WHIM: watermarking multicast video with a hierarchy of intermediaries

Abstract

Fingerprinting, watermarking content to identify the recipient, provides a good deterrence to unauthorized duplication and propagation of multimedia streams. This is straightforward in a unicast environment; however, in a multicast environment, inserting a fingerprint at the source does not provide any security since many receivers will share a common fingerprint. A simple solution would be to fingerprint the data for each user at the source and unicast the different streams. We aim to achieve a more scalable solution while maintaining and even increasing the level of security. To achieve this, we have developed WHIM, a scalable system that allows multicast content to be marked with distinct information for distinct receivers securely. This system introduces two new concepts: (1) generating a watermark based on the receiver's location in a tree overlaying the network and (2) inserting the watermark in the content incrementally as it traverses an overlay network. We propose and evaluate several forms of this architecture and show how it improves scalability while increasing security.

Introduction

As content distribution on the Internet becomes more pervasive and the value of the content being distributed increases, the security of this data has become a main concern of content providers. Encryption is generally used to safeguard the content while it is being transmitted so that unauthorized persons cannot read the stream from the network, but this offers no protection after the intended receiver receives the data. There is no protection against unauthorized duplication and propagation by the intended receiver. This additional protection can be obtained by watermarking the content. Watermarking is the embedding of some identifying information into the content in such a manner that it cannot be removed by the user but it can be extracted or read by the appropriate party. Watermarks can be used for copyright protection or for identification of the receiver. Copyright protection watermarks embed some information in the data to identify the copyright holder or content provider, while receiver-identifying watermarking, commonly referred to as fingerprinting [1], embeds information to identify the receiver of that copy of the content. Thus, if an unauthorized copy of the content is recovered, extracting the fingerprint will show who the initial receiver was.

Problems arise when attempting to fingerprint content in a multicast environment that do not arise in copyright protection watermarking. Copyright protection watermarks are embedded in the data at the source, then the watermarked data is multicast to the group of receivers. For fingerprinting, embedding the receiver's identification as the watermark at the source will not work since all the receivers will share the same watermark. It is necessary to watermark content with unique information for distinct receivers of the same multicast stream. A simple method to achieve unique watermarks for each receiver would be to watermark the stream differently for each receiver and to unicast the watermarked streams. Of course, the inefficiency of such a scheme calls for a better solution. We aim to maintain the security of this approach while achieving scalability.

We propose WHIM, a scalable system that allows multicast content to be securely marked with distinct information for distinct receivers. This system introduces two new concepts: (1) generating a watermark based on the receiver's location in the network and (2) inserting the watermark in the content incrementally as it traverses the network. WHIM makes use of a hierarchy of intermediaries for creating and embedding the fingerprint. This allows security and scalability. The use of a hierarchy allows a new type of security by having an User ID based on the user's location in an overlay network. Security is also maintained by using proven watermarking algorithms to embed this User ID. The hierarchy leads to scalability by capitalizing on the efficiency of multicast distribution and by distributing the watermark embedding load from the source to the different intermediaries.

This paper proceeds as follows. In Section 2 we enumerate the design objectives of WHIM. Section 3 gives an overview of the WHIM architecture. Section 4 discusses the WHIM-Backbone (WHIM-BB) component which is based on a hierarchy of intermediaries that provide an efficient distribution architecture that fingerprints the streaming content. Section 5 describes the WHIM-Last Hop (WHIM-LH) component, a secure protocol that fingerprints and distributes content between an intermediary and a group of receivers. Section 6 examines previous work in the area. Section 7 presents an analysis and simulation results of the efficiency of WHIM, and a comparison with previous solutions. Finally, Section 8 presents conclusions and discusses possible future work.