Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Semantic Communications: Principles and Challenges

Zhijin Qin, Xiaoming Tao, Jianhua Lu, Wen Tong, Geoffrey Ye Li

TL;DR

The paper surveys semantic communications as a shift from Shannon-symbol fidelity to task-driven semantic transmission, outlining core semantic theory concepts such as semantic entropy, semantic channel, and semantic rate-distortion. It discusses DL-enabled architectures (e.g., DeepSC, DeepJSCC) and their end-to-end designs for text, image/video, and speech, emphasizing task-oriented losses and semantic similarity metrics. The work highlights the lack of a universal performance metric and identifies open challenges in theory, robust transceivers, reasoning, and resource allocation, while presenting multimodal and unified frameworks for future semantic networks. Overall, the article provides a comprehensive map of concepts, methods, and open problems guiding the development of practical semantic communication systems.

Abstract

Semantic communication, regarded as the breakthrough beyond the Shannon paradigm, aims at the successful transmission of semantic information conveyed by the source rather than the accurate reception of each single symbol or bit regardless of its meaning. This article provides an overview on semantic communications. After a brief review of Shannon information theory, we discuss semantic communications with theory, framework, and system design enabled by deep learning. Different from the symbol/bit error rate used for measuring conventional communication systems, performance metrics for semantic communications are also discussed. The article concludes with several open questions in semantic communications.

Semantic Communications: Principles and Challenges

TL;DR

The paper surveys semantic communications as a shift from Shannon-symbol fidelity to task-driven semantic transmission, outlining core semantic theory concepts such as semantic entropy, semantic channel, and semantic rate-distortion. It discusses DL-enabled architectures (e.g., DeepSC, DeepJSCC) and their end-to-end designs for text, image/video, and speech, emphasizing task-oriented losses and semantic similarity metrics. The work highlights the lack of a universal performance metric and identifies open challenges in theory, robust transceivers, reasoning, and resource allocation, while presenting multimodal and unified frameworks for future semantic networks. Overall, the article provides a comprehensive map of concepts, methods, and open problems guiding the development of practical semantic communication systems.

Abstract

Semantic communication, regarded as the breakthrough beyond the Shannon paradigm, aims at the successful transmission of semantic information conveyed by the source rather than the accurate reception of each single symbol or bit regardless of its meaning. This article provides an overview on semantic communications. After a brief review of Shannon information theory, we discuss semantic communications with theory, framework, and system design enabled by deep learning. Different from the symbol/bit error rate used for measuring conventional communication systems, performance metrics for semantic communications are also discussed. The article concludes with several open questions in semantic communications.
Paper Structure (27 sections, 4 theorems, 22 equations, 11 figures, 1 table)

This paper contains 27 sections, 4 theorems, 22 equations, 11 figures, 1 table.

Table of Contents

  1. Introduction
  2. From Information Theory to Semantic Theory
  3. Difference between Conventional and Semantic Communications
  4. Information Theory
  5. Semantic Theory
  6. Semantic Entropy
  7. Semantic Channel
  8. Semantic Channel Capacity
  9. Semantic Rate Distortion and Information Bottleneck
  10. Components, Semantic Noise, and Performance Metrics
  11. Semantic Communication System Components
  12. Semantic Noise
  13. Performance Metrics
  14. Text Semantic Similarity
  15. Image Semantic Similarity
  16. ...and 12 more sections

Key Result

Theorem 1

For transmission over noisy channels, described by $p(y_j|x_i)$, channel capacity is given by where $I(X;Y)=H(X)-H(X|Y)$ is the mutual information between the input, $X$, and the output, $Y$, of the channel, and $H(X|Y)=- \sum\limits_{j = 1} {p(y_j)}\sum\limits_{i = 1} {{p(x_i|y_i)}{{\log }_2}} {p(x_i|y_i)}$ is the conditional entropy of $X$ for a given $Y$ as shown in Fig. fig_com(a).

Figures (11)

  • Figure 1: The illustration of a semantic communication system for object recognition.
  • Figure 2: Comparison of conventional and semantic communication systems.
  • Figure 3: The main components in a semantic communication system.
  • Figure 4: The novel layered architecture of semantic communication system Lan2021WhatIS.
  • Figure 5: An example of semantic mismatch for the word "earth" between two persons.
  • ...and 6 more figures

Theorems & Definitions (6)

  • Definition 1
  • Theorem 1
  • Theorem 2
  • Theorem 3
  • Definition 2
  • Theorem 4