The combination of source coding with decoder side information (the Wyner-Ziv problem) and channel coding with encoder side information (the Gel'fand-Pinsker problem) can be optimally solved using the separation principle. In this work, we show an alternative scheme for the quadratic-Gaussian case, which merges source and channel coding. This scheme achieves the optimal performance by applying a modulo-lattice modulation to the analog source. Thus, it saves the complexity of quantization and channel decoding, and remains with the task of "shaping" only. Furthermore, for high signal-to-noise ratio (SNR), the scheme approaches the optimal performance using an SNR-independent encoder, thus it proves for this special case the feasibility of universal joint source-channel coding.Index Terms-Analog transmission, broadcast channel, joint source/channel coding, minimum mean-squared error (MMSE) estimation, modulo lattice modulation, unknown signal-to-noise ratio (SNR), writing on dirty paper, Wyner-Ziv (WZ) problem.
I. INTRODUCTIONC ONSIDER the quadratic-Gaussian joint source/channel coding problem for the Wyner-Ziv (WZ) source [1] and Gel'fand-Pinsker channel [2], as depicted in Fig. 1. In the Wyner-Ziv setup, the source is jointly distributed with some side information (SI) known at the decoder. In the Gaussian case, the WZ-source sequence is given by(1)where the unknown source part, , is Gaussian independent and identically distributed (i.i.d.) with variance , while is an arbitrary SI sequence known at the decoder. In the Gel'fand-Pinsker setup, the channel transition distribution depends on a state that serves as encoder SI. In the Gaussian case, known as the dirty-paper channel (DPC) [3], the DPC output, , is given by (2)