Text-Guided Token Communication for Wireless Image Transmission

With the emergence of $\mathbf{6 G}$ networks and proliferation of visual applications, efficient image transmission under adverse channel conditions is critical. We present a text-guided token communication system leveraging pre-trained foundation models for wireless image transmission. Our approach converts images to discrete tokens, applies 5G NR polar codec on top of the tokenizeation, and employs text as a conditioning signal to generate lost tokens to mitigate the cliff effect at lower signal-to-noise ratios (SNRs). Evaluations on ImageNet show our method outperforms state-of-the-art deep joint source-channel coding scheme in perceptual quality and semantic preservation at extremely low bandwidth ratio, i.e., 1/96. In addition, Our system requires no scenario-specific retraining and exhibits superior cross-dataset generalization, establishing a new paradigm for efficient image transmission aligned with human perceptual priorities.