DiT-JSCC: Rethinking Deep JSCC with Diffusion Transformers and Semantic Representations

TL;DR

DiT-JSCC introduces a novel joint source-channel coding framework combining semantic-aware encoding and diffusion transformer decoding, significantly improving image transmission quality and semantic consistency under extreme wireless channel conditions.

Contribution

The paper proposes a semantics-prioritized encoder and a diffusion transformer decoder for GJSCC, addressing semantic mismatch issues and enhancing transmission efficiency with adaptive bandwidth allocation.

Findings

Outperforms existing JSCC methods in semantic consistency

Achieves higher visual quality in extreme channel conditions

Introduces a training-free adaptive bandwidth strategy

Abstract

Generative joint source-channel coding (GJSCC) has emerged as a new Deep JSCC paradigm for achieving high-fidelity and robust image transmission under extreme wireless channel conditions, such as ultra-low bandwidth and low signal-to-noise ratio. Recent studies commonly adopt diffusion models as generative decoders, but they frequently produce visually realistic results with limited semantic consistency. This limitation stems from a fundamental mismatch between reconstruction-oriented JSCC encoders and generative decoders, as the former lack explicit semantic discriminability and fail to provide reliable conditional cues. In this paper, we propose DiT-JSCC, a novel GJSCC backbone that can jointly learn a semantics-prioritized representation encoder and a diffusion transformer (DiT) based generative decoder, our open-source project aims to promote the future research in GJSCC. Specifically, we design a semantics-detail dual-branch encoder that aligns naturally with a coarse-to-fine conditional DiT decoder, prioritizing semantic consistency under extreme channel conditions. Moreover, a training-free adaptive bandwidth allocation strategy inspired by Kolmogorov complexity is introduced to further improve the transmission efficiency, thereby indeed redefining the notion of information value in the era of generative decoding. Extensive experiments demonstrate that DiT-JSCC consistently outperforms existing JSCC methods in both semantic consistency and visual quality, particularly in extreme regimes.