.png){kind=icon}
.png){kind=icon}
Return
Robust Semantic Communication: A Dimension Selective Fading Analysis and Repair Approach
.png){kind=icon}
PRE
AI
M
Y
L
Y
Y
J
DOI：10.1109/twc.2026.3717328.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Text semantic communication aims to preserve meaning rather than reproduce bit sequences, but wireless fading and noise can still cause nonuniform semantic degradation after semantic representations are quantized, encoded, and transmitted. This paper develops a robust and interpretable text semantic communication framework that links sentence representation, distortion caused by wireless channels, semantic repair, and semantic fidelity evaluation. First, an interpretable representation guided by WordNet is constructed, where each semantic dimension corresponds to a predefined attribute and each sentence is described by a semantic tensor over tokens with explicit normalization and quantization rules. Second, semantic dimension selective fading (SDSF) is characterized through a finite difference sensitivity model that measures how hard bit flips affect each semantic dimension without using derivatives over discrete bits. The resulting vulnerability model is calibrated over SNR and Doppler conditions and is used to guide protection and repair across semantic dimensions. Third, a semantic anchor assisted repair module estimates semantic distortion at the receiver from known semantic anchors and applies signed correction directly in the semantic space, with sensitivity weighting and validity constraints. Finally, a cooccurrence proportion (CoP) metric and a context aware variant are developed to evaluate semantic similarity while balancing accuracy and complexity. Experiments under fading channels with the same bandwidth budget, together with semantic textual similarity validation and complexity analysis, show that the proposed framework improves semantic robustness while keeping the system interpretable and lightweight.
Keywords:
Semantic communication
semantic dimension selective fading
semantic repair
semantic similarity
Journal
IF:
10.7
Papers:
1.3W
Citations:
5.3W
Organization
Cited Papers