JP2002539723A - Method and apparatus for generating multiple watermarked copies of an information signal - Google Patents

Abstract

SUMMARY A method and apparatus for watermarking an information signal to generate a plurality of different watermarked copies of the information signal is disclosed. The information signal is segmented into information segments. Next, a plurality of different watermarked versions of each information segment are generated, and one of the watermarked versions for each segment is selected for each of the plurality of different copies to be generated. A different sequence of watermarked segments is generated for each copy.

Description

DETAILED DESCRIPTION OF THE INVENTION

The present invention generally relates to a method of watermarking an information signal to generate a plurality of different watermarked copies of the information signal.

[0002] When a provider of electronic information, such as audio, video, TV, images, and documents, desires to be able to identify when an unauthorized copy of the electronic information was made, the field of electronic information It is well known that watermarking can include information identifying the recipient of the information, for example, a unique ID that does not perceptually alter the quality of the electronic information. Further, watermarking techniques for electronic information To reduce the likelihood of being able to breach security labeling, the watermarking technique must not be obvious within the electronic information. There are many different schemes for performing watermarking of electronic information known in the art, and the technique used will depend, in part, on the type of electronic information. For example, one form of watermarking a document is to vary the space within the document according to a particular algorithm. In the case of audio data, the audio signal can be modified in an imperceptible manner to provide watermarking. For images compressed using JPEG compression techniques, the image can be watermarked by changing the compression parameters. This is the MP for the video image
The same applies to the EG compression parameter.

[0003] Conventional watermarking techniques provide multiple copies of electronic information to multiple recipients in order to ensure that the source of each copy created can be tracked if an unauthorized copy is detected in distribution. Needed. Each copy must be uniquely watermarked to identify the source of any unauthorized copies.

[0004] Electronic information is typically transmitted from a source of information to a recipient over a network. Examples of such networks are cable networks and the Internet or intranets. If the copy is transmitted specifically from the source to the recipient (ie, a unicast transmission), the source watermarks the electronic information transmitted to the recipient with information identifying the recipient. Thus, when an illegal copy is detected, the source of the copy can be determined from the watermark.

In networks, it is very common for information to be sent to multiple recipients at once, which is called a multicast transmission. Multicast transmission over a network is advantageous over unicast transmission in that the sender need only transmit a single copy that is transmitted to multiple recipients. However, because the source transmits only a single copy, it was not possible for each recipient to receive a uniquely watermarked copy of the electronic information using prior art multicast transmission.

[0006] In a multicast session over a network, the recipient conventionally subscribes to the transmission, so that the source can transmit the electronic information in an encrypted form. Upon receiving the electronic information, the subscriber is given a key to decrypt it. This provides security by allowing only the recipient with the key to decrypt the electronic information. This, however, does not prevent the creation of unauthorized copies from legitimately received and decrypted electronic information. However,
Being able to track the origin of any pirated copy is very beneficial. Since the Internet requires routers within the Internet to route electronic information, possible sources of unauthorized copying of electronic information include legitimate recipients,
Also included are network operators who have control over the routers through which electronic information passes.

The present invention provides a method for watermarking an information signal to generate a plurality of different watermarked copies of the information signal. The information signal is segmented into information segments to generate a plurality of different watermarked versions of each information segment. For each of a plurality of different copies to be generated, one of the watermarked versions of each segment is selected to generate a different sequence of watermarked segments for each copy. This sequence is probabilistically unique for each copy. This choice depends on the identity of the recipient of each copy and / or
Or it is controlled based on information about a location.

Thus, according to the invention, each copy of the information signal is watermarked using a unique sequence of watermarks. This has the advantage that the number of watermarked versions of each information segment need not be as large as the number of copies made.
This reduces the amount of processing required to generate a watermarked information signal for the recipient.

The present invention is applicable to unicast transmissions, but is particularly beneficial when used in multicast transmissions over a network. In such embodiments of the present invention, the intelligence available at the nodes in the network is used to perform the selection process. The source of the information signal produces a plurality of different watermarked versions of each segment of the information signal. These are transmitted to nodes in the network. Then, before passing the packet downstream, each node in the network depends on the location of the node in the tree and the path along which the information signal must travel to watermark each segment. Filter out one of the incoming versions. This selection is preferably performed in a pseudo-random manner, using information about network node locations and paths as keys for pseudo-random number generation.

In one embodiment, at the last node in the network in front of the recipient, the selection takes the form of a selection of only one of the watermarked versions of each segment to be transmitted to the recipient. , The receiver receives only one watermarked version of each segment. The selection process within this last node further depends on the information identifying the recipient.

Thus, in a network embodiment, at the network recipient node, multiple users or subscribers may be operating (ie, multiple recipients are present at physical locations in the network). So that information that uniquely identifies the recipient (user or subscriber), in addition to location information (ie, path information), can more accurately track the source of the illicit copy of the information signal. Needed to The use of both routing information and user identification information allows the recipient to be identified if the unauthorized copy originated from the recipient, and that the unauthorized copy originated from within the network. In that case, the router involved can be identified from the route setting information.

[0012] The number of different watermarked versions of each segment generated by the watermarking means depends on the number of nodes in the network. Each node in the network filters out at least one of the watermarked versions so that the recipient has at least two watermarked versions received by the last node in the network connected to it. To do so, the number of watermarked versions generated must be greater than the longest path between the source and the recipient, ie, the path with the largest number of nodes. Particularly in IP (Internet Protocol) multicast sessions, the longest path in a multicast session can change during the session, because the recipient can actively participate in and leave the multicast session. is there. Thus, information about the number of nodes in each path between the recipient and the source is monitored at the source so that the number of different watermarked versions of each segment of the information signal can vary as needed. I do. If, during the transmission of the information signal, the number of watermarked versions of the different segments changes, the point in the information signal at which the change occurred is the sequence of the watermarked version that appears in each recipient's copy of the information signal received. This must affect and must be recorded at the source because it is needed to track the source of the pirated copy.

According to one aspect of the invention, the source of the copy of the information signal can be detected by using the stored information regarding the identity of the recipient and the copy of the information signal. By applying a watermarking and selection procedure to the original information signal and comparing it to the copy to be examined, it is possible to identify the recipient from which the copy originated.

In embodiments where the selection process is performed by a node in the network, the information needed to identify the source of the copy includes information about the identity of the recipient and the path followed by that copy in the network. Information. This information is stored during transmission and, if necessary, used to track the source of the suspected unauthorized copy.

Next, an embodiment of the present invention will be described with reference to the accompanying drawings.

FIG. 1 shows the functional components of a generalized embodiment of the present invention. The information signal source 2 provides the information signal for which a plurality of unique watermarked copies are required. The information segmenter 3 segments the information signal from the information signal source 2 and inputs it to the watermarking module 4, which
Watermarking of the information segment will be performed. The watermarking module 4 uses the watermarking parameters stored in the watermarking data store 1 to
Generate multiple different watermarked copies of each segment. A plurality of different watermarked copies of each information segment are then received by the selector 5, which depends on the information about the recipient 7 stored in the recipient data source 6, for a plurality of receptions. A watermarked copy for each information segment for each of the parties 7 is selected. Thus, the selector 5 will generate M unique sequences of differently watermarked segments from a stream of information segments each having N different watermarked copies. An advantage of the present invention is that the number N of different watermarked sequences generated in the watermarking module 4 is:
Number M of unique watermarked copies of the information signal generated and received at receiver 7
That is, M> N.

The processing required in generating copies with multiple unique watermarks will therefore be greatly reduced since the watermarking operation will be performed a smaller number of times. Selecting a watermarked copy for each segment is a simple operation that is not computationally intensive compared to generating a watermarked copy. The present invention therefore provides a considerable reduction in processing requirements, especially when M >> N.

An analyzer module 8 is also provided to analyze unauthorized copies of the information signal so that it can identify recipients originating therefrom. The analyzer module needs to have a copy of the information signal received by the receiver 7. These copies are provided directly by the selector 5 or, using information from the receiver data store 6, the watermarked data store 1, and the information signal source 2, which information signals are received by the receiver 7 Can be reconfigured to predict what has happened. An input device 9 is provided to allow a suspected fraudulent copy of the information signal to be input to the analyzer module 8 for comparison with an expected copy of the information signal for the recipient. This comparison will determine from which recipient the suspected fraudulent copy originated.

Embodiment of the Present Invention An embodiment of the present invention will be described below with reference to FIGS.

[0020] This embodiment of the invention is based on IP (Internet Protocol),
Internet Protocol) applied to multicast transmission of information signals over a network.

As explained below, this embodiment of the present invention particularly benefits from the advantages of the present invention. This is because the processing required to select a watermarked copy of each segment is distributed among the nodes of the network. Thus, the source need only generate multiple N watermarked versions of each segment without actually generating a unique watermarked copy for each recipient,
Additional computational savings are provided. This embodiment to the present invention can achieve this due to the intelligence that can be made available inside routers in IP networks.

FIG. 2 shows that the source S is the receiver R where he subscribes to the multicast session. ₁ To R₇Network that multicasts information signals over a network
It is a schematic diagram of a work. Source S is located on local area network 10
And the network is connected to a node N in the network.₁(Router)
Interface. Node N₁Is two separate nodes N₂And N₃ Has an IP connection to Node N₂Is the two recipients R₁And R₂Against
Has another IP connection and another node N₄And node N_AAlso has a connection to
. Node N_AWith respect to another node (not shown) and the node N_BContact with
Node N_BIs connected to the local area network (LAN) 11.
-Face. In the LAN 11, a multicast session is subscribed.
There is no recipient.

Node N₃Is the node N₅Has an IP connection to node N₅If you are receiving
Person R₅IP connection to other nodes (not shown), and
Code N₆Has an IP connection to Node N₆Is the two recipients R₆And R ₇ Has an IP connection to other nodes (not shown)
You. Node N₄Has IP connections to four recipients, two of which R₃ And R₄Only are the subscribers of the multicast session.

In the multicast session of this embodiment, when a source wants to start a multicast session, a source path message (SPM) is transmitted to node N _1, which transmits the SPM to other nodes in the network. Broadcast to nodes. This allows the node to identify the path of the multicast session. For example, the node _{N 4,} there are two possible paths from the node _{N 2} and _{N 3.} However, IP connection to _{N 3,} for example, preferred for shorter, therefore, IP connection between the _{N 2} and _{N 4} is for that multicast sessions are dropped. Node On N _A and N _B, there is no subscriber requiring the reception of multicast information, therefore,
Node N _A and N _B are not play a role nothing in the multicast session. That is, they do not require that information signals be transmitted to them. LA
New subscribers on N11 is the case so as to request to join the multicast session, the node N _B is found path back to the requesting connection to the multicast session, thereby, connected to the node N _A, then Node N _A will connect to node N ₂ , thereby forming a path for the new recipient on LAN 11. Thus, the nodes in the network store information about the multicast session and the IP connection, allowing the recipient to join and join the multicast session, which opens a new path through the network. Can be accompanied.

The path followed by the information signal through the network is therefore
Can be drawn again as a tree structure. Nodes that are not in the communication path between source and recipient are irrelevant, as will become clearer below. At each level in the tree, nodes are assigned unique labels. Each level corresponds to several "hops" between the nodes. In FIG. 3, characters indicate the interface IP address. Thus, node 1 is labeled ab,
Node 2 is labeled abce, node 3 is labeled abdf, node 4 is labeled abdfkm, node 5 is labeled abdfer, and node 6 is labeled abdflrtv. I have. These labels are determined by the node from its location in the network using the source path message. With reference to the flow chart of FIG. 4, the manner in which the labels are determined will now be described.

FIG. 4 is a flowchart showing the steps performed at each router to set up the routers ready to receive and filter the information signal. A source path message (SPM) is transmitted by a source desiring to set up a multicast session. This is then received by the first hop router and propagates through the network from router to router as described below.

In step S 1, the router sends one or more source path messages (
SPM) and add the incoming interface IP address to it. When receiving an SPM from a plurality of routers, the router selects the SPM received first, that is, the one that provides the shortest path. In step S3, the router
Next, a label is determined from the received SPM and the SPM including the added incoming interface IP address. Next, in step S4, the outgoing interface IP address is added to the SPM, which is transmitted over that interface. Next, in step S5, the router determines whether there are any more interfaces that require the SPM message to be transmitted over it. If so, the process returns to step S4.

Each router includes a random number generator as described later with reference to FIG. In step S6, a random number generator is initialized for each interface using the router label and the interface IP address as keys. The interface address is the IP address of the next router or router on the interface to the recipient. If more than one router or receiver is connected to a router, a unique random sequence must be generated for the segments passing through each connection, and the pseudo-random number generator must Must be initialized with a unique key. Therefore, each IP address (ie,
For each interface), a separate pseudo-random sequence is generated. The IP address provides uniqueness indicating a path from the router, and the router label provides information identifying the router. The router is the last router in the route (the last "
Hop "router), and if one or more recipients are connected to it, the random number generator key is unique to the recipient, not just the router label and the IP address to which the recipient is connected. An ID, for example, a subscriber ID is also included. Next, in step S7, the router is ready to receive the information signal and performs a filtering (selection) procedure to determine which packets pass through the router using the random number generator as a selector. .

Thus, the last hop router stores not only the label but also the recipient ID information for the recipient connected to that router. This information, along with the source path message (SPM), is transmitted back to the source S and, if necessary, stored for later analysis to determine the source of the unauthorized copy of the information signal.

With reference to FIG. 5a, the operation of a router when passing a packet to another router will now be described.

In step S 10 A, the packet arrives at the router. Step S11
At A, the router waits until all of the multiple copies for the entire transmission group, ie, segment, have been received. When the entire transmission group has been received, in step S12A, one packet identified in the group by the random number generator is discarded. Next, the remaining packets in the transmission group are transmitted to the next router in step S13A. Therefore, the subsequent router will receive a smaller transmission group.

FIG. 5 b is a flowchart illustrating the steps for transferring data at the last hop router to the recipient. In step S10B, the packet arrives at the last hop router, which waits in step S11B for the entire transmission group to be received. When all packets in the transmission group have been received, one of the packets identified by the value generated by the random number generator is selected in the group, and in step S13B, the selected packet is received. Transmitted to the user. Thus, each recipient receives only one packet in the transmission group, ie, only one watermarked copy of each segment of the information signal.

An example of how an information signal is multicast over a network will be described with reference to FIGS.

FIG. 6 shows the data structure of a watermarked information signal generated at the source.
The transmission group T contains a plurality of different watermarked versions of this segment of the information signal. In this embodiment, five different watermarked versions A, B
, C, D, E, that is, N = 5. Each transmission group (1, 2, ... P
) Are transmitted sequentially. As can be seen in FIG. 6, this information signal is P
There are P transmission groups, ie, 5P packets, which are segmented into P segments.

As can be seen in FIG. 8, the data structure of FIG.₁By
Received. The random number generator uses the same label (1) for the node,
However, it is initialized using the interface IP address of each path,
Pseudo-run one of the packets in each transmission group to be filtered
Select by dam method. Therefore, node N₂And N₃Is a transmission group of length 4
Receive a loop. Node N₄And N₅Receives the transmission group of length 3, and
Node N away₆Receives a transmission group of length 2. Recipient connects to it
Node N₂, N₄, N₅, N₆In each case, the random number generator
To select one packet from each of the loops to be transmitted to the receiver,
Initialized using label, IP address, and recipient ID. Accordingly
Thus, each recipient receives only one packet for each segment. Recipient R ₁ FIG. 7 shows an example of a unique watermark sequence of a segment.

As seen in this example, each node in the network receives a transmission group of length ND, where D is the depth of the node on the path, ie, the transmission group is The number of "hops" before passing through the node. The number of watermarked versions of each segment received at the node with the recipient must be at least two, so that the number of packets in the transmission group is therefore relative to the longest path for the recipient , At least D. If N >> D, the number of segments required for the watermark segment sequence to be unique is reduced.

Although not shown in FIGS. 6 and 7, each transmission group will include a header indicating the number of segments in the transmission group. This is because as the number of packets in a transmission group decreases as this group traverses the network,
Changed continuously at each node. Each packet in the transmission group also contains a header identifying its sequence in the transmission group. Again, this needs to be updated as the packets are filtered out by the router, eg, the router receives A, B, C, D, E and If filtered out, E must be relabeled as the fourth packet in the transmission group.

Alternatively, a placeholder packet can be transmitted instead of a filtered-out (dropped) packet to indicate to a subsequent router that the packet was intentionally discarded. . This placeholder packet can be transmitted in conjunction with the next data packet, avoiding the need to transmit a separate packet.

A multicast IP session allows a subscriber to join and leave the session at any time, ie, while receiving an information signal. Also, the route followed by the information signal through the network may change, for example, when a router fails. So the source is
Updated information about the path followed by the information signal must be continuously received. This is achieved using a source path message (SPM). The SPM is broadcast periodically over the network so that the information
Received from the last hop router connected to the recipient, and provides routing information to the source. If the path information indicates that the maximum number D of routers in the path for the recipient is equal to or greater than the number N of different watermarked versions of each segment, then the source determines that each segment has a different watermark. The number of versions must be increased so that N> D. Also, the longest path is shorter,
Thus, if D decreases, the bandwidth required for the transmission of the watermarked information signal over the network can be reduced by reducing N while still maintaining the relationship N> D. Thus, the source monitors the network to monitor the route to the recipient. If the number of different watermarked versions of each segment is changed, the time at which it was changed, ie, the segment of the information signal at which the change was made, is stored. This will affect the sequence of the watermarked segment at each recipient.

With reference to FIG. 9, the manner in which new subscribers can join an IP multicast session will now be described.

In step S 20, the source assigns a unique ID to the subscriber and stores the subscriber information. The recipient gets its subscriber ID from the source. Step S
At 21, the new subscriber contacts the nearest router (last hop router) and informs this router of its unique subscriber ID needed to join the multicast session. Next, in step S22, the router sets S
If PM has not been received yet, wait for the reception. Next, the router performs step S
At 23, the source has a source path message (SPM) indicating the route to the router
, The subscriber ID that identifies the new recipient, and the ID of the first data packet sent to the recipient, ie, the time proof that the recipient then first started receiving a copy of the information signal (Identification of time). This information is stored at the source for later use to identify the source of the unauthorized copy of the information signal if needed.

At the source, in step S 24, the new depth d of the network for the new subscriber is determined from the source path message (SPM). Step S2
At 5, whether this new depth is greater than the current depth, ie, d>
D is determined. If so, in step S26, the source increases the number of different watermarked versions of each segment N by increasing the relationship N>
Then, in step S27, the last hop router receives the information signal and forwards it to the receiver. If it is determined in step S25 that the new depth is not greater than the current depth d used by the source to determine N, then in step S27, the last hop router receives the information signal And forward it to the recipient. Note that in this case, the last hop router before the recipient will not transmit the packet to the recipient unless it has received multiple copies to choose from. This is necessary to avoid having some recipients connected to the last hop router to receive the same sequence of watermarked segments. The last hop router can receive at least two different watermarked copies of each segment and use a pseudo-random number generator to randomly select from them for some recipients. I have to do it. These recipients can be connected to a router such that each recipient receives a unique sequence of watermarked segments.

In this embodiment, in addition to using different watermarked versions of each segment, conventional IP multicast security techniques such as encryption of the input information signal can also be used. The subscriber will be provided with a key that allows the subscriber to decode the information signal. The key for encryption is usually common to all recipients of the multicast session. Within the packet, the header used to identify the number of packets in the transmission group and to identify the sequence of packets in the transmission group is not encrypted. This is because the router needs to use the header information for the filtering process.

In fact, the tree topology together with the recipient ID is the secret used by the source to perform the watermarking. Participating routers must therefore reject requests that reveal any part of their topology. Even if several routers and clients collude, finding something useful would require collusion from the client to just before the source.

The selective discard function aims to provide the multicast router and its clients with the least possible degree of freedom to facilitate tracking after a rogue router or user. All routers in the tree have a persistent effect on the stream by dropping certain packets. Thus, a fraudulent router is identifiable because all users downstream from the router need to collude to reconstruct the watermark data passing through the router.

The probability decreases as one goes up the tree. The further down the tree, the easier it is to remove targets from the investigation. Introducing redundant data into the multicast stream is the first hop router, the number of packets per transmission group,
Increase the size of the stream, then decrease by one at the second hop router, and so on, the packet arriving at the receiver, where the traffic size is the same as the unwatermarked stream. is there. N is the group size,
If H is the maximum number of “hops” in the tree from the source to the recipient (ie, H = D + 1), this will increase the amount of traffic by the following factor.

[0047]

(Equation 1)

This is still much less than the traffic that would be generated by unicasting a unique version within each stream for all recipients. For N = H, which generates minimal additional traffic but takes a longer watermark sequence, this factor is:

[0049]

(Equation 2)

When the last hop router connects to a subnetwork (local area network) such as Ethernet, any host on that network can receive the same packet without any special effort. . Non-subscribers on the network can receive such packets, but do not have the necessary decryption keys to read them. However, 2 on the same subnetwork
One legitimate subscriber will receive the same watermarked data. However, this is usually not a problem. Because a multi-access subnetwork is
This is because it is usually under the administrative control of a single agency. Should one of the users of such a network illegally copy the information signal, it would be traceable to the institution controlling that network and in most cases this would be sufficient.

The source of this embodiment for the invention may include any suitably programmed computer with network access. With reference to FIG. 10, a schematic diagram of the structure of such a device will now be described.

All components of the device are linked by a computer bus 20. The computer device is a conventional BIOS for a computer system.
And a read-only memory (ROM). Random access memory (RAM)
21 is provided as working memory used by processor 23 during processing operations. Processor 23 accesses storage device 24 to load a program that implements watermarking application 23a, encryption application 23b, network monitoring application 23c, and analyzer application 23d. The watermarking application 23a performs segmentation and segment watermarking and uses the watermarking parameters stored in the storage device 24 in conjunction with information about the tree depth D to generate a plurality of different watermarks for each segment. , I.e., determine the number N different watermarked copies of each segment to be generated. The encryption application 23b performs conventional encryption of the watermarked segment for transmission over the network. The network monitoring application 23c provides information from the network regarding the route to the receiver, the ID of the receiver, information on the depth of the tree that can be searched from the source path message (path information), A monitoring function is performed to receive the time at which one segment was received, ie, the identification of the first segment received by the recipient. This information is then stored in the storage device 24. The processor 23 also implements a program module from the storage device 24 and an analyzer application 23d that analyzes the watermarked copy of the information signal to determine one or more recipients from which the copy was issued. Will also be implemented. This process is described in further detail below. A removable memory device 33 is provided to allow input of a copy of the information signal to be analyzed. Such a removable memory device 33 may be, for example, a floppy disk drive or a CDRO.
M. Alternatively, a copy of the information signal to be analyzed may be input over network 30 via network interface 25.

The information signal to be multicast can be obtained from an information source in the form of a database 26, or it can be obtained in real time via an input / output device 27 connected to a video camera 28 in this example. can do.
Therefore, the information signal input from the input / output device 27 or the database 26 is transmitted to the watermarking application 23 executed by the processor 23.
a, and encrypted by the encryption application 23b, and then multicast by the network interface 25 over the network 30.

The computing device also comprises a conventional display 29, a keyboard 31, and a pointing device 32 that allow a user to interface with the computing device.

Thus, in the source, the storage device 24: 1. Subscriber ID and 2. the time at which the subscriber joined (or left) the multicast session; 3. the route to the subscriber; 4. N for each segment; It stores the watermarking scheme and watermarking parameters used for each packet.

The copy of the multicast information signal is stored in the storage device 24 or
Alternatively, a reference to that storage location in database 26 is stored.

Other parameters stored in the storage device 24 include a database of legitimate users that enables the request to be validated when a request is made during a multicast session, and encrypts the multicast information signal. This is data about the encryption scheme used to encrypt. The watermarked segment is temporarily stored in the RAM 21. Instead of storing the watermarking scheme, the parameters used for each packet, and a copy of the original information signal in storage 24,
The watermarked segment can be stored. The storage device 24 also stores a multicast address used for one or more multicast sessions and transmission protocols. Further, if the information signal is compressed or encoded, information about the compression or encoding scheme is stored.

With reference to FIG. 11, a functional diagram of the functional components of the router 40 used in the network 30 for switching the information signal via the router in the network will be described next. Many of the functions shown and described below are performed by processors that operate in accordance with processor instructions.

The router 40 has a first interface 41 for receiving data from another router or a source further up the tree. Also, the SPM is received and transmitted. The received data is passed through a filter module 42 for filtering the data to remove at least one of the packets in the transmission group. The filter module 42 operates under the control of a random number generator (RNG) 43. The random number generator 43 is initialized using the parameters stored in the volatile memory 44. The volatile memory 44 stores the label for the router, the IP address for each interface, and the current state of the random number generator 43. The data is therefore filtered by the filter module 42 and output to the second interface 45 and the third interface 46.
The SPM is received by the interface 41 and utilized by a practical pragmatic general multicast (PGM) protocol engine 47. The PGM protocol engine 47 adds the IP address of the router 40 to the SPM, and returns the SPM to the source via the interface 41.
This engine also outputs its S for output to the next router or recipient.
The PM message is also passed to the second interface 45 and the third interface 46. The router 40 also includes a subscription manager 48 for managing subscriptions by recipients. Thus, if any router 40 is the last hop router in the path and has a recipient connected to it, the join message (SM) will be sent when the recipient wishes to join the IP multicast session. Sent to the respective interface 45 or 46 and then to the subscription manager 48
Passed to. The join message includes the subscriber ID, which is stored in volatile memory 44 for use in addition to the router label and interface IP address for initialization of the random number generator 43 for the interface to the recipient. Is done. Also, the subscriber ID in volatile memory 44 is passed back to the source and returned. Join messages also include requests to join or leave a multicast session.

In the embodiment of FIG. 11, the PGM protocol can be replaced by any suitable protocol, for example, by Multicast Transport Service (GMTS).

Referring to the flowchart of FIG. 12, an analysis method for analyzing a watermarked copy of an information signal to identify its source, ie, to identify the router or receiver that has received the sequence of watermark segments, will now be described. Will be described. In this flowchart, a sequence of packets is sequentially analyzed to determine, at each point in the sequence, which recipient may have received the packet. The "criminals" of the group are therefore determined. If there is a direct match between the expected sequences for the subscriber or router, it is clearly the most likely culprit.

In step S30, a packet is selected, and in step S31,
All recipients of the packet, both the router and the user, are marked as suspect. This list of suspects is reduced during subsequent steps in an attempt to predict the most likely suspect.

In step S 32, the operation starts by predicting the operation of the first hop router. In step S33, the next interface on the list of suspects at the router is considered, and
Determines whether a random number generator has been initialized for this interface of this node. If not initialized, step S
At 35, the random number generator checks its router label and interface ID (
And, in the case of the last router, the random number generator is initialized with the subscriber ID). Next, in step S36, a random number generator is executed to determine a packet sequence.
Next, in step S37, the router determines whether the packet has been transmitted on this interface. If yes, in step S38, it is determined whether a router exists at the end of the interface. If there is, the process moves down the tree to the router at the end of the interface in step S39, and returns to step S33. If it is determined in step S38 that the receiver is at the end of the interface, the process proceeds to step S41.

If it is determined in step S37 that the router did not transmit the packet, the interface and all its children are removed from the suspect list in step S40. Next, in step S41, it is determined whether there is a suspicious interface remaining on the router. If there is, the process returns to step S33. If not, it is determined in step S42 whether the router under consideration is the first hop router. If it is not the first hop router, then the parent router is considered in step S43, and the process returns to step S41. In this manner, the tree can be traversed and returned to the first hop router by continuously looping in steps S41, S42 and S43.

If it is determined in step S42 that the router under consideration is the first hop router, the entire tree has been scanned for the selected packet, and therefore, in step S44, Determine if this is the last packet. If it is not the last packet, the process proceeds to step S3
Returning to 0, it is repeated for the next packet. If all of the packets have been considered, then in step S45, the union of the set
Is found as the most likely criminal group.

This method will identify multiple culprits if the information signal is spliced from sections from different recipients attempting to break through the watermarking security measures. Since the comparison is made on a packet-by-packet basis, the recipients who collude and generate the illicit copy will be identified.

This technique allows for the identification of any router, not just the recipient, from which the pirated copy may have originated.

Other Embodiments While the embodiments of the present invention have been described with reference to IP multicast over the Internet or an intranet, the present invention is not limited to any form, such as a cable network having some intelligence within a network switch. It can be applied to other networks.

In IP multicast, the protocol allows a maximum of 32 nodes in the router, and thus the depth D is limited. This limits the number of different watermarked versions of the segment that need to be generated and transmitted over the IP multicast system. However, the invention is applicable to any network, and the benefits from the invention are obtained as long as the number of nodes in the network is less than the number of recipients.

The present invention is particularly suited for multicast transmissions over a network where the selection process is performed within the network, but also offers advantages for unicast transmissions. Because it reduces the processing required by the information signal source. Less watermarked copies of the information signal need to be generated, and a unique sequence of watermarked segments for each user may be provided by a simple selection or filtering process.

A further advantage of the present invention may furthermore be provided, in fact, in non-networked systems where D = 1. This technique again benefits from a reduction in the number N of watermarked copies that must be generated.

The claims state that the system is implemented by separate means, any of which may be combined in an embodiment of the system. For example, the present invention is ideally suited for software implementation on a general purpose computer. Thus, the segmenting means, the watermarking means and the selecting means can be realized as instructions for controlling the processor. In a network embodiment, the segmenting means and the watermarking means are implemented by software at the source computer, and the selecting means is implemented by software in each of the routers in the network. Since the present invention can be implemented as software for controlling a multipurpose computer, a CD-ROM or a floppy (registered trademark) disk in which instructions in the form of computer code for controlling a processor that executes the processing are stored. , A solid state memory device, or a storage medium such as a tape. Further, the computer code can be transmitted over a network and installed on a computer so that it can implement the present invention;
Thus, the invention is also implemented as a signal carrying computer code.

The watermarking techniques used to form different watermarked versions of each segment may include the same watermarking technique using different watermarking parameters or different watermarking techniques. Different watermarking techniques can be used as long as they do not appreciably affect the data.

In the embodiment of FIGS. 5 a and 5 b, the router waits for the reception of the entire transmission group before filtering, but the filtering or selection process need not wait for the reception of the entire transmission group.

Furthermore, the selection process is not limited to the use of a pseudo-random selection procedure, but can use any procedure that will result in multiple deterministic sequences keyed on recipient information.

Although specific embodiments of the present invention have been described above with reference to the drawings, the present invention is not limited to such embodiments, and those skilled in the art It will be apparent that modifications may be made which fall within the spirit and scope of the invention.

[Brief description of the drawings]

FIG. 1 is a functional diagram of the present invention.

FIG. 2 is a schematic diagram illustrating an example of a network in which a source S multicasts an information signal therefrom to a plurality of recipients R ₁ to R ₇ .

FIG. 3 is a schematic diagram showing a network tree structure of the network of FIG. 2;

FIG. 4 is a flowchart illustrating steps performed by a router in a configuration of a network to perform a selection process according to an embodiment of the present invention;

FIG. 5a is a flowchart illustrating steps performed by a router in a network to pass a packet on another router, according to an embodiment of the present invention.

FIG. 5b is a flowchart illustrating steps performed by a final router to pass a packet to a recipient, according to an embodiment of the present invention.

FIG. 6 is a diagram illustrating a structure of watermarked data generated by a source according to an embodiment of the present invention;

FIG. 7 is a diagram illustrating an example of a data structure received by a receiver according to an embodiment of the present invention;

FIG. 8 is a diagram illustrating a tree structure of the network of FIG. 3, illustrating a flow of a packet through a network to a receiver according to an embodiment of the present invention;

FIG. 9 is a flowchart illustrating steps performed when a new subscriber desires to join a multicast session, according to an embodiment of the present invention.

FIG. 10 is a schematic diagram illustrating a structure of a source device for an information signal according to an embodiment of the present invention;

FIG. 11 is a schematic diagram illustrating a router according to an embodiment of the present invention.

FIG. 12 is a flowchart illustrating an analysis method according to an embodiment of the present invention.

──────────────────────────────────────────────────続 き Continuation of front page (81) Designated country EP (AT, BE, CH, CY, DE, DK, ES, FI, FR, GB, GR, IE, IT, LU, MC, NL, PT, SE ), OA (BF, BJ, CF, CG, CI, CM, GA, GN, GW, ML, MR, NE, SN, TD, TG), AP (GH, GM, KE, LS, MW, SD, SL, SZ, TZ, UG, ZW), EA (AM, AZ, BY, KG, KZ, MD, RU, TJ, TM), AE, AL, AM, AT, AU, AZ, BA, BB, BG, BR, BY, CA, CH, CN, CR, CU, CZ, DE, DK, DM, EE, ES, FI, GB, GD, GE, GH, GM, HR, HU, ID , IL, IN, IS, JP, KE, KG, KP, KR, KZ, LC, LK, LR, LS, LT, LU, LV, MD, MG, MK, MN, MW, MX, NO, NZ, PL, PT, RO, RU, SD, SE, SG, SI, SK, SL, TJ, TM, TR, TT, TZ, UA, UG, US, UZ, VN, YU, ZA, ZW (72) Invention Ian Brown United Kingdom EN 26 3 AF Tin and Weir White Ray Bay Holywell Avenue 98 F Term (Reference) 5B057 AA20 CE08 CG07 CH18 5C063 AB07 AC01 CA23 CA29 CA34 CA36 DA13 5C076 AA14 BA06 5J104 AA14

Claims (69)

[Claims] 1. A method for watermarking an information signal to generate a plurality of different watermarked copies of the information signal, the method comprising: segmenting the information signal into information segments; Generating a watermarked version; selecting, for each segment for each of a plurality of different copies to be generated, one of said watermarked versions to provide a different, differently watermarked copy for each copy; Generating a sequence of segments of the sequence. 2. In the selecting step, using information about the recipient of each copy, a sequence of different watermarked segments of each copy is performed depending on the recipient of the copy. The method of claim 1, wherein: 3. The method of claim 2, wherein the step of selecting is performed in a pseudo-random manner based on recipient identification information. 4. The information about the recipient includes at least one for the recipient.
4. A method according to claim 2 or 3, including one unique identification and information about the location of the recipient. 5. A copy of the information signal is transmitted to the recipient over a network and is performed in the selecting step depending on the path taken by each copy to reach the respective recipient. A method according to any one of the preceding claims. 6. The network includes a plurality of switching means for switching a path followed by information within the network, wherein a plurality of different watermarked versions of each information segment are generated at an information source, The switching means executing the step of selecting by selecting at least one watermarked version for each segment not transmitted further from the received watermarked version for each segment that is input to the network. The method of claim 5, wherein: 7. In a network, the switching means, to which the other switching means is connected, selects one of the received watermarked versions for each segment not transmitted to the other switching means. The method of claim 6, wherein: 8. The switching means to which one or more recipients are connected, wherein said switching means receives a watermarked version for each segment transmitted to each recipient connected to said switching means, for each segment. The method according to claim 6 or 7, wherein one of the watermarked versions is selected. 9. The method according to claim 6, wherein the number of different watermarked versions of each segment generated is dependent on the maximum number of said switching means in the path between a source and said recipient. The method according to any one of the preceding claims. 10. The number of different watermarked versions of each segment is:
The method of claim 9, wherein the number of the switching means in a path between a source and the recipient is greater than a maximum number. 11. The number of said switching means in a path from a source to a receiver is monitored, and the number of different watermarked versions of each segment is changed depending on the monitored number of said switching means. The method according to claim 9, wherein the method is performed. 12. The switching device according to claim 6, wherein each of the switching devices transmitting to another switching device performs selection based on a position of the switching device in a path from a source to a receiver. The method according to any one of the preceding claims. 13. A method according to claim 6, wherein each of said switching means for transmitting to the recipient makes a selection based on unique identification information for each recipient. 14. The method according to claim 6, wherein each of the switching means transmits information on a position of the switching means in a path between a source and a receiver to the source. The described method. 15. The method of claim 13, wherein each switching means for transmitting to a recipient receives information about the identity of the recipient from the recipient and transmits the information to the source. 16. A method as claimed in any preceding claim, wherein each watermarked version of each segment is watermarked using the same watermarking technique and a different watermark is applied. Method. 17. The method as claimed in claim 1, wherein each watermarked version of each segment is watermarked using one of several possible watermarking techniques. The method described in. 18. A system for watermarking an information signal to generate a plurality of different watermarked copies of the information signal, comprising: segmenting means for segmenting the information signal into information segments; Watermarking means for generating different watermarked versions of each of the plurality of different copies to be generated, for each segment, selecting one of said watermarked versions to provide a different Selecting means for generating a sequence of different watermarked segments. 19. The selecting means performs the selecting step using information about the recipient of each copy of the information signal to perform different watermarking of each copy depending on the recipient of the copy. The system of claim 18, adapted to generate a sequence of segments. 20. The system of claim 19, wherein said selection means is adapted to perform said selection in a pseudo-random manner based on recipient identification information. 21. The selection means is adapted to perform the selection using at least one of unique identification information for the recipient and information regarding the location of the recipient. 21. The system of claim 19 or claim 20. 22. The selecting means receives different watermarked versions of each information segment and differs for each copy, depending on the path each copy takes to reach its respective recipient. 22. The system according to any one of claims 18 to 21, comprising a node in the network configured to generate a sequence of different watermarked segments. 23. Each node in the network is configured to control the routing of an information signal in the network, and the segmenting means and the watermarking means are connected to a first node in the network. Provided at a source, wherein the first node is adapted to receive a plurality of different watermarked versions of each segment, the node comprising: 23. The system of claim 22, wherein the system is adapted to select at least one watermarked version for each segment. 24. The node to which the other node is connected is adapted to select one of the received watermarked versions for each segment not transmitted to the other node. 24. The system of claim 23, wherein: 25. The node to which one or more recipients are connected, wherein the node selects one of the received watermarked versions for each segment to be transmitted to each recipient connected to the node. The system according to claim 23 or 24, wherein the system is adapted to: 26. The method according to claim 23, wherein the watermarking means generates a different watermarked version of each segment of a number depending on the maximum number of nodes in the path between the source and the receiver. 26. The system according to any one of claims 25. 27. The watermarking means is adapted to generate a different watermarked version of each segment between a source and a receiver, the number being greater than the maximum number of nodes in any path. 3. The method according to claim 2, wherein
7. The system according to 6. 28. A system comprising monitoring means for monitoring the number of said nodes in a path between a source and a receiver, wherein said watermarking means is responsive to said monitoring means for monitoring said monitored number of said nodes. 28. The system according to claim 26 or claim 27, wherein the number of different watermarked versions for each segment to be generated is changed according to the number. 29. The node for transmitting to another of the nodes is adapted to make the selection based on a location of the node in a path from a source to a recipient. 29. The system according to any one of the preceding claims. 30. The node according to claim 23, wherein the node transmitting to the recipient is adapted to make the selection based on unique identification information for the recipient. System. 31. The method according to claim 23, wherein each of the nodes is adapted to transmit to the source information regarding the location of the node in a path between the source and the recipient. The system according to paragraph. 32. Each of the nodes transmitting to a recipient is adapted to receive information regarding the identity of the recipient from the recipient and transmit the information to the source. The system of claim 30. 33. The method according to claim 18, wherein the watermarking means is adapted to watermark each version of each segment using the same watermarking technique. System. 34. The watermarking means, wherein the watermarking means is adapted to watermark each version of each segment using one of several possible watermarking techniques. The system according to any one of claims 18 to 32. 35. An apparatus for watermarking an information signal, comprising: segmenting means for segmenting the information signal into information segments; and watermarking means for generating a plurality of different watermarked versions of each information segment. An apparatus comprising: 36. An interface means for connection to nodes in a network, wherein a recipient requesting a copy of an information signal is connected to a node in said network, said node in said network being connected to each recipient. Selecting one of the watermarked versions for each segment for each recipient to generate a different sequence of watermarked segments, different for each recipient, the apparatus comprising: Storage means for storing information relating to a path followed by transmitting a copy of a person, wherein such an interface is adapted to receive said information over a network. An apparatus according to claim 35. 37. The apparatus according to claim 35, wherein said watermarking means is adapted to store parameters used to watermark each version of each segment. 38. The interface means is adapted to receive, from a network, information indicating when the recipient has begun receiving a copy of an information signal and to store the information in the storage means. 37. The method according to claim 36, wherein
Or the apparatus of 37. 39. The interface is adapted to receive information identifying a number of nodes in a network between the device and each recipient, and wherein the watermarking means comprises a means for communicating between the device and a recipient. 39. The method according to any one of claims 36 to 38, wherein the method is adapted to set a number of different watermarked versions of each segment depending on a maximum number of identified nodes. apparatus. 40. The apparatus according to claim 39, wherein said storage means is adapted to store the number of watermark versions used for each segment. 41. The storage device according to claim 1, wherein the storage means stores information for identifying a storage location of the information signal, or is adapted to store the information signal. An apparatus according to claim 1. 42. Receiving a copy of an information signal having a sequence of different watermarked segments to be tested and using the information in said storage means to compare with said sequence of different watermarked segments to be tested. 42. The method of claim 35, further comprising the step of: calculating a sequence of different watermarked segments for the recipient and determining if a match exists between at least a portion of the sequences. The device according to item. 43. A method for watermarking an information signal at a source, comprising: segmenting the information signal into segments; and generating a plurality of different watermarked versions of each information segment. A method comprising: 44. Information about the recipients of the copy of the information signal and a watermark for each segment, for each recipient connected to a node producing a different sequence of watermarked segments for each recipient. One of the included versions
44. The method of claim 43, comprising storing information about a path taken to transmit a copy to each recipient over a network having a node to select one. 45. The method of claim 44, comprising storing parameters used to watermark each version of each segment. 46. The method of claim 44 or 45, comprising receiving information from a network indicating when the recipient has begun receiving a copy of the information signal and storing the information. Method. 47. receiving from a network information identifying the number of nodes in the network between a source and each recipient; and depending on a maximum number of nodes identified between the source and the recipient. 47. The method of any of claims 44 to 46, comprising: setting a number of different watermarked versions of each segment. 48. The method of claim 47, comprising storing a number of versions of the watermark to use for each segment. 49. The method according to claim 48, further comprising the step of storing information for identifying a storage position of the information signal or storing the information signal.
The method according to any one of the preceding claims. 50. Receiving a copy of an information signal having a sequence of different watermarked segments to test, and using the information in said storage means to sequence the different watermarked segments to test. 50. The method of claim 43, further comprising calculating a sequence of different watermarked segments for the recipient as compared to and determining whether a match exists between at least a portion of the sequence. The method described in all items. 51. Receiving means for receiving a plurality of different watermarked versions of each of a plurality of information segments forming an information signal, and selecting at least one watermarked version for each segment not transmitted further. Apparatus according to any one of claims 6 to 15, comprising: selecting means; and transmitting means for transmitting the watermarked version or each remaining watermarked version for each segment over a network. . 52. The apparatus according to claim 50, wherein said selection means is adapted to control said selection using information regarding a location of a switch device in a network. 53. The method according to claim 5, wherein the selection means is adapted to control the selection using information about a path to be used for transmission.
3. The device according to 2. 54. The selection means is adapted to control the selection using information identifying the recipient to select only one watermarked version for each segment to be transmitted further. Wherein the recipient is the ultimate recipient of the information signal, and wherein the transmitting means is adapted to transmit the selected watermarked version for each segment to the recipient. 54. The apparatus according to any one of claims 51 to 53. 55. The selection means includes a pseudo-random number generator and is adapted to perform the selection in a pseudo-random manner, wherein the information is used as a key to initialize the pseudo-random number generator 55. Apparatus according to any one of claims 52 to 54, characterized in that: 56. The receiving means is adapted to receive a message indicating a path that an information signal will follow, and the transmitting means adds an identity of a device in the path. 56. The apparatus according to any one of claims 51 to 55, wherein the apparatus is adapted to retransmit the message. 57. A method performed at a node in a network, comprising: receiving a plurality of different watermarked versions of each of a plurality of information segments forming an information signal; A method comprising selecting at least one watermarked version for each segment, and transmitting the watermarked version or each remaining watermarked version for each segment over a network. 58. The method of claim 57, wherein the step of selecting controls the selection using information about a path to be used for transmission. 59. The step of selecting, wherein the selecting is controlled using information identifying the recipient to select only one watermarked version for each segment to be transmitted further, 60. The method according to any one of claims 57 to 59, wherein is the final recipient of the information signal and the selected watermarked version for each segment is transmitted to the recipient. 60. A method according to claim 58, wherein a pseudo-random number generator is used for said selection and said information is used and controlled as a key for initializing said pseudo-random number generator. 60. The method according to any one of claims 59. 61. Receiving a message indicating a path to be followed by an information signal, and retransmitting the message with identification of a node in the path. 61. The method according to any one of 7 to 60. 62. Receiving means for receiving a copy of a watermarked information signal including a sequence of different watermarked segments; storage means for storing information about a recipient of the copy of the information signal; and accessing the information signal. Access means for determining a watermarked version for a segment for a recipient using the accessed information signal and information about the recipient; and a watermarked information for a watermark for a segment for the recipient. 35. A system as claimed in any one of claims 19 to 34, comprising: matching means for matching against a segment of a received copy of the signal to identify a recipient of the copy of the watermarked information signal. An apparatus for identifying a recipient of a copy of a watermarked information signal using the same. 63. The storage means is adapted to store information regarding nodes in a network forming a path between a source and a receiver of an information signal, and wherein the determining means uses the path information to 63. The apparatus of claim 62, wherein the apparatus is adapted to determine an identity of the recipient. 64. Receiving the copy of the watermarked information signal including a sequence of different watermarked segments; reading stored information about a recipient of the copy of the information signal; Using the accessed information signal and the read information about the recipient to determine a watermarked version for the segment for the recipient; Identifying the recipient of the copy of the watermarked information signal against a segment of the received copy of the watermarked information signal to identify a recipient of the copy of the watermarked information signal. Identify the recipient of the watermarked copy of the information signal using the described method. How to. 65. Reading information about nodes in a network forming a path between a source of information signals and a recipient, and using the path information to determine the identity of the recipient. 64. The method of claim 63, wherein: 66. The method according to claim 1, comprising a plurality of information transmission groups, each information transmission group comprising a plurality of different watermarked versions of the same information packet. Information signal watermarked using the method. 67. A medium for carrying the information signal according to claim 66. 68. Controlling a processor to carry processor-executable instructions for performing a method according to any one of claims 1 to 17, 43 to 50, or 57 to 61. Characteristic signal. 69. A method for controlling a processor to carry processor-executable instructions for performing a method according to any one of claims 1 to 17, 43 to 50, or 57 to 61. Conveyance medium characterized.