Project Website http://www.rle.mit.edu/sia/technology.htm
In recent years, there have been a growing realm of applications that require one to embed digital information into some host signal, which may be inherently analog or digital and which is generally not known at the decoder. There is a fundamental tradeoff between the number of bits embedded, the distortion incurred by the embedding process on the host signal, and the robustness with which the bits are embedded. Good information embedding systems make these tradeoffs efficiently. Spread-spectrum systems are examples of information embedding systems that are not efficient, except when the host signal is known at the decoder. Quantization Index Modulation (QIM) systems, by contrast, do achieve the fundamental performance limits in many practical scenarios of interest, particularly when combined with a technique referred to as distortion-compensation (DC-QIM) and error control coding (coded DC-QIM). Moreover, QIM systems can be implemented in a computationally very efficient manner.
At one end of a spectrum, applications of QIM and its variants include digital watermarking, data hiding and covert communication. At the other end of the spectrum, there provide important solutions to problems of backwards-compatible upgrading of existing analog and digital communications infrastructure, and efficient multiplexing of transmissions to different users in multiple-antenna wireless systems.