Symbol Distributions in Semantic Communications: A Source-Channel Equilibrium Perspective

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Yoo, Hanju, Choi, Dongha, Kim, Songkuk, Chae, Chan-Byoung, Heath Jr, Robert W.
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909964898926592
author Yoo, Hanju
Choi, Dongha
Kim, Songkuk
Chae, Chan-Byoung
Heath Jr, Robert W.
author_facet Yoo, Hanju
Choi, Dongha
Kim, Songkuk
Chae, Chan-Byoung
Heath Jr, Robert W.
contents Semantic communication systems often use an end-to-end neural network to map input data into continuous symbols. These symbols, which are essentially neural network features, usually have fixed dimensions and heavy-tailed distributions. However, due to the end-to-end training nature of the neural network encoder, the underlying reason for the symbol distribution remains underexplored. We propose a new explanation for the semantic symbol distribution: an inherent trade-off between source coding and communications. Specifically, the encoder balances two objectives: allocating power for minimum \emph{effective codelength} (for source coding) and maximizing mutual information (for communications). We formalize this trade-off via an information-theoretic optimization framework, which yields a Student's $t$-distribution as the resulting symbol distribution. Through extensive studies on image-based semantic systems, we find that our formulation models the learned symbols and predicts how the symbol distribution's shape parameter changes with respect to (i) the use of variable-length coding and (ii) the dataset's entropy variability. Furthermore, we demonstrate how introducing a regularizer that enforces a target symbol distribution, which guides the encoder towards a target prior (e.g., Gaussian), improves training convergence and supports our hypothesis.
format Preprint
id arxiv_https___arxiv_org_abs_2512_14022
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Symbol Distributions in Semantic Communications: A Source-Channel Equilibrium Perspective
Yoo, Hanju
Choi, Dongha
Kim, Songkuk
Chae, Chan-Byoung
Heath Jr, Robert W.
Information Theory
Signal Processing
Semantic communication systems often use an end-to-end neural network to map input data into continuous symbols. These symbols, which are essentially neural network features, usually have fixed dimensions and heavy-tailed distributions. However, due to the end-to-end training nature of the neural network encoder, the underlying reason for the symbol distribution remains underexplored. We propose a new explanation for the semantic symbol distribution: an inherent trade-off between source coding and communications. Specifically, the encoder balances two objectives: allocating power for minimum \emph{effective codelength} (for source coding) and maximizing mutual information (for communications). We formalize this trade-off via an information-theoretic optimization framework, which yields a Student's $t$-distribution as the resulting symbol distribution. Through extensive studies on image-based semantic systems, we find that our formulation models the learned symbols and predicts how the symbol distribution's shape parameter changes with respect to (i) the use of variable-length coding and (ii) the dataset's entropy variability. Furthermore, we demonstrate how introducing a regularizer that enforces a target symbol distribution, which guides the encoder towards a target prior (e.g., Gaussian), improves training convergence and supports our hypothesis.
title Symbol Distributions in Semantic Communications: A Source-Channel Equilibrium Perspective
topic Information Theory
Signal Processing
url https://arxiv.org/abs/2512.14022