Coding Schemes for Relay-Assisted Information Embedding

摘要

Cooperative information embedding deals with the problem of embedding unperceived information into some cover signal by different users or partners, cooperatively. It models applications in which embedded signals, or watermarks, transmitted over wireless networks need to be reinforced in order to withstand channel impairments. In cooperative information embedding, each embedder that can reinforce the embedded signal may or may not know the original cover signal. In this paper, we concentrate on the two user cases: 1) an initial embedder and 2) an assisting embedder or helper collaborate to embed some watermark into given digital media content which is transmitted over a wireless network. One important application is that of infrastructure-aided information embedding, a case in which the network provider plays the role of a helper and contributes to securing the distribution of the media, not only by blocking unauthorized signals but also by reinforcing the watermarks in legitimate signals. We investigate the two scenarios in which the helper does or does not know the cover signal. For each scenario, we derive lower and upper bounds on channel capacity. Furthermore, we also design implementable coding schemes and derive the embedding rates practically allowed by these schemes, for both scenarios. Among others, the performance characterization shows that for cooperative information embedding to be effective, careful code design is required at the initial embedder and the helper. The careful design concerns the joint conception of the embedded codes and the exploitation of the knowledge of the cover signal, if any.