We consider a multiple access channel (MAC) with channel state information, or side information, available at one encoder. Focusing on the case of one informed encoder and one uninformed encoder, we develop achievable rate regions in the discrete memoryless and Gaussian memoryless cases. The Gaussian case combines the techniques of partial state cancellation and dirty paper coding by the informed encoder. Both encoders benefit in terms of achievable rates from the informed encoder’s use of the channel state. Although this region is larger than that obtained by dirty paper coding alone, the state cannot be perfectly canceled as in the case when both encoders are informed. 1

A Gaussian channel, when corrupted by an additive Gaussian interfering signal whose complete sample sequence is known non-causally to the transmitter but not to the receiver, has the same capacity as if the interfering signal were not present. This is true even when the noise and interference are not necessarily stationary or ergodic. 1

A coding theorem is proved for memoryless channels when the channel state feedback of finite cardinality can be designed. Channel state information is estimated at the receiver and a function of the estimated channel state is causally fed back to the transmitter. The feedback link is assumed to be noiseless with a finite feedback alphabet, or equivalently, finite feedback rate. It is shown that the capacity can be achieved with a memoryless deterministic feedback and with a memoryless device which select transmitted symbols from a codeword of expanded alphabet according to current feedback. To characterize the capacity, we investigate the optimization of transmission and channel state feedback strategies. The optimization is performed for both channel capacity and error exponents. We show that the design of the optimal feedback scheme is identical to the design of scalar quantizer with modified distortion measures. We illustrate the optimization using Rayleigh block-fading channels. It is shown that the optimal transmission strategy has a general form of temporal water-filling in important cases. Furthermore, while feedback enhances the forward channel capacity more effectively in low-signal-to noise ratio (SNR) region compared with that of high-SNR region, the enhancement in error exponent is significant in both high- and low-SNR regions. This indicates that significant gain due to finite-rate channel state feedback is expected in practical systems in both SNR regions.

In this paper, we introduce a new approach to passive-warden steganography in which the sender embeds the secret message into a certain subset of the cover object without having to share the selection channel with the recipient. An appropriate informationtheoretical model for this communication is writing in memory with (a large number of) defective cells [1]. We describe a simple variable-rate random linear code for this channel (the “wet paper” code) and use it to develop a new steganographic methodology for digital media files – Perturbed Quantization. In Perturbed Quantization, the sender hides data while processing the cover object with an information-reducing operation, such as lossy compression, downsampling, A/D conversion, etc. The sender uses the cover object before processing as side information to confine the embedding changes to those elements of the processed cover object whose values are the most “uncertain”. This informed-sender embedding and uninformed-recipient message extraction improves steganographic security because an attacker cannot easily determine from the processed stego object the location of embedding changes. Heuristic is presented and supported by blind steganalysis [2] that a specific case of Perturbed Quantization for JPEG images is significantly less detectable than current JPEG steganographic methods.

Abstract—This paper characterizes the capacity of a class of modulo additive noise relay channels, in which the relay observes a corrupted version of the noise and has a separate channel to the destination. The capacity is shown to be strictly below the cut-set bound in general and achievable using a quantize-andforward strategy at the relay. This result confirms a previous conjecture on the capacity of channels with rate-limited side information at the receiver for this particular class of modulosum channels. This paper also considers a more general setting in which the relay is capable of providing noncausual ratelimited side information to both the transmitter and the receiver. The capacity is characterized for the case when the channel between the source and the destination is binary symmetric with a uniform noise distribution. It is shown that the rates available for conveying information about the noise to the source and the destination can be traded off with each other arbitrarily in this case; the capacity is a function of only the sum of the two rates. Index Terms—modulo-sum channel, channel with rate-limited side information, cut-set bound, quantize-and-forward, relay channel I.

We consider a state-dependent multiple access channel p(y|x1, x2, s) whose output Y is controlled by the channel inputs X1 and X2 from two encoders and the channel state S. It is assumed that the channel state is known non-causally at one encoder, called the informed encoder. We derive the capacity region for the case of degraded messages in which the informed encoder knows the message of the uninformed encoder.

In a variety of applications, there is a need to authenticate a source that may have been degraded, transformed, edited, or otherwise modified, either intentionally or unintentionally. We develop a formulation of this problem, and identify and interpret the associated informationtheoretic performance limits. The results are illustrated through application to binary sources with Hamming distortion measures, and to Gaussian sources with quadratic distortion measures.

We consider a multiple access channel (MAC) with state information non-causally known at some encoders. For simplicity of exposition, we focus on a two-encoder model in which one of the encoders is non-causally informed of the the channel state information (CSI). The results can in principle be extended to any number of encoders with a subset of them being informed. We derive an inner bound for the capacity region in the general discrete memoryless case and specialize to a binary noiseless case. We also derive an inner bound for the capacity region of an additive white Gaussian MAC with one encoder being informed of the CSI. In both binary case and Gaussian case, we compare the inner bounds with trivial outer bounds obtained by giving the CSI to the decoder. For both the binary and the Gaussian inner bounds, the informed encoder uses a slightly generalized dirty paper coding (DPC) scheme that allows arbitrary correlation between the codeword and the known CSI with negative correlation being viewed as partial state cancellation. It appears that the generalized DPC can not completely eliminate the effect of the CSI, in contrast to the case of all encoder being informed.

Consider the optimum strategy for using channel state ("side") information in transmission over a modulo-additive noise channel, with state dependent noise, where the receiver does not have access to the side information. Recent work showed that capacity-wise, the optimum transmitter shifts each code letter by a "prediction" of the noise sample based on the side information. We show that this structure achieves also the random-coding error exponent, and therefore is optimum at some range of rates below capacity. Specifically, the optimum transmitter-predictor minimizes the R'enyi entropy of the prediction error; the R'enyi order depends on the rate, and goes to one (corresponding to Shannon entropy) for rates close to capacity. In contrast, it is shown that this "prediction strategy" may not be optimal at low transmission rates. Key words: Time varying channels, side information, R'enyi entropy, prediction, error exponent. I. Introduction The somewhat uncommon scenario of a time vary...

Digital watermarking can be viewed as channel coding with side information at the encoder (CC-SI); the original data to be watermarked is known to the encoder but not the decoder. Likewise, distributed source coding is rate distortion with side information at the decoder (RD-SI); a noisy observation of the source data to be compressed is available to the decoder but not the encoder. For a Gaussian channel or source, CC-SI and RD-SI are shown to be informationtheoretic duals. Ideal coding schemes are presented, and the schemes are interpreted geometrically to highlight dual elements and quantities. 1. Introduction The duality between channel coding (CC) for the Gaussian channel and rate distortion (RD) for a Gaussian source has been known for years [5]. Recently, interest has been renewed in two similar scenarios: channel coding with side information at the encoder (CC-SI) and rate distortion with side information at the decoder (RD-SI). CC-SI relates directly to digital watermarking o...