GB2492939A - Digitally watermarking media files and tracking to identify original recipient - Google Patents

Abstract

In what is sometimes referred to as â traitor tracingâ , an arrangement is disclosed for tagging (figure 1), delivering and tracking (figure 6) copies of media files. One or more copies of a media file are digitally watermarked to uniquely identify the original recipient of a watermarked copy. Following delivery of watermarked copies of files to recipients, an online tracking search, preferably automated, is performed. This forensic search is carried out to detect watermarked copies of files and to determine the respective original recipient of a detected watermarked copy. Implicitly, it is the original recipient who must have made the detected copy available and the purpose of the invention is to help protect the rights of the originator against illegal file sharing.

Description

Media tagging, tracking and delivery system This invention relates to the unique tagging and tracking of audio and video.

When video and audio is currently distributed through various different transport mechanisms such as the internet, CD, USB memory stick, hard drive, digital and analogue discs keeping track of each item is very difficult without affecting the experience for the intended recipient. This is nowhere more of an issue than the online illegal sharing of music and video which is currently costing the creative industries many millions of pounds per year.

To overcome this, the present invention proposes to uniquely watermark each copy of the audio whether it is on it's own or part of a video in an efficient scalable way and to search the internet for possible watermarked (protected by the system) copies. The invention provides the ability to decode the watermark within the audio of a supplied file either from the internet or uploaded from another source (such as USB stick, CD, hard drive or any other medium capable of carrying audio or video) and to pin point the person that the copy was originally delivered to thus identifying the responsible party for making it available through the internet or other transport means.

The invention provides all the tools necessary to protect copies of the audio and video files within it's protection. The delivery of the files which is discussed in more detail later in this section creates a record of every transaction against each unique watermark. The invention searches many online sites known for giving access to copyrighted files and is able to alert the content owner to each suspected occurrence. The content owner is then able to load the suspected file indicated or indeed a file from any other source into the system. In some cases the system will be able to automatically download the suspected file to process making the system completely automatic. The system then decodes the watermark extracting the unique reference if found. By matching this with a delivery transaction record in the database it is able to feedback to the content owner the details of the end user delivered the item originally that were stored at the time of the transaction such as name, email address, phone number along with the file name, type, time it was delivered, the IP (Internet Protocol) address, the ISP (Internet Service Provider) used and the geographical location assigned to theiP address for that user. This detailed report is conclusive proof that this is the responsible party who originally received that copy of the audio or video file.

There are two main ways to control the management and delivery of the files within it's protection.

Either through the online web interface or though the API (Application programming interface). All of the features of the system are available through each different interface.

The architecture of the system uses a number of distributed servers which are each capable of securely watermarking the audio (and thus the video files) and delivering it to the end user either directly through a download or stream or indirectly through a third party (usually the clients) website, application or duplication system. Each watermarking server is able to pass the request on to the centralised watermarking server should it need to. This occurs if there is no copy of the requested file ready to serve which ensures a quick response to the download requests.

The watermarking servers each feature an algorithm to manage the caching of the audio and video files as well as to manage the processing queue efficiently. This ensures quick deliveries by pre empting the requirement for specific files as well as keeping the servers resource usage optimal.

Advantages * No special player or software is required to play the watermarked files other than that used normally to play that medium format.

* Able to integrate easily with any existing online delivery system using the API.

* Can generate copies for use in duplication systems.

* Very robust watermarking able to withstand repeated processing, audio signal degradation and even passing through an air transducer.

* Scalable to suit any application.

* Secure to the level of credit card company requirements (PCI compliant).

* Automatic searching for potential copies of protected content.

* Accurate identification of the original recipient of the copy of the content.

* Compatible with any format of audio and video * Very fast download/streaming delivery and high availability due to built in edge delivery architecture ensuring every request is dealt with by the nearest node in the network.

An example of the invention will now be described by referring to the accompanying drawings.

Figure 1 shows how an unprotected audio file is uploaded into the system which then creates watermarked copies of the original file. The system creates unique watermarks for each copy of the file and is able to output the file in many different formats including mp3, mp4, way, aiff. The unique watermark is recorded in the database against the details of the person who has downioaded that copy. If the user downloads multiple copies each copy will have a different watermark allowing the system to be able to pinpoint the time, location, IP, ISP of every download.

The drawing in Figure 1 shows 2 examples of how the downloads may be delivered. Two of the downloads occur through the direct download interface which is described in more detail in Figure 2.

The third download is done through a third party site such as a digital music store or a music promotion system which might be delivering music for promotional purposes. The download in this case is done using the API which records the details of the recipient to the copy of the file and delivers the file. The API can be used in 2 different ways. Either delivering on demand or pre-emptive caching which is covered in Figure 3.

Figure 2 shows how the system can be used to deliver direct download via an email to the intended recipient. An email is generated and sent to the user containing a list of the tracks they have been sent as well as a supporting message which can include images. The email contains a unique link which is a secure web address to a download page. On clicking this the user is shown a secure webpage listing the tracks. The page allows the user to select to download each track in a specific format or to securely stream the track without downloading. The stream is delivered through a flash player allowing the user to seek easily through the tracks timeline without having to download or buffer the whole track.

Figure 3 shows that the API can be used to download and store local copies of the files ready to deliver to users of the third party site. This is a great way to allow the third party site to manage their bandwidth and speed of delivery by creating their own caching rules. Then API is then used to update the record when a file is delivered to an end user associating the details of that user against the copy delivered.

Figure 4 shows how rather than caching locally or passing the data stream for the file delivery through a third party site the user can be presented with direct links to the watermarking download and streaming servers. These act as edge servers in a content delivery network which means the downloads are always delivered from the nearest server in the network to the user and therefore the fastest. This means that the third party site benefits from the security of using the invention and no longer has to provide the download bandwidth for delivering the files to the user.

Figure 5 shows how the system caters for scalability with the use of edge servers'. In the invention they are each able to watermark files, stream and deliver direct to the end user or through the API.

As each is able to deliver a large number of files an hour scaling the system is a matter of establishing more of these edge servers. As the user clicks the download link either in an email or on a third party site their browser will request the IP address for the domain name which will be done by the DNS servers. These will use the location of the user, the load on the edge servers and other static routing factors to decide which edge server to connect the user to. This configuration also makes it easy decommission edge servers for maintenance by simply blocking them in the DNS servers.

By delivering from the nearest server in the network makes the connection time very fast and should avoid the majority of the congestion on the internet. This is mainly achieved by the setting up of peer connections to lSPs and network providers at the location of each edge server. In many cases this will mean that files are delivered direct to the user through their ISP without using a carrier in between. In addition using the peer technique ensures a high level of resilience even with multiple network failures.

Figure 6 shows how the system searches a large number of sites, file sharing systems, blogs and forums for any download links matching any of the protected items in the system. Should a new instance be detected then an alert is sent to the content owner notifying them of the new instance.

The content owner can then download the file and use the testing tool to see if the file is a copy delivered through the audio protection system. In some instances the file will be downloaded automatically and checked for a watermark providing automatic detection. The system then searches the file for a watermark. If found the system then decrypts the watermark to find the unique reference for the file. Once found this can then be used to look up the details for the person who was originally sent the copy of the file. This information is shown to the content owner allowing them to proceed as they see fit.

The key to the system is the management of the copies of audio files to ensure a fast delivery to the end user without having to store thousands of every format of a file which would use up a huge amount of resource. Therefore edge servers that do the watermarking and the delivery use logic as stated in Figure 7.

Each of the edge servers will record transactions whether they be download s or streaming requests.

These are stored on the central log database.

Another example of the invention in use would be to protect video files. This is done in exactly the same way as the audio files as it is the audio file that is watermarked. As video files without the audio track are useless in the majority of cases then this is a robust way to protect the video files.

The difference is that the video part of the file is stored in the formats it is to be delivered in and when a copy is required the video part is muxed with the audio which contains the watermark into a single file to deliver through to the user. This is shown in the diagram FigureS.