… to hierarchical joint source-channel coding (JSCC), studied in … multiple layers using successive
refinement source coding [3], … neural networks (CNNs) and the communication channel is …

… First, we review separate source and channel coding in Section 2.2. We then present how
DNNs have been used for compression and channel coding in Section 2.3. A neural network–…

… As it was already suggested in [10], we introduce an interleaver between the source coder
and the channel coder. The Bayesian Network formalism shows that there is no need for a …

… Abstract—We propose a convolutional neural network (CNN) architecture with lower
reconstruction distortion and stronger noise adaptability for deep joint source-channel coding for …

… We now consider a joint source and channel coder optimization where residual redundancy
… If the source coder is DPCM and the channel coder is a convolutional coder, Fig. 1 becomes …

… The properties of jointly typical sequences can be used to prove one of the principal
theorems in information theory: the channel coding theorem. For lack of space we cannot do this …

… have led to considering joint source-channel coding (JSCC) … communication over noisy
channels. Researchers have taken … coder, that can encompass a source or/and a channel coder …

… vector quantization, JPEG coding, and MPEG coding, are … channel SNR reduction can be
achieved by the joint decoder without increasing computational cost for various channel coding …

… NN-based joint source channel coding (JSCC) is proposed for images in [4], and for text in
[5]. Though traditional techniques are optimal in the asymptotic regimes, in practical scenarios…

… The significance of joint source–channel codes extends beyond memoryless point-to-point …
Example 2 considered a simple model of neural communication. This also illustrates the …