech, the original data may represent pressure versus
time, magnitude versus
frequency, or a specific frequency magnitude versus time. With images, the
original data may represent
gray code versus space, separate or combined RGB or equivalent values versus
space, or magnitude
versus frequency. Video data encompasses the image data with an added
dimension of time available.
For example, with MPEG bit-reduced audio or images the auxiliary data may be
embedded in scaling
factors or frequency coefficients versus frequency or time or both.
Usually one of the detection criteria is a large threshold. With 16 bit audio,
a threshold greater
than 48 dB above the minimum value is desirable. This threshold allows the
data to be changed with
minimal perception due to masking. Masking is the psychological term defined
as the increase in
threshold for steady-state stimuli. Use of the term in this disclosure is much
broader than that definition,
and describes how one set of data reduces the perception of other data.
Specifically, for uncompressed,
magnitude-time data, the sensitivity of the sensory system decreases with
increased input level, thus the
small adjustment of a neighboring data point is masked by the large value of
the threshold. For bit-rate
reduced, time-frequency data, such as MPEG data, the masking is minimal and
more similar to the
textbook definition since masking has been used to reduce the bit rate.
Finally, this method is applicable to data where masking is not used, but the
efficiency of the
process in that it is does not require a key, such as a PN sequence, or
original data for retrieval of the
auxiliary information, is an advantage. In summary, the parameters of the
detection criteria will
determine the interaction between the data rate, process complexity and
perceptual quality.
CA 02364433 2001-08-08
WO 00/54453 PCTIUSOO/06296
13
Embodiment I
The first particular embodiment is based upon hiding the auxiliary information
in large peaks
within the original data. In this embodiment, the auxiliary information is
preferably broken into N bit
words, with synchronization data placed between the words for better error
recovery. The auxiliary
information does not need to include sync pulses between the words if
robustness to noise or modified
files is not needed.
FIG. 4 conceptually shows that the first embodiment detects a peak or local
maximum and sets
the value of the subsequent point in relation to the peak to indicate the
value of the embedded bit.
FIG. 5 includes the pseudocode in the form of a flowchart for the embedding
process. The
process begins by searching the original data until a positive peak that lies
above a large threshold,
labeled thr, and has a relatively small decrease after the peak, labeled dS,
is found. This process is
demonstrated in boxes 200, 210 and 220. The detection criteria are checked in
the most computationally
efficient order, which includes first checking to see if the point represents
a peak since peaks are the least
likely criterion.
When a desirable peak is found, the data point after the peak is adjusted
according to a user
defined bit depth, b, to carry the auxiliary information. Specifically, if it
is the beginning of an auxiliary
word, the synchronization code is embedded by adjusting the point after the
peak, x[n+1 ], to be equal to
the peak, x[n], minus half of the maximum allowable change, dS/2, between the
peak and the next point,
as shown in boxes 242, 230 and 250. An auxiliary information bit of one is
encoded by adjusting the
point after the peak, x[n+1 ], to be equal to the peak, x[n], minus half the
maximum change, dS/2, and
plus the half the bit depth magnitude, 2b-'. Correspondingly, an auxiliary
information bit of zero is
encoded by adjusting the point after the peak, x[n+l ], to be equal to the
peak, x[n], minus the sum of half
the maximum change, dS/2, and half the bit depth magnitude, 2b-'. This
embedding of zeros and ones is
shown in boxes 242, 240, 260, 270 and 280. The next two points after embedding
the data should be
skipped so one does not create another peak for very slow changing (i.e. flat)
data, as shown in box 290.
These steps are repeated until the auxiliary information, box 242 and 240, has
been hidden in the
original data or the original data is finished.
FIG. 6 displays the pseudocode in the form of a flowchart for the retrieval
process of the first
particular embodiment. The process begins by searching the original data until
a positive peak that lies
above a large threshold, labeled thr, and has a relatively small decrease
after the peak, labeled dS, is
found. This process is demonstrated in boxes 300, 310 and 320. Again, the
search first looks for a peak
to improve efficiency.
When a desirable peak is found, the difference between the peak and the data
point after the
peak is measured to retrieve the auxiliary information. Specifically, if the
peak minus the point after the
peak, x[n]-x[n+1 ], is close to half of the maximum allowable change, dS/2, a
new auxiliary word is
beginning, as shown in boxes 330 and 350. If the peak minus the point after
the peak, x[n]-x[n+]], is
approximately equal to half the maximum change, dS/2, minus half the bit depth
magnitude, 2b-', an
auxiliary bit of one is found. If this difference, x[n]-x[n+l ], is close to
the sum of half the maximum
CA 02364433 2001-08-08
WO 00/54453 PCT/US00/06296
14
change, dS/2, and half the bit depth magnitude, 2b-1, an auxiliary bit of zero
is retrieved. This retrieving
of zeros and ones is shown in boxes 340, 360, 370, 380, and 382. The two
points immediately after
retrieving the data can be skipped as shown in box 390.
These steps are repeated until the auxiliary information has been retrieved in
the original data or
the original data is finished.
There are three user-defined parameters, including threshold, thr; bit depth,
b; and maximum
allowable change after the slope, dS. For 16 bit audio, the threshold is
usually around 48 dB above the
minimal quantization, as discussed above. For data with more bits per sample,
the threshold may be
increased to reduce perception. The bit depth is an indication of the relative
change to be made to the
sample point to embed the data. Thus, the smaller the bit depth, the less
disturbance of the original data,
making the embedded data less perceptible to the listener, but less robust,
that is, more susceptible to
being lost to noise or attack. Minimal perception in 16 bit audio is found
when bit depths are between I
and 6 bits. However, higher bit depths can be used if one desires more
robustness to noise in trade for
more perceptual degradation. The maximum allowable change after the peak, dS,
must be at least the
desired bit depth magnitude, 2b. On the one hand, one can gain better
robustness to noise at the expense
of more distortion, if dS is set to twice the bit depth magnitude, 2b+1. On
the other hand, if one desires to
keep the threshold undetectable to statistical cryptoanalysis (labeled
statistically invisible), dS should be
set at 2b, and b should be small, probably below 3 bits. If dS is not 2b, one
can use the discrepancy of the
average difference between large positive peaks and their next points between
embedded file and regular
file data to determine if the file contains embedded data or not. Finally, if
dS is much greater than 2b, the
auxiliary information embedding rate will be increased, because more peaks
will be found suitable for
data embedding. Using the principles explained above, skilled persons will be
able to set the user-
defined parameters to values appropriate to the requirements of a particular
application.
As discussed above, the large threshold usually reduces the perceivable effect
of adding the
auxiliary information, and may even cause the auxiliary data to be non-
perceivable, depending upon the
data type. In addition, many data points satisfy the small difference between
the peak and data point after
the peak, because with a slope near 0 at the peak, the data is changing the
least. This small difference
means that the adjustment will be small as compared to the threshold, thus
reducing the chance of
perceiving the embedded auxiliary data
The pseudocode is shown using a buffer with what appears to be look ahead
capabilities (i.e,
x[n+1 ]). This makes the process easier to explain and understand. However,
the process is causal, as
determined by replacing n+1 with k, and keeping track of the last two points,
x[k-1 ] and x[k-2].
Finally, one can add more criteria to define the peak. For example, the peak
extends for one
more point each direction where x[n]>x[n-2], x[n]>x[n+2], x[n]>x[n-3],
x[n]>x[n+3], and so on, or the
peak is of minimal sharpness, i.e. x[n]-x[n-I ]>5. Both of these criteria
produce better robustness to noise
and less distortion since it will take more noise to move the location of the
peak, although changes in the
peak criteria affect the rate at which auxiliary data can be embedded.
CA 02364433 2001-08-08
WO 00/54453 PCTIUSOO/06296
The embedded data density and bit rate will vary with the original data and
with the user-defined
parameters. For example, bit rates of between 99 and 268 bits per second were
achieved in CD quality
audio data using a bit depth of 5 and a threshold of 5,000 (74 dB). Using a
bit depth of 8 and maintaining
a threshold at 5,000, the average embedding rate was 1,000 bits per second.
When the threshold is lowed
5 to 2,000 at a bit depth of 8, an average embedding rate of 2,000 bits per
second was found.
Embodiment 2
The second particular embodiment hides the auxiliary information in large,
steep threshold
crossings which do not have a large change in slope. The method is more robust
to noise changing the
10 detected location. This occurs because it is less likely that noise changes
the location of a threshold
crossing as compared to a peak, since a threshold crossing usually has a slope
larger than the slope at the
peak, which, by definition, has a slope near zero. Testing with audio data has
shown this embodiment, as
compared to the first embodiment, to produce a lower embedded data rate and is
more perceivable at a
lower bit depth, in trade for the robustness to noise. One will probably find
the optimal embodiment
15 dependent upon the application.
FIG. 7 conceptually shows that data is embedded by setting the slope after the
threshold crossing
in relation to the slope at the threshold crossing.
In FIG. 8, the pseudocode for hiding the auxiliary information using the
second preferred
embodiment is presented in the form of a flow chart. The process begins by
searching the original data
until a positive, large, steep threshold (labeled thr) crossing with minimal
change in slope (labeled dS) is
found. This process is demonstrated in boxes 400, 410 and 420.
When the desirable threshold crossing is found, the data point after the
threshold crossing is
adjusted according to a user defined bit depth (b) to carry the auxiliary
information in the change in
slope. Note that the change in slope is defined as (x[n+1 ]-x[n])-(x[n]-x[n-
1]), or equivalently as x[n+]]-
2*x[n]+x[n-1]. Specifically, if it is the beginning of an auxiliary word, the
synchronization code is
embedded by adjusting the point after the threshold crossing, x[n+l ], so that
the change in slope is zero,
as shown in boxes 442, 430 and 450. An auxiliary bit of one is encoded by
adjusting the point after the
threshold crossing, x[n+l], so that the change in slope is positive by an
amount equal to half the bit depth
magnitude, 2b-l. Correspondingly, an auxiliary bit of zero is encoded by
adjusting the point after the
threshold crossing so that the change in slope is negative by an amount equal
to half the bit depth
magnitude, 2b-1. This embedding of zeros and ones is shown in boxes 442, 440,
460, 470 and 480. The
point after embedding the data can be skipped for efficiency as shown in box
490.
These steps are repeated until the auxiliary information has been hidden in
the original data or
the original data is finished.
FIG. 9 shows the pseudocode in the form of a flowchart for the retrieval of
the auxiliary
information in the second preferred embodiment. The process begins by
searching the original data until
a positive, large, steep threshold (labeled thr) crossing with minimal change
in slope (labeled dS), is
found. This process is demonstrated in boxes 500, 510 and 520.
CA 02364433 2001-08-08
WO 00/54453 PCTIUSOO/06296
16
When a desirable threshold crossing is found, the change in slope around the
threshold is
measured to retrieve the auxiliary information. Again, the change in slope is
defined as (x[n+l]-x[n])-
(x[n]-x[n-1 ]), or equivalently as x[n+]]-2*x[n]+x[n-1 ]. Specifically, if the
threshold crossing has almost
zero change in slope, a new auxiliary word is begun, as shown in boxes 530 and
550. If the threshold
crossing has a positive change in slope approximately equal to half the bit
depth magnitude, 2b-', an
auxiliary bit of one is found. If the threshold crossing has a negative change
in slope approximately equal
to half the bit depth magnitude, 2b-', an auxiliary bit of zero is retrieved.
This retrieving of zeros and ones
is shown in boxes 540, 560, 570, 580, and 582. The point after retrieving the
data can be skipped for
efficiency as shown in box 590.
These steps are repeated until the auxiliary information has been retrieved in
the original data or
the original data is finished.
As mentioned above, one does not want the embedding process to eliminate the
embedded
location from fulfilling the detection criteria. Specifically, in this
embodiment, the pre-threshold change
condition, x[n]-x[n-I ]>dS+2b"', in the detection criteria of box 420 and 520
requires that the adjustment
of the next data point does not bring the point back below the threshold. An
alternative approach, is to
ignore this condition and to set either the current or next point (x[n] or
x[n+1 ], respectively) to the
threshold if the embedding process would cause the next point to move below
the threshold, and ignore
any data points that are equal to the threshold in both the embedding and
retrieving process.
Interestingly, only when embedding a sync or 0 could the next point move below
the threshold. Given
these options, the described embodiment is chosen so the process is causal,
thus incorporating the known
advantages of causal processes.
Once again, the large threshold and maximum allowable change in slope
condition, dS, reduce
the perception of embedding the auxiliary data, and depending upon the data
type can cause the
embedding process to be completely non-perceivable. The maximum allowable
change in slope
condition, dS, can have any value. A larger value allows a higher data rate
with more perceivable
distortion, whereas a smaller value produces minimal distortion with a lower
data rate. Our preferred
setting for dS in 16 bit audio is equal to the bit depth magnitude, 2b. Again,
bit depths below 6 bits
produce minimal distortion, but higher bit depths can be used for robustness
to noise and attack.
Using a threshold of 2,000 (i.e. 66 dB) and a bit depth of 5, data rates
between 40-100 bits per
second are expected, with an average of about 75 bits per second, for CD
quality audio. At a bit depth of
8, the bit rate increases to an average of 100 bits per second.
Modifications
Particular embodiments have been described in detail above. However, there are
many simple
modifications that can be made to optimize the process for each use.
In some applications, a very simple embodiment could use a simple threshold to
determine a
local masking opportunity and then encode the auxiliary data in the LSB of the
point exceeding the
threshold or of another point in the vicinity of the point exceeding the
threshold. Such a variation is
CA 02364433 2001-08-08
WO 00/54453 PCTIUSOO/06296
17
extremely simple, yet provides reduced perceptibility compared to prior art
LSB schemes. As with the
other embodiments, one must ensure that changing the value does not remove the
point for the detection
criterion. In this case, one could simply skip embedding where the change
brings the data below the
threshold, and change the current value of the data point to the threshold so
that the data point will be
skipped in the retrieving phase.
To increase the robustness to attack or noise, the following changes could be
made. (Attack is
defined as a person or machine trying to remove the auxiliary information from
the combined signal
without distorting the perception of the original data.)
Using a dynamic threshold can make it harder to remove the auxiliary
information. An example
dynamic threshold is an offset sinusoidal waveform. When using a dynamic
threshold, dS should be
small and close to 2b so that the process does not change the distribution of
the differences between
neighboring points, i.e. be statistically invisible; thus, an attacker cannot
use this data to find the
threshold.
One can also use the statistical gaps when dS is larger than 2b to find the
threshold if the attack
uses a DC shift. A DC shift is obviously a more potent attack for the second
preferred embodiment than
the first, but could affect the first preferred embodiment since threshold is
one of the detection criteria.
(Attack-resistant methods are further considered below.)
The process may use more global definitions for peaks and threshold crossings,
for better
robustness to noise. Specifically, a peak or threshold crossing definition may
be used that includes more
points on each side.
Finally, the process can use any type of error correction in the auxiliary
information to increase
the robustness.
To increase the data rate, the following changes may be made. The auxiliary
information does
not need to include the extra sync pulses between the N-bit words, especially
if robustness to noise is not
needed. In addition, negative going peaks and/or more thresholds can be used
to increase bit rate.
Finally, the process can use more than a binary system in adjusting the second
bit to encode more
information. However, the result is more likely to be perceivable or less
robust to attack.
An interesting twist is to embed different auxiliary information on positive
and negative peaks,
and/or on various thresholds. In addition, with stereo files, the channels can
be coded separately, or
encoding can move between channels with consecutive points moving between left
and right channels.
A change that can improve the perception is to move the data point after the
embedded point
towards the value of the embedded point if combining the auxiliary information
causes a large value
change in the embedded point.
As mentioned above, the data does not have to be relative to time. For
example, the data may
represent magnitude versus frequency. In addition, the data could be viewed as
magnitude of a specific
frequency versus time. All frequencies may be included for an increased data
rate. In other words,
embedding may be performed in the spectrum or spectrogram. Advantageously, one
doesn't have to
change the format of the given data to use this process and corresponding
apparatus.
CA 02364433 2001-08-08
WO 00/54453 PCT/US00/06296
18
Consider, for example, bit-reduced data, such as MPEG compressed data. MPEG-
compressed
data comprises a series of data points that represent scaling factors and
frequency coefficients. Auxiliary
data may be embedded in the series of MPEG data points, using, for example,
one of the two particular
embodiments described above. When using the first particular embodiment, one
may want to increase the
peak or modify its LSB such that the term is only increased, rather than
decrease the point after the peak,
such that quantization error is not increased in the MPEG data, especially
when dealing with scaling
factors. Skilled persons will recognize that, in using data like MPEG data
that is divided into time
frames, one may use, for example, scaling factors or frequency coefficients
from consecutive frames, as
well as data points representing scaling factors and coefficients of different
frequencies within a frame,
when determining where to embed data. For example, the coefficients for a
particular frequency in
consecutive frames could be considered as a series of consecutive data points,
and those data points
analyzed in accordance one of the embodiments above to determine where to
embed data in the series. In
an alternative example, the series of data points representing scaling factors
or frequency coefficients for
different frequencies within a frame may be analyzed in accordance with the
first or second particular
embodiments above to decide where to embed data.
Example Utilizations
Below are included some example utilizations of an illustrative algorithm to
aid in its
understanding. This list is not complete, and only highlights the usefulness
of the disclosed technology.
Moreover, the applications given below are not reliant on the particular form
of encoding used - any
other form of digital watermarking, steganography, or data hiding, can
alternatively be used.
The process can be used to embed copyright information. This information may
include a code
to determine if the data can be copied. Copying devices, such as CD writers,
can include an inexpensive
integrated circuit that could interpret embedded data and prohibit copying.
In addition, author's or artist's name and affiliation can be embedded. In
this utilization, the
auxiliary information is small and would be repeated over and over with
synchronization pulses between
each duplication. Alternatively, the copy code could be embedded using
embodiment 1, and the creator's
name and affiliation using embodiment 2 (i.e., several embedded data may co-
exist in a work).
Such technology can be used to send additional information. This information
may be
transmitted in ASCII or ANSI with 8 bit "words" (not to be included with
digital words being defined as
32 bits) and synchronization pulses between these words, if desired. The
information may be a secret
message, lyrics to the song, or a description of the artwork. For lyrics, this
could be useful for karaoke
machines and CD or DVD players.
Digital Compression
The main problem with hiding data and digital compression (reducing bit rate
not dynamic
range) is that the process of hiding data is at odds with digital bit-rate
reducing techniques known as
compression (a.k.a. encoding and decoding). This incompatibility occurs since
the goal of data hiding is
CA 02364433 2001-08-08
WO 00/54453 PCT/US00/06296
19
to make the data minimally perceivable and the goal of compression is to
remove minimally perceivable
parts.
To this end, FIGS. 10A and B demonstrates an illustrative process for data
hiding, if at some
point the data must be compressed. For example, this may happen while
transmitting the data.
In FIG. I OA, the auxiliary information is embedded in the non-compressed data
using the above-
detailed process or any other method, as shown in box 600. Then, when the data
needs to be compressed,
the auxiliary information is retrieved via the above-detailed or the other
appropriate method, and re-
embedded in the compressed data with the above-detailed process or any other
method, as shown in box
610. The algorithm for data hiding in the compressed and non-compressed data
may be the same
algorithm, differing by only using different original data. Or they may be
different.
In FIG. 1013, the auxiliary information is retrieved from the compressed data
by the above-
detailed or other method, the data is uncompressed, and the auxiliary
information is embedded in the
uncompressed data, as shown in box 620. Finally, when needed, the auxiliary
information can be
retrieved from the data using the above-detailed or other method, as shown in
630. Once again, the
algorithm for data hiding in the compressed and non-compressed data may be the
same algorithm,
differing by only using different original data, or not.
Apparatus
As described above, FIG 2 demonstrates that the detailed process can be
implemented via logic
processor and storage unit 18. FIG. 12 shows the implementation with a digital
processor 1200 and
digital memory 1210. The digital processor 1200 may be defined as the
equivalent of a digital signal
processor (DSP), general-purpose central processing unit (CPU), or a
specialized CPU, including media
processors. A likely DSP chip is one of the Texas Instruments TMS320 product
line. A CPU could
include one of Intel's Pentium line or Motorola/IBM's PowerPC product line.
The design is
straightforward for someone familiar with the state of the art given the
pseudocode in Figs 5 through 9.
In addition, as shown in FIG. 13, a person familiar with the state of the art
could implement the
process with analog and digital circuitry, either separate or in an
application specific integrated circuit
(ASIC). The analog and digital circuitry could include any combination of the
following devices: a
digital-to-analog converter (D/A), comparators, sample-and-hold circuits,
delay elements, analog-to-
digital converter (A/D), and programmable logic controllers (PLC).
Programmable logic arrays (PLDs)
can likewise be used. Someone familiar with the state of the art given the
previous description and
pseudocode in FIGs. 5 through 9 could easily design the circuit.
FIGs. I1A and B show that the logic processor and storage unit typically
comprise an
embedding apparatus 700 and retrieving apparatus 770. The embedding apparatus
700 includes the
following. A data reader 710 to read original data 720 and auxiliary data 730.
A comparer 740, that is, a
circuit or device for comparing data points with known values or other data
points. A data writer 750 to
write the combined data 760 to a permanent or temporary storage media.
CA 02364433 2001-08-08
WO 00/54453 PCT/USOO/06296
The retrieving apparatus 770 includes the following. A data reader 715 to read
the combined
data. The data reader 715 maybe identical to the embedding data reader 710,
but it also may be
different. A comparer 745, that is, a circuit or device for comparing data
points with known values or
other data points and, if necessary, producing the auxiliary bit or bits. Once
again the comparer 745 may
5 be identical or different that the embedding comparer 740. A data writer is
not always necessary since
the auxiliary information may be taken from memory or only displayed for the
corresponding use.
Ramifications and Scope
As the reader can see from the description above, and can demonstrate by
testing the process
10 with CD quality audio, the above-detailed processes and apparatuses for
hiding auxiliary information
within original data are efficient and have configurations which are non-
perceivable. These advantages
are mainly due to finding locations to hide the auxiliary data without needing
to transform the signal to
the frequency domain, so that masking may block or reduce the perception of
the auxiliary data.
15 Corruption- and Attack-Resistance
As noted, a further aspect of the technology detailed herein relates to
methods for increasing the
resistance of embedded data to corruption and attack. For expository
convenience, the word "attack" is
used herein, but is meant to include both deliberate efforts to remove
embedded data, and incidental
removal of such data. Attack may include duplication, which is defined as
being able to replicate or
20 impersonate the embedded data from one data segment to another. Attack may
also include modification,
which is defined as changing the embedded data for a desired affect, such as
from "no copying" to
"copying allowed".
The below-detailed technology sets forth two embodiments describing ways of
using embedded
data such that the embedded data is more robust to attack: the enabling and
registration process. In
addition, embodiments improving the robustness of embedded data to duplication
or modification are
disclosed, including dynamic locking and unlocking.
The first embodiment utilizes an enabling process, which involves using
embedded data to
enable an action, such as copying, playing or otherwise rendering. Thus, if
the embedded data is
removed by attack, the end-user has gained nothing because the original data
has become unusable.
Improvements in this process occur when the embedded data is robust against
duplication and
modification.
The second embodiment utilizes a registration process, where the recording
device embeds its
registration in the data. In this embodiment, the recording device can refer
to a physical device, such as a
CD or DVD burner, or virtual device, such as an MP3 or AAC encoder. This
registration process allows
any illegal media to be traced back to the original owner assuming that
recording devices are registered
when purchased. At the very least, the illegal media may be traced back to the
specific recording device's
place of purchase, providing law enforcement with a good starting point.
CA 02364433 2001-08-08
WO 00/54453 PCTIUSOO/06296
21
The dynamic locking and unlocking embodiments improve the robustness of
existing or future
embedded data techniques to duplication and/or modification. Dynamic locking
causes the embedded
data to be dependent upon the media, e.g., by including one or both of the
following steps. The first step
includes modifying the auxiliary information by the media. The second step
includes encrypting the
auxiliary information, possibly modified in the first step. The encryption
technique may be RSA, DES or
any appropriate algorithm. After dynamically locking the auxiliary
information, it is embedded in the
original data. Each step of dynamic locking provides its own independent
advantages. However,
incorporating both steps produces auxiliary information that cannot be
transferred between media,
modified, or created.
The dynamic unlocking process performs the inverse steps, assuming each
specific step was
performed in the dynamic locking process. The first step involves decrypting
the retrieved data. The
second step involves unmodifying the output of the first step or the retrieved
data directly, depending
upon whether the first step was performed, and thus producing the original
auxiliary data.
Five exemplary utilizations of the enabling and dynamic locking process and
apparatus are
described briefly here and in detail below to aid in the understanding of both
processes and apparatus.
These utilizations include (1) distribution of compressed media such that it
can only be played by the
requester's playback device, (2) using the presence of the embedded data to
specify copy-once access, (3)
protection of DVD media, (4) photo-card validation, and (5) sending secure
secret messages.
In the first example utilization, a media player, such as a computer with MP3
software player,
contacts an Internet site to download media, such as a song in MP3 format. The
player sends its unique
identifier to the Internet site, where the identifier is modified using the
original data and the result is
encrypted. The modified and encrypted identifier is then embedded in the
original data, and the
combined data is downloaded to the player. The media player is able to extract
the identifier from the
combined data, and compare it to its own identifier. If these identifiers are
identical and any additional
information, such as a date limit, is verified, the player will play the data.
If the combined data is copied
to a second player having a different identifier, the second player will not
play the combined data.
If an unauthorized person were able to determine the identifier, he could then
embed it in other
songs and play them on his player. By encrypting the identifier, an
unauthorized person would be unable
to determine the identifier, even if he were able to extract the auxiliary
data from the combined data. In
addition, if the process did not include modifying the auxiliary information
with the original information,
the embedded data could be copied between media. Finally, the encryption key
also requires proper
handling, and the identifier may include additional information besides the
player identifier.
An example of the second utilization includes, rather than a unique
identifier, a predefined copy
code such as "allow no copying," "allow copying one time, but not copying of a
copy," and "allow
unlimited copying." The recorder would retrieve the copy code and not copy
unless permitted by the
code. The copy would either contain no "allow copying one time..." code, or
contain an "allow no
copying" code. For broadcasts, both the player and the broadcast unit would
know the code beforehand
(i.e. predefined) or the code would be included in the broadcast.
CA 02364433 2001-08-08
WO 00/54453 PCTIUSOO/06296
22
In the third example utilization, two approaches are described. In the first
approach, a DVD
player will not play the DVD without retrieving the predefined identifier
embedded in the original data.
For extra security the identifier could be encrypted with a key located at a
central database or in a section
of the DVD not available for copy. In the second approach, the identifier
could control the number of
generation of copies allowed, noting that if no identifier exists, no copies
can be made. Or, there could
be two layered identifiers for both types of copy management.
The fourth example utilization involves embedding secure data in the picture
of a photo-card, as
in a photo used for identification purposes like a driver's license or credit
card. If the retrieved
information at the photo-card reader does not match that of the central
database, the card is recognized as
a fake and will not be authorized for use. Note that the information and key
exchange must be securely
transmitted.
The fifth example utilization allows the secure transmission of secret
information, hidden in the
media. Most bystanders will not know the secret message is attached. If found,
the hidden information
cannot be read by, modified by, and/or transferred to other media by an
imposter when the embedded
data is dependent upon the media and encrypted. Different types of encryption,
symmetric or
public/private key, can be used for creating the desired protection or
authentication of the embedded data.
This hidden information enables a person or machine on the receiving side to
perform an action.
The exemplary apparatus for these processes involves a logic processor,
possibly including DSP
chips, host CPUs or custom analog or digital circuitry, and memory. The
configuration and machine
code are easily designed given this disclosure and familiarity with the state
of the art in cryptology and
electrical engineering.
Before turning to a detailed exposition of the foregoing, consider some
definitions. Media or
content includes, but is not limited to, audio, video, still images,
combinations of the above, and forms
related to other senses. The terms media and content are used interchangeably.
Media does not refer to a
storage medium. A media or content segment includes, but is not limited to, a
song, part of a song,
movie, part of a movie, part or all of a sound track, part or all of a still
image, a taste, a touch, and an
odor. Original data is the raw, unprotected data. The auxiliary information
refers to any data that is to be
embedded in the original data. The ID 140 in Fig. 21 refers to this auxiliary
information, and may
include but is not limited to, information such as the player ID, number of
copies allowed, usage time or
date limits, and content enhancement information such as author, copyright,
publisher, song lyrics or
image details. The embedded data is the data that is actually embedded in the
original data. The
embedded data differs from the auxiliary data by the transformation used in
the embedding process. This
transformation can include the modifying process and/or encryption involved in
dynamic locking, and the
embedding process such as bit manipulations, pulse-width modulation, or
spreading with frequency
transformation or pseudo-random noise sequence. The combined data results from
adding the embedded
data to the original data. Robustness to attack is defined as getting around
what the embedded data is
supposed to provide or prevent. Finally, a pirate is a person who tries to
illegally obtain the data or use
the protected device.
CA 02364433 2001-08-08
WO 00/54453 PCTIUSOO/06296
23
Enabling Process
Fig. 14 demonstrates an exemplary enabling process. This process uses a logic
processor 900
and memory 910, as shown in Fig 22. First, as shown in box 10, processor 900
retrieves the auxiliary
data from the combined data 5 and stores it in memory 910. Then, processor 900
determines whether the
embedded data allows the desired action, as shown in box 20. If so, the
desired action is allowed, as
shown in box 30. If not, the desired action is disallowed, as shown in box 40.
Registration Process
Fig. 15 demonstrates the registration process. This process involves assigning
a unique
registration code 305 to each recording device 300, and embedding the
registration code 305 into the
media when it is recorded, as shown in box 310. Then, when illegal media is
found in an open-market
320, it can be traced to the owner of the recording device via the
registration code 305, as shown in box
330.
The process is similar to gun registration assuming the recording device is
registered upon
purchase. At the very least, the illegal media could be traced back to the
recording device and its place of
purchase, thus aiding law enforcement.
This recording device may be a physical device or virtual device. A physical
device could
include a CD or DVD burner. A virtual device could include a software program
using processor 900
and memory 910 to digitally compress (bit rate reduce) audio, such as a MP3
ripper or AAC encoder.
Remember that media refers to the perceived data and not the storage medium.
Dynamic Locking
Fig. 16 displays the way in which dynamic locking blocks the duplication of
the auxiliary data.
Duplication is blocked for both bit-for-bit copying of the embedded data
between content, and retrieving
the embedded information, and re-embedding it into different content, such
that the different content
appears authentic.
Specifically, when only modifying the auxiliary information, a pirate will not
be able to move
the embedded data from one media segment to another without figuring out how
to correctly unmodify
and re-modify the embedded data. When only encrypting the auxiliary
information, a pirate will not be
able to obtain the auxiliary information. The pirate will be able to retrieve
the embedded data, but not
decipher it since it is encrypted. When both steps, the auxiliary information
cannot be moved from one
media segment to another. If it is moved directly, the modification step of
dynamic locking causes the
embedded data in the new media to be incorrectly unmodified because the values
in the new media
segment used for unmodifying the retrieved data don't match the values in the
original media where the
data was modified. If the pirate tries to unmodify and re-modify the embedded
data (since the details of
this step may be known), he/she must first have the key to decrypt the data in
order to move it to new
media segment.
CA 02364433 2001-08-08
WO 00/54453 PCT/US00/06296
24
Fig. 17 displays the input and output for the exclusive-or function (XOR). The
XOR is its own
inverse and extremely efficient.
Fig. 18A displays an overview of the dynamic locking and embedding process.
The whole
process contains three steps, but either one (not both) of the first two
steps, i.e. those steps of dynamic
locking, can be skipped. However, when both dynamic locking steps are
performed the difficulty in
duplicating the data is improved. In addition, the order of the last two steps
can be switched. This switch
is beneficial when the content, including the embedded data, is encrypted,
usually for other content
protection reasons, or when the modification step has some of the desirable
features such as requiring a
key to be unmodified.
In the first step, box 600, the auxiliary data (d), which is to be embedded,
is modified based
upon the original content (c). This step is designed to modify the auxiliary
data to be dependent upon the
original content such that the embedded data cannot be copied bit-for-bit
between content. The chosen
content bits should be critical to the content, such that they cannot be
changed in new content to make it
appear authentic. A desirable function is the exclusive-or (XOR) operator
since this function is its own
inverse and efficiently implemented on digital processors.
In the second step, box 610, the modified data is encrypted such that the
original auxiliary bits
cannot be obtained from the embedded data. Thus, the original auxiliary bits
cannot be re-embedded in
different content, making this different content appear authentic. If the
auxiliary data is not modified by
the original content before being encrypted, it could be copied bit-for-bit
from the original content to new
content making the new content appear authentic. Any existing or future
methods of encryption,
including DES and RSA, can be used, with known methods of key management, all
of which is well
described in the prior-art.
In the third step, box 620, the encrypted and modified (labeled dynamically
locked) auxiliary
data is embedded into the original content.
Fig. I 8B displays an overview of the process used to retrieve and dynamically
unlock the
auxiliary data. The whole process contains three steps, and each step should
only be performed if the
corresponding step was performed when the data was embedded. In addition, if
the order of the last two
steps was switched while embedding, these two corresponding steps should be
switched during this
retrieval process.
In the first step, box 630, the embedded data is retrieved from the content.
At this time, the
embedded data consists of encrypted and modified auxiliary data (assuming both
dynamic locking steps
were performed). In the second step, box 640, the retrieved data is decrypted.
In the third step, box 650,
the output of step two is unmodified. The result is the original auxiliary
data.
In addition, dynamic locking and unlocking can use correlated data. Correlated
data may
include information such as song lyrics or the address of the person in a
photographic identification card.
Fig. 19 shows several example implementations of the modification part of
dynamic locking and
unlocking when data is embedded such that it will not be perceived (i.e.
watermarking). Although, only
the modification part is shown in Fig 19, the modified auxiliary information
may be encrypted before
CA 02364433 2001-08-08
WO 00/54453 PCT/USOO/06296
being embedded and decrypted after being retrieved (but before being
unmodified), if desired. In
addition, the modification of the auxiliary information may be skipped, and
the auxiliary information may
be only encrypted before being embedded and decrypted after being retrieved.
The cryptology process is
not discussed in detail since someone familiar with the state of the art
easily understands its
5 implementation.
Fig 19A shows dynamic locking and unlocking as applied to an apparatus earlier-
described. For
dynamic locking, the peak value, box 200, or threshold crossing value, is used
in the exclusive-or (XOR)
calculation to modify the next N auxiliary information bits, where N is the
number of bits per sample in
the data (such as 16 bits for CD audio). Then, these modified N bits of the
auxiliary information are
10 optionally encrypted and embedded (e.g., by the above-detailed methods)
using locally masked bit
manipulations of difference A. This process is repeated for the next group of
N peaks, and so on, until
the whole modified auxiliary information is embedded or all the original data
has been used with
modified auxiliary information being repetitively embedded.
The embedded data can be retrieved using the process described above,
decrypted (if required),
15 and unmodified. The unmodifying process is the inverse of the modifying
process. Since the XOR
function is its own inverse, the peak values of the combined data and the
decrypted auxiliary information
are applied to the XOR function. Importantly, the peak values are identical to
those of the original data
since they were not changed during the embedding process.
For example, when using CD-audio, N is 16 bits. Thus, for this example, the
first 16 bits of the
20 auxiliary information are modified by the first peak value using the XOR.
Then, these modified auxiliary
information bits are optionally encrypted and embedded in the data points
after the current peak and the
next 15 peaks. This process is repeated for the following group of 16 peaks
and auxiliary information
bits, and so on, until all the data is embedded or all the original data has
been used. The modified and
optionally encrypted auxiliary information can be embedded over an over again
within the data, by
25 restarting the process with the first 16 bits of the auxiliary information
after all of the bits have been
embedded.
The embedded data can be retrieved, decrypted (if encrypted), and unmodified
with the inverse
of the XOR calculation, which is an XOR calculation. Thus, the first original
16 bits of the auxiliary
information can be obtained by performing the XOR calculation with the
retrieved and decrypted
embedded data and first peak value. The retrieving process is continued for
the next group of 16 peaks of
the combined and embedded data, and so on, until the whole auxiliary
information is found or all of the
combined data has been traversed.
It is very important to keep proper track of the position of the groups of 16
bits in the auxiliary
information when modifying for embedding, and unmodifying after retrieving.
The above-detailed technology allows sync pulses in the combined data. These
sync pulses can
be used to align the auxiliary information with the value used in modifying
the auxiliary information. For
example, rather than embedding data after the peak used to modify the
auxiliary information, a sync pulse
could be embedded and used for re-alignment during the retrieval process.
CA 02364433 2010-08-09
WO 00154453 PCT/US00/06296
26
Fig I 9B shows dynamic locking and unlocking as applied to Patent #5,774,452
"Apparatus and
method for encoding and decoding information in audio signals" by Jack
Wolosewicz of Aris
Technologies;. For this case, the data values occurring previously in
time to the embedding of the pulse-width modulated (PWM) bit stream and shown
in box 220, could be
used in an XOR operation with auxiliary information to modify and unmodify the
embedded data. In this
case, several data values would need to be used to modify all of the auxiliary
information. For example,
when using 16 bit data and embedding 256 bits of auxiliary information, the
dynamic locking and
unlocking process would use the previous 16 original data points to modify all
of the auxiliary
information. As long as the data is received in the same order as it.was
embedded, it does not matter if
the data values used to modify the auxiliary information overlap with the
previous embedded bit stream.
If one finds a configuration where the above does matter, it can easily be
handled by skipping the second
embedded bit stream and marking it as skipped in the combined data,
Fig 19C shows an overview of applying dynamic locking and unlocking to
embedded data
schemes based upon pseudo-random noise (PN) sequences. In one embodiment, the
PN sequence could
skip the Mth data point, as shown in boxes 250 and 270, where M is equal to
the number of bits per
sample in the data (N) times the length in bits of PN sequence segment applied
to each auxiliary
information bit. This M`" data point would be used in an XOR operation with b
bits of the auxiliary
information to modify the auxiliary information. For example, lets assume each
auxiliary information bit
is embedded with a 1024 bit segment of the PN sequence in 16 bit audio and the
auxiliary information is
64 bits long. Then, after adding the PN sequence to 16384 (M = 1024 bit PN
segment*1 6 bit audio) bits
of original data, another original data point is skipped to modify the
auxiliary information. It will take 4
(64 bit auxiliary information/I6 bit audio) of these segments to embed each
auxiliary information.
Equivalently, 4 adjacent original data points could be skipped every 65536
(1024 bit PN segment* 16 bit
audio*4 PN segments) original data points and embed the whole modified
auxiliary information in one
continuous stream of four PN segments.
This modified and optionally encrypted auxiliary information can be used to
control the fashion
in which the PN sequence is added to the original data, as well known in the
state of the art of spread
spectrum technology. Specifically, in many applications, the PN sequence will
be phase shifted by the
modified auxiliary information (i.e. where 0 scales and adds the negative
value of the PN sequence and 1
scales and adds the positive value) or simply multiplied by the auxiliary
information. Once retrieved, the
modified auxiliary information could be unmodified using the inverse XOR
calculation with the skipped
data point.
Another embodiment for PN sequences is using the skipped data point to modify
the next N bits
of the PN sequence, not the auxiliary information. If one point is skipped,
the number of PN bits
modified, M, should be equal to N. the number of bits in the data. If two
points are skipped, M is equal
to 2*N, and so on. Modifying the PN sequence using an XOR calculation and
optional encryption is one
scheme. However, this may reduce the randomness of the PN sequence, and other
modification functions
CA 02364433 2010-08-09
WO 00/54453 PCT/tJS00/06296
27
can be employed to maintain randomness. Finally, the modified and optionally
encrypted PN sequence is
embedded in the media data and used to retrieve the embedded data.
In a final implementation of dynamic locking, it is applied to embedding
methods that use PN
sequences to determine where to place the auxiliary information in the
original data, possibly after being
transformed into the frequency domain. Such methods include that of patents
#5,613.004 and #5,687.236
"Steganographic method and device" by Marc Cooperman and Scott Moskowitz of
the Dice Company,
incorporated herein by reference, and patent-pending technology of AT&T labs
(Lacy J, Quackenbush
SR. Reibman AR, Shur D, Snyder JH. (1998) "On combining watermarking with
perceptual coding."
ICASSP'98 Seattle, WA.), detailed in EP889471. For these methods,
the PN sequence used to embed the data could be required to never have more
than N continuous embed
bits and start with a non-embed bit, where N is again the number of bits per
sample in the original data.
Then, the original data point adjacent and previous to the first embed bit
modifies the next N bits of the
auxiliary information. This process may be repeated until all the original
data is embedded, such that the
modified auxiliary information is embedded repetitively. For example, when
embedding in the frequency
domain from low to high frequency with 16 bit data and a 32 bit auxiliary
information, the non-embed bit
in the frequency bin just below first embed bit is used to modify the next 16
auxiliary information bits.
Then, the non-embed bit in the frequency bin just below the 17d' embed bit is
used to modify the next 16
auxiliary information bits. Next, the non-embed bit in the frequency bin just
below the 33rd embed bit is
used to modify the first 16 auxiliary information bits, and so on.
In a similar scheme, the PN sequence could be applied to every other or k'h
(where k < N bits per
sample in original data) data point, such that no limitations need be applied
to the PN sequence. The
process guarantees to have a non-embed bit next to the N'h embed bit, and is
implemented is similar
fashion the previous method.
For all these methods of dynamic locking using PN sequences, the dynamic
unlocking process is
the inverse and obvious to a person familiar with the art given the previous
disclosure.
Fig. 20 demonstrates applying dynamic locking and unlocking to data embedded
in header, not
content, data. Fig. 20A displays the pseudo-code for the dynamic locking
process. In general, the
auxiliary data bits, of length L, are locked and placed in the header of
frames of the content and
repetitively embedded.
Specifically, the process of Fig. 20A starts at the beginning of the content
bits (box 700) and
auxiliary data bits (box 705). Then, L auxiliary data bits are locked by being
modified with L bits of the
content using the XOR or applicable function, and/or encrypted (box 735).
These L content bits should
be critical to the either or both file format and content, such that they
cannot be replicated in a different
media segment without disturbing it. Next, M bits of locked auxiliary data are
embedded in the frame
header (box 710). These M bits should be less than L. and preferably L is
divisible by M, such that the L
bits are embedded in L/M frame headers. If L is not divisible by M. a person
familiar with the state of the
art can easily handle the offset. Then, the content is checked to see if more
frames exist (box 715). If
there are no content frames left, the process is completed (box 730). If there
are more content frames, the
CA 02364433 2001-08-08
WO 00/54453 PCT/USOO/06296
28
auxiliary data is checked to see if any previously modified bits exist (box
720). If there are previously
modified auxiliary bits left, the next frame is read (box 725), and the
process is continued at box 710. If
there are no previously modified auxiliary bits left, the next content frame
is read (box 740), the auxiliary
data is re-started at bit 0 (box 705), and the process is continued at box
710.
This process assumes the auxiliary information is of length L and L is
reasonably short for ease
of explanation. It is obvious that if you have a very large number of
auxiliary bits, you can break them
into segments of length L, and rather than starting at the first auxiliary bit
each time, start at the kth
segment. To this end, the auxiliary bits are embedded within the data, broken
into segments of length L
and each segment is embedded in L/M frame headers.
To increase robustness to attack, a pseudo-random noise (PN) bit sequence
could be used and
the first N critical content bits with a corresponding PN bit value of I could
be used for modifying the
auxiliary information.
Alternatively, only the first important M content bits in each frame, rather
than L bits every L/M
frames, are used in the XOR calculation when embedding M locked auxiliary data
bits in each frame. In
this case, the auxiliary data bits are modified in each frame, specifically,
between boxes 725 and 710 in
Fig 20A. Once again, a PN sequence could be used to randomize which M bits of
original audio are
used. Importantly, M must be large enough so that error correction in new
content cannot repair all the
content bits that need to be changed, such that a bit-for-bit transfer of the
auxiliary data makes the new
content appear authentic. The value of M depends upon the frame size and
desired bit rate.
When using compressed content, such as MPEG data, specifically Layer III (MP3)
or AAC
audio as specified in the MPEG2 specifications, including the MPEG I and 2
specifications, ISO 11 172-
3 and ISO 13818-7 respectively, herein by reference, the frames and header
bits are pre-defined. The
private, copyright, or ancillary bits can thus be used to embed the data. When
using content without pre-
defined frames, such as in raw PCM audio, databases, or software applications,
the frames can simply be
created. For example, the content could be arbitrarily divided into 1024 bit
frames with header bits for
the embedded data.
Alternatively, the locked auxiliary data could be placed only in the global
header, defined as the
header for the complete file, or in a linked but separate file. These two
cases are less secure than
embedding the data throughout the file. More bits mean the data will be more
robust to attack via brute
force. For broadcast content, the data should be embedded throughout the
content as described above so
the rendering device or person can receive the auxiliary information and
respond accordingly from any
point in the broadcast.
Fig. 20B displays the pseudo-code for the retrieval and dynamic unlocking
process for the
auxiliary data embedded in Fig 20A. In general, the auxiliary data bits are
retrieved by reading them
from the header of the content frames and unlocking them, in a repetitive
manner.
Specifically, the process of Fig. 20B starts at the beginning of the content
bits (box 750) and
auxiliary data bits (box 755). Then, N content bits are saved, such as in
memory 910 of Fig. 22, so they
can be used to unlock the next N retrieved auxiliary data (box 785). Next, M
bits of locked auxiliary data
CA 02364433 2001-08-08
WO 00/54453 PCTIUSOO/06296
29
are read from the frame header (box 760). Then, the content is checked for
existing frames (box 765). If
there are no content frames left, the process is completed (box 780). If there
are content frames left, the
auxiliary data bits are checked to see any exist (box 770). If there are
auxiliary bits left, the next frame is
read (box 775), and the process is continued at box 760. If there are no
auxiliary bits left, the retrieved
auxiliary data is unlocked (box 790), the next frame is read (box 795), the
auxiliary data is re-started at
bit 0 (box 755), another N content bits are saved (box 785) and the process is
continued at box 760.
For this example, unlocking the retrieved embedded data (box 790) involves
performing an
XOR operation (since it is its own inverse) on the N content bits that were
saved in box 785 and the last
N retrieved embedded data bits, and decrypting, if required. In addition,
since the data is repetitively
embedded in each frame, the retrieving process must overlay the auxiliary data
bits after the last bit as
received (box 790) and make sure the auxiliary data bits do not change
throughout the file. If the
auxiliary data bits change throughout the file, the file is not authentic.
Alternatively, if another modification function was used, its inverse should
be used.
Importantly, the same retrieved auxiliary bits and original content bits
should be used in the inverse
calculation as were used in the modifying calculation. For the embedding
example where the first M
audio bits of the frame were used to modify the auxiliary data, the first M
audio bits of the frame should
be used to unmodify the modified auxiliary data, which was retrieved and
decrypted. If a PN sequence
was used to modify the auxiliary data, the same PN sequence should be used to
unmodify the data.
If an alternative embedding step was used, the auxiliary bits are retrieved
accordingly. For
example, if bits are embedded in the global header or linked file, the are
read from the global header or
linked file, respectively.
Finally, the appropriate steps should be taken if the auxiliary data is longer
than L or L is not
divisible by M. These steps are obvious to a person familiar with the state of
the art given the above
explanations about dynamic locking and unlocking.
Example Utilization
These five example utilizations are described to aid in understanding the
enabling and dynamic
locking process and apparatus. The general underlying process for these
examples is displayed in Fig. 21
and the corresponding apparatus is shown in Figs. 22 and 23. The process, in
general, begins with a
sending device 100, dynamically locking an ID 140 as shown in box 110, and
embedding the locked ID
within the media as shown in box 120. Remember, as defined at the beginning of
this section, the term
ID usually refers to an identifier, but can include any auxiliary information.
The sending device 100 may
be an encoder, recorder, transmitter, storage medium, or the like.
The media is then transmitted to a receiving device 130 in which the locked ID
is retrieved as
shown in box 160, and dynamically unlocked as shown in box 170. Then, the
proper action is enabled if
allowed by the retrieved ID 140, as shown in box 180. The receiving device 130
may be a decoder,
player, recorder, and/or the like.
CA 02364433 2001-08-08
WO 00/54453 PCT/US00/06296
When the dynamic locking and unlocking processes include encryption and
decryption, the
encryption key must be located somewhere and transmitted safely, as shown in
boxes 151, 152, and 153.
Transmitting the key safely is well understood by one familiar with the state
of the art in cryptology. The
location of the key depends upon the requirements of the utilization. The five
utilizations demonstrate
5 various key locations. For most utilizations, the key will be available only
in one of the three possible
locations. In addition, the encryption and decryption key will usually be
identical (symmetric), and
referred to as the encryption key in the discussion below. However,
public/private key encryption could
also be used in many of these situations. When discussing private/public
encryption below the key will
be specified as the public or private encryption key. Finally, certain
utilizations may not need to transmit
10 the auxiliary information since the values are predefined.
In addition, the use and location of ID 140, the types of sending devices 100
and receiving
devices 130 are also explained in more detail in these example utilizations.
The five example utilizations include distribution of MP3 data, copy-once
access to broadcast
data, DVD copy protection, photo-card verification, and secret data
transmission. From these
15 explanation, many more utilizations are obvious.
Regarding distribution of MP3 data, the concept is explained using several
scenarios. All
scenarios include both a software PC-based and portable MP3 audio player, and
distribution via the
Internet.
In the first scenario, the MP3 data exists on the Internet and is purchased by
an end-user. The
20 delivery system interacts with the end-user's player, securely transmitting
ID 140 and the encryption key,
shown in box 151, from the receiving device to the sending device, and
dynamically locking, including
encryption, the ID 140 in the MP3 data. In this scenario, the encryption key,
shown in box 151, is
located on the end-user's player. After the MP3 file is delivered (i.e.
downloaded) only the end-user's
player can play the data since other players will have different IDs. A
portable and PC-based player may
25 share the ID 140, and this is easily implemented by a software program and
current digital electronics,
such as EPROM or flash memory. Since the ID 140 is dynamically locked the end-
user cannot extract
the ID 140 and use it in another song or MP3 file.
In another scenario, the MP3 encoder and player may be part of one software
program, which
transforms CD, DVD or broadcast audio into MP3 audio with the embedded data
containing the
30 dynamically locked, including encryption, ID 140. In such an example, there
is no need for the exchange
of an encryption key, as displayed in box 151, and ID 140. The software
applications should be
programmed such that the key and ID 140 are protected from the end-user, as
well known in the state of
the art in software. Again, the key, shown in box 151, is located within the
end-user's player. The
transformed MP3 audio is now only playable on the end-user's system and/or
portable player, and it is
not possible to move the ID 140 to another song as described above.
In yet another scenario, the key could be located in a central database, as
shown in box 152.
This configuration allows a different key for each player and MP3 audio
sample. This configuration
CA 02364433 2010-08-09
WO 00/54453 PETIUS00/06296
31
increases robustness to attack since new keys are used for each song, but
involves extra management
tools and responsibilities.
In a final scenario for MP3 audio, the ID 140 could contain time limits for
listening to the audio
or a date limit that, when exceeded, the audio will not play. The player will
keep track of how many
times the song has been played or whether the date has expired. The ID 140
could contain a demo code,
which does not limit the song to one player.
Regarding copy-once access (defined as allowing an end-user to copy the media
only once,
perhaps for time shifting purposes), the concept is explained in terms of the
broadcast of a movie. With
broadcast media, it is best if everyone shares the same encryption key. The
key, as shown in box 153,
could be broadcast embedded in the movie. and changed for each broadcast. In
addition, if the embedded
data is not encrypted, there is no need for a key, thus simplifying
transmission. Finally, the copy-once ID
140 will be predefined, meaning that it is already defined in the transmitting
and receiving device, as
shown in Fig. 21 where ID 140 has an optional location in the transmitting
device. Once the broadcast is
received and the retrieved ID 140 enables the data to be recorded, the
recorder can record the movie and
either remove the copy-once ID 140 or change the ID 140 to a predefined code
that informs other
recorders that the media has been copied once.
Regarding DVD copy protection, there are two scenarios. In the first scenario,
the player will
not play the media unless the embedded ID enables the action. The encryption
key, as shown in box 153,
is be included on the DVD in a non-copy access location, This means the user
will be able to play the
media only when the DVD disk is present since the player will not play the DVD
data without retrieving
the correct ID. A copy of the entire DVD (minus the encryption key since it is
unable to be copied) or a
copy of a content file will be unusable since the key to decrypt the embedded
data will not be found and
without it the player will not work.
In addition, the key could be located in a centrally accessible database, and
possibly linked to
the requesting end-user player, as shown in box 152. This configuration
increases robustness to attack
since access to the key is monitored, but includes extra management
responsibilities for the content
provider and additional time for the end-user. The key could also be purchased
and encrypted by the key
in your player as described in U.S. patent 5,933,498 to Paul Schneck..
Once again, the ID 140 will be predefined, and exist in the sending device
100.
In a different scenario. the predefined ID 140 could be used to enable the
recorder, and allow a
certain number of copy generations, or a copy of only the original, known as
serial copy management. ID
140 could be modified to allow one less recording generation each time the DVD
is recorded. Possibly
through keeping track of the recorded generation and originally allowed count
or by reducing the allowed
count. For serial copy management. the watermark could be removed in the
second generation DVD.
Remember that in this approach if the watermark does not exist, no copies can
be made, Finally, there
could be two-layered ID 140s for both types of copy management.
The photo-card utilization example involves having the picture in the photo-
card embedded with
the ID 140. If the correct information is not present, the card is a fake and
will not be authorized for use.
CA 02364433 2005-03-09
WO 00/54453 PCT/USOO/06296
32
To increase the security of the method dynamic locking is applied; the ID 140
is reversibly modified by
the photograph or connected data such as the corresponding name and address,
and encrypted, such that
the information cannot be copied between cards or from a legitimate card to an
illegal card. The
matching ID 140 and encryption key can be stored at a database only accessible
by every sending device
(i.e. in the sending device) and securely transmitted between the database and
the photo-card reading
device, such as using RSA key exchange or any other method known in the state
of the art of cryptology.
Besides being as secure as other cryptology techniques, another advantage of
this process is that it
requires transmission of minimal data, including the short ID 140 and
encryption key.
The last example utilization allows the secure transmission of secret
information in ID 140,
hidden in the media. Most bystanders will not know the secret message is
attached. Once the receiving
device extracts the hidden message, the receiving device, a connected device,
or a human will be enabled
by the hidden information contained in ID 140. If found, the hidden
information can be protected from
being moved to other media segments and/or interpreted by using dynamic
locking with various
encryption schemes. For example, if the secret information is encrypted with
your public key, only you
can recover it. Or if it is encrypted with your private key, people or devices
receiving the message using
your public key know it was signed by you and is authentic. If it is encrypted
with a symmetric key, only
the holders of the key could have created and read the message. Finally, if
the modification step of
dynamic locking is used, the receiver knows the message was not transferred
from a different media
segment.
Apparatus
Fig. 22 shows an exemplary apparatus used to implement the enabling,
registration, and dynamic
locking processes. The hardware includes a logic processor 900 and memory 910.
The logic processor
900 may be defined as the equivalent of a digital signal processor (DSP),
general-purpose central
processing unit (CPU), or a specialized ASIC chip. A likely DSP chip is one of
the Texas Instruments
TMS320 product line. A CPU could include one of Intel's Pentium line or
Motorola/IBM's PowerPC
product line. The design is straightforward for someone familiar with the
state of the art given the
description of these processes. The memory 910 includes any type of memory.
Fig 23 shows more detail of the apparatus for dynamic locking. Specifically,
the logic
processor 900 and memory 910 must work together to act as the modifier 1010
and encrypter 1040.
Modifier 1010 performs the modification step of dynamic locking. Encrypter
1040 performs the.
encryption step of dynamic locking.
Fig 2 4 shows more detail of the apparatus for dynamic unlocking.
Specifically, the logic
processor 900 and memory 910 must work together to act as the decrypter 1045
and the unmodifier 1015.
The decrypter 1045 performs the decryption step of dynamic unlocking. The
unmodifier 1015 performs
the unmodifying step of dynamic unlocking: The unmodifier 1015 and 'decrypter
1045 of dynamic
unlocking may use the same or different circuitry as the modifier 1010 and
encrypter 1040 of dynamic
CA 02364433 2001-08-08
WO 00/54453 PCT/USOO/06296
33
locking. However, when using the same circuitry, the dynamic locking and
unlocking processes would
use different control programs.
Binding and ID Assignment
As noted, another aspect of the technology detailed herein relates to media-
binding, e.g., the
fashion in which consumers legitimately access protected content while
controlling piracy. A basic
concept is that the content contains an ID that locks it to a particular user
or broadcast and the rendering
device automatically determines whether the content can be accessed based upon
the current and
previously rendered IDs and rules. Such technology may result in increased
sales of content for the
content providers.
One aspect of the technology resides in having the rendering device keeping
track of the IDs
contained in both the current and previously accessed content. This allows the
rendering device to
control access to new content based upon the new content's ID, the rules
provided with the content (by
the content providers) and/or within the device, and the IDs from previously
rendered content by the
device.
The ID may be linked to the user or the broadcast. User IDs work well for
content that is sold for
a user's continued use, whereas broadcast IDs work well for content recorded
by the user from a
broadcast.
An example implementation is as follows. For user-linked content, the
rendering device includes
constraints that limit the number of content tracks with different user IDs
that can be accessed in a certain
amount of time, possibly influenced by the number of times content with each
user ID has already been
accessed. For broadcast content, broadcast IDs and the optionally included
rules can be used to limit
rendering or copying of each broadcast. In other words, with broadcast IDs,
the limits are based upon
date or number of times that ID is played, not on the total number of
broadcast IDs.
More specifically, a portable MP3 player can keep track of each song's user
ID, and if the
previously played songs contain more than N different user IDs, the player
decides if it can replace an old
user ID with the new one due to the old user ID's date and number of times
songs with that ID have been
played. Similarly, if a broadcast ID is contained in memory, the MP3 player
notes that the user has
played the audio X times and Y times is allowed by the broadcast, or the date
is past the broadcast's
allowable usage date.
To this end, it is easy for the consumer to use the device, as he/she is not
required to posses an ID
card. In addition, there is no need for a global database linking the user to
the ID; thus, the user's privacy
is not compromised. For example, if a user loses his/her ID, it can be
obtained from previous content.
However, the user or broadcast ID can be kept secret and other privacy methods
can be used. Most
importantly, access to the media is limited, as the content providers wish,
but the user is not
inconvenienced.
Again, a review of relevant terminology may be in order. The rendering device
is a device that
can play, view, record or perform a similar action upon the data. The
rendering device can provide any
CA 02364433 2001-08-08
WO 00/54453 PCT/US00/06296
34
type of perceived data, including but not limited to images, audio and video.
If the rendering device has a
portable section, such as with a MP3 player, the loader, which puts the
content onto the rendering device,
is considered as part of the rendering device. The ID may be a user or
broadcast ID. For example, many
MP3 players can also record broadcasts, and these broadcasts will, in the
future, contain embedded
broadcast IDs, possibly as watermarks or header data with digital broadcasts.
Content refers to the
desired audio, video, image, or other relevant perceived data. Content
providers include but are not
limited to record labels, movie studios, and independent artists. The ID may
be embedded within the
content such as bits in the header file or a watermark, or the ID can be
linked to the encryption and
decryption of the content. Finally, this automatic ID management may be used
in conjunction with other
methods, such as media-binding.
Fig. 25 displays an overview of ane automatic ID management process. In the
process, the
rendering device 100 keeps track of the IDs contained within the content it
has previously accessed (box
110). The rules 120 may be provided in the device hardware and/or contained
with the content. The
rules 120 decide whether or not the device can access the new content based
upon its ID (box 130).
If the rendering device has a portable section, such as with a MP3 player, the
loader, defined
above as part of the rendering device, can be used to lower the amount of
memory required within the
portable section, thus lowering its costs. This means that with a portable
rendering device, the portable
section may contain all of the memory and processing hardware (described in
detail below) required to
perform this automatic ID handling, or the hardware may be split between the
loader and portable section.
For example, when a computer uses a software loader to put MP3 files onto a
portable MP3 player, the
loader may store all the information about IDs on the computer and all the
rendering device needs to do is
count the number of times each song is played and maintain date information
for its current list of
content.
Fig. 26 displays the pseudo-code to implement an example of the process. In
this example, the
rules 120 include constraints 245, which are contained within the content as
specified by the content
provider, as well as default rules contained with the rendering device
hardware. The constraints 245 are
retrieved from the content 200 (box 240). The constraints 245 may limit the
number of content tracks
with different IDs that a device can access during a set time-period. The
constraints 245 may also change
the time-period an ID is stored dependent upon the number of times content
with a specific ID was
accessed. The constraints 245 may be embedded within the content or attached
as a header information
or a linked file.
For ease-of-use, it is better to not change these constraints per song because
it may confuse the
user. Ideally, the constraints should be agreed upon and set in the rendering
device. However, including
the rules in the content is a viable option.
Before describing further details of this exemplary process, it is important
to understand the
illustrative apparatus for implementing the automatic ID management process
(Fig. 27). The hardware
includes a logic processor 300 and a memory 310. The logic processor 300 may
be defined as the
equivalent of a digital signal processor (DSP), general-purpose central
processing unit (CPU), or a
CA 02364433 2001-08-08
WO 00/54453 PCTIUSOO/06296
specialized CPU, including media processors. Again, a likely DSP chip is one
of the Texas Instruments
TMS320 product line. A CPU could include one of Intel's Pentium line or
Motorola/IBM's PowerPC
product line. The design of code for controlling logic processor 300 is simple
for someone familiar with
the state of the art given the above pseudo-code and description.
5 In addition, a person familiar with the state of the art could implement the
logic processor 300
using analog and digital circuitry, either separate or in an application
specific integrated circuit (ASIC).
The analog and digital circuitry could include any combination of the
following devices: digital-to-analog
converters (D/A), comparators, sample-and-hold circuits, delay elements,
analog-to-digital converters
(A/D), and programmable logic controllers (PLC). Programmable logic arrays
(PLDs) can likewise be
10 used.
The memory 310 stores the information required by rules 120, such as IDs, last
play date, and
the number of times that content with each ID has been accessed. Memory 310
may consist of standard
computer random access memory (RAM). It is also desirable if memory 310
maintains this information
even without power in the rendering device, perhaps but not limited to using
ROM with backup and
15 chargeable battery power, or memory that is stable without power, such as
EPROM. As discussed above,
memory 310 may consist of two separate modules when using a portable section
and loader.
Now, back to a detailed description of the example process. It begins with the
device 100
receiving new content 200. From the content 200, an ID 210 is retrieved. The
ID 210 is checked to see
if it is a user or broadcast ID (box 215).
20 For user IDs, the following happens. If the ID 210 already exists in the
memory 310 of device
100 (box 220), the play count and last access date are updated (box 222) and
the content 200 is rendered
(box 230). If the ID 210 does not exist in memory 310 (box 220), the rules 120
are checked. If another
ID can exist in memory 310 (box 250), ID 210 and the current date are added to
the memory 310 (box
260) and the content is rendered (box 230). If another ID cannot be added, the
rules 120 are checked to
25 see if any existing IDs can be replaced because they are too old (box 270).
If any IDs can be replaced,
the old ID is replaced with ID 210 (box 280) and the content is rendered (box
230). If no IDs can be
replaced, the user may be warned and access to content 200 is denied or
limited (box 290). The user may
also be presented with a link to buy the content (box 290).
More specifically, the rules may allow a device to store 10 IDs, and IDs can
be replaced if they
30 have not been accessed for a week.
In addition, the number of times an ID has been rendered could be used to
determine whether or
not to replace the old ID with a new one (box 270). This count value could
influence the time period an
ID is held is memory 310; thus allowing ID 210 to replace a stored ID (boxes
270 and 280). For
example, if content associated with the stored ID has not been accessed in a
week, it can be replaced.
35 Conversely, if content associated with the stored ID has been played at
least 7 times, it should be held for
at least a month since its last access.
There are many other simple rules that can be designed to meet the specific
needs of the content
provider. Some may involve using difference equations to decide whether or not
an ID can be replaced.
CA 02364433 2001-08-08
WO 00/54453 PCT/US00/06296
36
For example, the count for an ID can be reduced by one each day and
incremented by one for each
rendering of content containing the ID, and the ID can be replaced (box 270)
if the count is zero or less,
or the date of last access is over a week.
For broadcast IDs, the following happens. The ID 210 is examined to see if it
already exists in
memory 310 (box 255). If not, the ID 210 and current date are added to the
rendering devices memory
310 (box 265), and the content is rendered (box 230). If the ID 210 does exist
in memory, the play count,
record date and/or last access date are checked to see if the content can be
rendered (box 275). The
broadcast may allow only two renders, or one week of rendering, or rendering
until a specific date. If the
broadcast is allowed to be rendered, the count and last access date are
updated (box 285) and the content
is accessed (box 230). If the broadcast is not allowed to be rendered, the
user is notified, the access is
limited and a link to buy the broadcast or similar content may be provided, if
applicable (box 295).
In addition, the device should probably have some way to reset all of the
information, such as IDs,
date and count. The reset function may require a password code that is pseudo-
random, thus requiring
the user to contact support to reset the device. For example, the password may
depend upon the day and
year and obtained from an automation system. The reset button may also delete
all the current content as
well as ID information. This allows people to use one portable player with
many friends at a party, but
the loss of content will discourage piracy since it will be cumbersome.
Figure 28 shows a portable MP3 player 400 that contains the described
apparatus implementing
the described pseudo-code. In this case, the logic processor 300 could be a
separate processor, or share
access with the processor that decompresses the audio. The device also
contains the necessary memory
310 to store the required information, such as ID, data and count, possibly
even when the player 400 is
without power. The device may share this memory with a software loader.
Finally, in any rendering device, the logic processor 300 could be a separate
processor or share
time with the processor handling content for the device, such as compressing
or decompressing digital
content.
Multiple Watermarks
Various advantages can accrue from using multiple watermarks instead of just
one. In an
exemplary system, one watermark is robust and declares that the media is
protected. This watermark is
embedded when the media is encoded into the desired format, such as MP3. This
means that the intensity
of adding the watermark is not an issue because the watermark is only added to
the audio once, and
copied with the audio by the distributor.
The other watermark declares that it is okay to play or record the media. It
is efficient, and does
not need to be difficult to remove, since removing it produces no advantageous
results. The efficiency of
this watermark is desirable since it must be embedded each time the audio
reproduced, such as
downloaded on the Internet, to link the media to the user, player, recorder
and/or storage device. Thus, it
greatly reduces the cost of copy management for the distributor. In addition,
it lowers the cost of the
players, since usually they only have to find this efficient watermark. Only
when it does not exist, does
CA 02364433 2001-08-08
WO 00/54453 PCTIUSOO/06296
37
the player need to determine if the audio is protected with the robust but
computationally intense
watermark.
Most importantly, non-protected media may contain neither watermark and can be
played by any
device from any storage.
In more detail, Fig. 29 displays a process employing two watermarks. Media 100
exists in an
insecure format, meaning that devices can play the media 100 even if it does
not contain any copy
protection and/or authentication watermarks. It is a format in which some
artists wish to freely distribute
their content, such as MP3. However, there are interested parties who don't
want to distribute their media
in the same format without allowing it to be freely copied and redistributed.
Watermark 110 declares that the media is protected. Watermark 110 must be
extremely difficult
to remove, and is allowed to be computationally intense. Many existing
watermark methods meet this
description, and future ones will certainly be designed.
Watermark 120 links the media to the user, player, recorder and/or storage
device. This link
determines if the user may copy and/or play the media. Watermark 120 must be a
computationally
efficient method that is hard to imitate.
Both watermarks are embedded at specific times in the reproduction process, as
shown in Figs.
29 and 30. Watermark 110 is embedded when the audio is encoded, and copied
with the audio when
distributed. Thus, the computational intensity of adding the watermark is not
that important. Watermark
120 is embedded when the media is reproduced, such as being distributed,
placed on permanent storage,
or encoded from an alternative form by a personal encoding device. The term
reproduced refers to the
legal transformation or distribution of the media, whereas copying refers to
an individual producing an
exact bit-for-bit replication of the media for legal or illegal utilization.
Since watermark 120 is embedded
every time the media is reproduced, its efficiency creates a reduction in
cost. Since watermark 120 is
embedded after watermark 110 it must be okay to layer the watermarks, as known
to be possible with
existing technology.
Optimally, the watermarks are search and retrieved in a specific order, as
shown in Figs. 29 and
31. First, the media is searched for watermark 120 (box 300). If watermark 120
is retrieved (box 310)
the embedded information is evaluated (box 320). If the embedded information
is correct, the desired
action is enabled (box 330). Alternatively, if the embedded information is not
correct, the desired action
is disabled (box 340). Only if watermark 120 is not found does the media need
to be searched for the
computationally intense watermark 110 (box 350). If watermark 110 declares the
media protected, then
the desired action is disabled (box 340). If watermark 120 is not present (or
declares the media to be
free), the desire action is allowed (box 330).
The just-detailed process can be used to restrict copying and/or playing of
the media.
Fig. 32 shows hardware apparatus that may be used to implement the invented
processes. The
hardware includes a logic processor 400 and a storage unit 410. The logic
processor 400 may be defined
as the equivalent of a digital signal processor (DSP), general-purpose central
processing unit (CPU), or a
specialized CPU, including media processors. A likely DSP chip is one of the
Texas Instruments
CA 02364433 2001-08-08
WO 00/54453 PCT/USOO/06296
38
TMS320 product line. A CPU could include one of Intel's Pentium line or
Motorola/IBM's PowerPC
product line. The design is simple for someone familiar with the state of the
art given the above
pseudocode and description. The storage unit 410 includes RAM when using a
digital processor.
A person familiar with the state of the art could alternatively implement the
process with analog
and digital circuitry, either separate or in an application specific
integrated circuit (ASIC). The analog
and digital circuitry could include any combination of the following devices:
a digital-to-analog converter
(D/A), comparators, sample-and-hold circuits, delay elements, analog-to-
digital converter (A/D), and
programmable logic controllers (PLC). Programmable logic arrays (PLDs) can
likewise be used.
Content Scrambling
As noted, it is often desirable to scramble content signals. The following
discussion reviews
certain improvements to such scrambling technology.
One such scrambling technique involves searching through the original digital
data for detection
criteria and then adjusting neighboring points to degrade the content, either
without affecting the location
of the detection criteria or affecting it in a known fashion so that the
original signal may be recovered.
The detection criteria may include the relationship between several points, or
be as simple as a threshold
crossing or include every M`h point. The adjustment of the neighboring points
may be as simple as
multiplying the point after the threshold crossing by N. It is advantageous if
N is less than one but not
equal to zero so saturation and data points equal to zero are not a problem,
and if the threshold is positive
and the data is decreasing towards zero during the threshold crossing.
The process can include searching through the data for the detection criteria
and then re-
adjusting neighboring points to their original value. For example, if the
adjustment in the degradation
process uses multiplication by N, the recovery process multiplies by I/N.
In the following discussion, digital content refers to digital data
representing a perceived
physical item, including but not limited to audio, video, and images. Digital
data refers to the grouping of
bits (I's or 0's) that represent a sample of the original digital content at
an instant in time. Each bit group
is equivalently referred to as a data point or sample. The data points are
arranged in an order, many times
representing a sequence versus time or frequency. In addition, the data points
may be grouped again to
form a subgroup, possibly used to represent a sequence versus frequency versus
time, as is the case in
MPEG standard compressed digital audio and video. Most importantly, the
digital data has an order,
with a beginning and end, such that searching the data is possible, and
neighboring points can be defined
as points close to each other. Finally, point(s) refer to one or several
points.
Fig. 36 displays an overview of the degradation and recovery process, and Fig.
37 displays the
corresponding pseudocode to be implemented by the apparatus.
To degrade the digital content (box 100), the samples are searched for the
detection criteria
(boxes 200, 210 and 220). The searching stops after the last data point in the
buffer has been examined
(box 210), and a new buffer may be presented if available. As known in the
state of the art, data values
must be saved between buffers and properly initialized for the first buffer so
as the initial points are
properly searched.
CA 02364433 2001-08-08
WO 00/54453 PCT/US00/06296
39
When the detection criteria are found, the neighboring data point(s) are
adjusted so as to cause
content degradation (box 230). The adjustment of these points should not
change the location of the
detection criteria or change it in a known fashion; otherwise, the detection
of the correct location to re-
adjust the data to its original value (recovery) is not easy. In addition, it
is desirable to prevent the
adjustment from causing saturation or resulting in a value of zero, because
then the original data point(s)
will not be easily recoverable.
To recover the original digital content (box 110), the degraded data is
searched for the detection
criteria defined by the degradation process (box 200, 210, and 220). If the
degradation process has
changed the detection criteria in a known fashion, then the detection criteria
in box 220 for recovery is
different than that used in degradation. When the criteria location is found,
the neighboring data point(s)
are re-adjusted by the inverse of the method used in the degradation process
(box 230).
An example of this process is shown in Figs. 38 and 39. In this case (boxes
300 and 310), the
detection criterion is a threshold crossing (using c-notation: x[n-I ]>thr &&
x[n]<thr) with a positive
threshold (thr>0) while the data goes towards zero (boxes 400, 410 and 420).
The neighboring point(s)
include only the point after the threshold crossing (box 430). To degrade the
data, the adjustment
involves multiplying the data point after the threshold crossing (x[n]) by N,
where N is less than 1 (box
430). By reducing the value of this data point, the detection criteria
location is not changed. In addition,
the closer N is to 0 (but not equal to 0), the more the digital content is
degraded. To recover the original
digital data, the point after the threshold crossing (x[n]) is multiplied by
I/N (box 430).
There are additional simplistic detection criteria that can be used. For
example, every Mt' data
point may be degraded. In this case, synchronization for recovery may require
scanning the data for M
points until the correct degraded locations are found. In addition, peak
values could be used, and the
point after the peak could be reduced in value. As desired, this will not
affect the detection criteria for
the recovery process. Alternatively, threshold crossings with a negative
threshold and the data moving
towards zero are viable. Again, the data point after the threshold is reduced
in absolute value towards but
not equal to zero. For these last two cases, synchronization for recovery
automatically occurs when
searching the data.
Although, in the exemplary, the detection criteria do not change between
degrading and
recovering the original digital data, this is not a requirement. The detection
criteria may change, if in a
known fashion, such that the recovery process uses a different (but known)
detection criteria than the
degradation process. In other words, box 420 (or 220, as discussed above)
would be different for the
degradation and recovery process.
The original content need not be represented by digital samples versus time,
as one may have
assumed. In many cases, such as using MPEG compression (i.e. MP3 audio), the
digital samples
represent subgroups of frequencies versus time. In this case the data may be
searched across frequency
for each subgroup, or across time for each frequency, or in any other but well
defined combination. The
data may also represent either the frequency magnitude or corresponding
scaling factors.
Additionally, there are alternative ways to recover the data while removing
most of the
CA 02364433 2001-08-08
WO 00/54453 PCT/US00/06296
perceptual degradation. For example, one could use a low-pass filter to
recover the data. The recovered
digital data will not exactly match the original digital data, but its
perception may be acceptable. As well
know by one familiar with the state of the art in DSP, filter characteristics
such as type and order will
affect the recovered data.
5 Alternatively, one could use pseudo-random sequences (a.k.a. a key) to set
the detection criteria
(box 220) or the adjustment or re-adjustment of the data (box 230). This
randomness increases the
difficulty to illegally recover the data. For example, a pseudo-random number
greater than zero but less
than one could be used as the scaling value N (box 430). Or, a pseudo-random
number between
minimum and maximum threshold could be used for the threshold (box 420). All
that matters is that the
10 degradation and recovery process use the same pseudo-random sequence.
However, this configuration
requires sending a key along with the data. The key may be embedded within the
data using known
techniques, such that the original data is still recoverable from the degraded
data.
Fig. 40 shows illustrative hardware used to implement the described
degradation and recovery
processes. The hardware includes a logic processor 500 and a storage unit 510.
The logic processor 500
15 may be defined as the equivalent of a digital signal processor (DSP),
general-purpose central processing
unit (CPU), or a specialized CPU, including but not limited to media
processors. A likely DSP chip is
one of the Texas Instruments TMS320 product line. A CPU could include one of
Intel's Pentium line or
Motorola/IBM's PowerPC product line. The design of code for controlling logic
processor 500 is simple
for someone familiar with the state of the art given the above pseudo-code and
description. The storage
20 unit 510 includes RAM when using a digital processor, and is required to
store the current buffer and/or
previous point(s) for the detection criteria.
In addition, a person familiar with the state of the art could implement the
logic processor 500
with analog and digital circuitry, either separately or in an application
specific integrated circuit (ASIC).
The analog and digital circuitry could include any combination of the
following devices: digital-to-analog
25 converters (D/A), comparators, sample-and-hold circuits, delay elements,
analog-to-digital converters
(A/D), and programmable logic controllers (PLC).
In accordance with another improvement to scrambling technology, a process is
provided the
avoids scrambling the header or other important information about the content.
An advantage to leaving
the header alone is that applications or devices can quickly read information
about the content before de-
30 scrambling and accessing the content. For example, with a scrambled MP3
file, a user can quickly learn
about the song's length, artist, resolution, etc., before choosing to de-
scramble and play it. Alternatively,
the header may contain copyright information that the player is required to
check before playing.
The scrambling process scrambles some or all of the non-header content. If
only some of the non-
header content is scrambled, it must be more than error correction, if
present, can repair. When the
35 content contains frames, each with its own header, as done with Motion
Pictures Expert Group (MPEG)
compressed audio or video, the header of each frame is avoided while
scrambling some or all of the non-
header content. The de-scrambling process recovers the original content from
the scrambled content,
similarly avoiding the header information.
CA 02364433 2001-08-08
WO 00/54453 PCTIUSOO/06296
41
An exemplary process involves using a pseudo-random noise (PN) sequence and
the XOR
function to scramble the content while avoiding the headers of each frame. The
de-scrambler is identical
since the inverse of the XOR function is the XOR function.
Again, a review of terminology may be in order. The header of a file contains
important
information about the file. This information may include the type of file,
author, place of origin, date of
origin, last modified date, file size, structure allocations, copyright codes,
unique IDs, usage rules, etc.
The header may exist only at the beginning of the file, at the beginning of
frames within the file, or both.
Frames are common for compressed digital media, such as MPEG audio and video.
More specifically,
with MP3 data, the header may include what the MPEG standard labels as header,
error correction and
side information. In addition, if the content does not contain frames or a
header, such data can easily be
created in a new structured file format.
Fig. 33a displays an overview of the scrambling process. If the file has only
a global header or
frames of known size without synchronization (sync) codes, the headers are
located and skipped (box
105) during the scrambling step (box 110). In other words, there is no reason
to look for the sync code
(box 100). The scrambling step may scramble part or all of the non-header
content. If the file is broken
into frames with additional sync codes, the sync codes that define the frames
are found (box 100), the
header information is skipped (box 105) and the content is scrambled (box
110). Usually, the header
contains information about the frame's size, which aids in locating the next
sync code, as the sync code
may also randomly occur within the data. Once again, the scrambling step may
scramble part or all of the
non-header content.
The scrambling step can consist of methods used in the prior art. Standard
modern encryption,
such as DES or RSA, is an excellent choice. With this encryption, although one
may be able to de-
scramble one file by brute-force, another file can remain secure even when
using the same key. Other
scrambling options may include simple mathematical operations with a PN
sequence, such as
multiplication, addition, subtraction, or exclusive-or (XOR). Division should
be used carefully since it
may cause bit error due to the imprecise nature of limited bit-length
division.
Fig. 33b displays an overview of the de-scrambling process. De-scrambling is
the inverse of
scrambling, and only the content bits that were scrambled should be de-
scrambled. If the file has only a
global header or frames of known size without synchronization (sync) codes,
the headers are located and
skipped (box 155) during the de-scrambling process (box 160). In other words,
there is no reason to look
for the sync code (box 150). If the file is broken into frames with sync
codes, the sync codes that define
the frames are found (box 150), the header information is skipped (box 155)
and part or all of the
remaining content in that frame is de-scrambled (box 160). Usually, the header
contains information
about the frame's size, which aids in locating the sync code, as the sync code
may not be unique and, thus,
occur within the data. The de-scrambling step may de-scramble part or all of
the non-header content,
depending upon what was scrambled.
The de-scrambler should use the inverse of the function used by the scrambler.
When scrambled
with standard modern encryption, the de-scrambler requires a decryption key,
and the key may be
CA 02364433 2001-08-08
WO 00/54453 PCTIUSOO/06296
42
different than the encryption key. For the mathematical operations,
subtraction and addition are inverses,
XOR is its own inverse, and division is multiplication's inverse. Division is
okay to use in de-scrambling,
since it is already known that there will be no remainder as the divisor was
the multiplier in the
scrambling process.
For both the scramble and de-scramble, the key is expected to remain the same
for many frames
and most likely for the whole content track, where a track can consist of a
song or movie. Thus, with
broadcasts, the key will probably be changing each track, and there are many
ways to send keys for
someone familiar with the state of the art in cryptology. When using PN
sequences, as described below,
the key for the PN sequence is the generator function, and it does not change
for each MP3 song, i.e.
defined as a track. Note that the generator function creates a random sequence
that is identical each time,
and is well know in the state of the art of cryptology. Alternatively, every
content track, such as a song,
may use one or more keys from a limited global list.
Fig. 34a displays the pseudo-code for an example of the scrambling or de-
scrambling process.
In this example, the content contains frames that begin with a sync code, and
a header exists for each
frame. Since the inverse of the XOR function is itself, the pseudo-code for
the scrambling and de-
scrambling process is identical.
Content scrambled or de-scrambled by this simple example could include MPEG
audio data, such
as Layer III (MP3) or AAC. MPEG audio's sync code is ' 1 1 1 1 1 1 1 1 1 1 1 1
' . The advantage of such
approaches are numerous. For example, portable players can quickly display
information about the
song's length, artist, resolution, etc., before a user decides to play the
song. Likewise, the header may
contain copyright information that the player is required to check before
playing.
The process begins at the beginning of the content (box 200). Then, the sync
code is found,
usually being the first few bits of the content (box 205). Next, the header
data is skipped, possibly
reading its own size from data after the sync code (box 210). Then, the M
content bits for that frame are
scrambled using an XOR operation with the M content bits and M bits of a PN
sequence (box 215). Fig
34b shows the input and output for the XOR function. Next, the content is
checked to see if another
frame exists (box 220). If another frame exists, the process continues at box
205 where the next sync
code is located. Usually, the size of the frame can be read from the frame's
header, which aids in
searching for the next sync code. If no content remains, the process is
complete (box 225).
In this example, the size of M determines the robustness to brute-force
attack, where the
attacker's purpose is to obtain the original content. The larger M, the more
robust the scrambled content
is to attack. However, the smaller M, the more efficient the scrambling and de-
scrambling processes can
be. M can be any number which is both greater than the number of bits that can
be repaired by error
correction, and less than the number of non-header content bits for that
frame.
The location of the M bits, which are scrambled in the frame, must be known
and contain bits
that are critical to the content's integrity. They could be the M bits after
the header. However, in MP3
data, the frame data may not start after the header. In this case, the
scrambled bits could be the first M
CA 02364433 2001-08-08
WO 00/54453 PCTIUSOO/06296
43
bits of the data for that frame. These bits determine the allocation of audio
bits and are critical to the
integrity of the file.
Fig. 35 shows hardware suitable for use in implementing the scrambling or de-
scrambling
process. The hardware includes a digital logic processor 300 and a digital
memory 310. The logic
processor performs the calculations and logic for this process. The logic
processor 300 may be defined
as the equivalent of a digital signal processor (DSP), general-purpose central
processing unit (CPU), a
specialized CPU, including media processors, or application specific circuitry
(ASIC). A likely DSP chip
is one of the Texas Instruments TMS320 product line. A CPU could include one
of Intel's Pentium line
or Motorola/IBM's PowerPC product line. The ASIC can easily be designed by
someone familiar with
the state of the art and the above pseudo-code and description. The design of
code for controlling logic
processor 300 is also simple for someone familiar with the state of the art
given the above pseudo-code
and description. The memory 310 includes RAM when using a digital processor,
and is used to store the
program and other necessary variables.
Conclusion
Having described and illustrated the principles of my technology with
reference to various
embodiments, it should be apparent that the technology can be modified in
arrangement and detail
without departing from such principles.
For example, while many of the embodiments employed watermark technology to
identify the
object or content, this may not be essential. Other marking techniques may be
employed in the
appropriate circumstances.
Similarly, while certain of the processes are detailed as being performed at a
certain location
relative to the user, the location of such processes is not generally
critical. That is, tasks can be allocated
among processing devices as best befits the context (provided security issues
are adequately addressed).
While reference has occasionally been made to applications involving images
and video, the
focus on exemplary audio applications may obscure this fact. Thus, it should
be emphasized that the
techniques detailed above are equally applicable to other forms of media
beyond audio.
While the detailed embodiment was described as changing the values of single
samples, in other
embodiments it may be desirable to change the values of plural neighboring
samples, e.g., to increase
durability of the watermark in the presence of corruption.
Similarly, while the detailed embodiments were described as embedding
auxiliary data into
content, the form of representation of the auxiliary was not much discussed.
In some embodiments a
payload of N bits may be encoded as M bits, where M>N (i.e. with partial or
complete redundancy). The
redundancy can include repetition of the N bits payload through the content;
BCH-. convolutional-, turbo-
, etc-coding of the N-bits to provide robustness and/or error correction; CRC
or ECC codes, etc.
While the detailed embodiments are systems of many parts, it will be
recognized that novelty
also resides in individual components thereof, and that such components can
also be employed in other
systems and devices.
CA 02364433 2010-08-09
WO 04/54453 PCT/FJS00/06296
44
The particular combinations of elements and features in the above-detailed
embodiments are
exemplary only, the interchanging and substitution of these teachings with
those of other embodiments.
In view of the wide variety of embodiments to which the principles and
features discussed above
can be applied, it should be apparent that the detailed technology is
illustrative only and should not be
taken as limiting the scope of the invention. Rather, I claim as my invention
all such modifications as
may come within the scope and spirit of the following claims and equivalents
thereof.