TONIC: Token-Centric Semantic Communication for Task-Oriented Wireless Systems

2026-05-20

AI Summary

This work addresses the mismatch between conventional bit-fidelity-oriented wireless communication and the semantic units (e.g., tokens) required by downstream foundation models, which leads to inefficient resource usage and performance degradation. To bridge this gap, the authors propose a task-oriented semantic communication framework that treats tokens as the fundamental transmission unit. At the transmitter, utility-aware unequal error protection is applied based on task relevance, while at the receiver, a confidence-gated mechanism combined with a Transformer-based completion model converts detrimental errors into recoverable erasures. The architecture is modular and interpretable, underpinned by a utility-aware Bayesian risk theory that guides gate design. Experiments demonstrate that the proposed method significantly outperforms traditional separation-based schemes, pixel-domain DeepJSCC, and token-domain baselines in image classification tasks across AWGN, Rayleigh, and Rician fading channels.

Abstract

Tokens are becoming the basic units through which foundation models represent and process information for understanding and inference. However, traditional wireless communication, centered on bit-level fidelity, faces a mismatch between what is transmitted reliably and what downstream models actually consume. This mismatch calls for a communication design that directly accounts for token-level task relevance and downstream model requirements, rather than treating all transmitted bits as equally important. In this paper, we propose TONIC, a token-centric semantic communication framework for task-oriented wireless systems. The transmitter converts each source sample into a sequence of tokens, estimates token-level task relevance, and allocates protection through utility-aware unequal error protection under a fixed channel-use budget. At the receiver, token-level confidence is used to gate unreliable decisions, turning harmful substitutions into recoverable erasures before a Transformer-based completion model restores the masked tokens for final task inference. Our framework combines transmitter-side semantic-aware protection with receiver-side confidence-aware gating in a modular and interpretable architecture, rather than relying solely on fully black-box end-to-end learning. We further establish a utility-aware Bayes-risk interpretation for the receiver-side gating rule and study its interaction with unequal protection and completion. Experimental results on image classification show that TONIC consistently outperforms separation-based schemes, the pixel-domain DeepJSCC baseline, and token-domain baselines under matched communication budgets over AWGN, Rayleigh, and Rician channels.

Problem

• semantic communication
• token-centric
• task-oriented
• wireless systems
• unequal error protection

Innovation

• semantic communication
• token-centric
• unequal error protection
• confidence-aware gating
• task-oriented wireless systems