SpatialMix-JSCC: Spatial Mixing State Space Duality for Adaptive Wireless Image Transmission

Abstract

SSMs offer linear-complexity modeling for JSCC-based wireless image transmission, but 1-D token serialization weakens 2-D spatial priors and structural fidelity, especially at low CBRs. We propose SpatialMix-JSCC, an SSM-based JSCC framework that injects 2-D spatial awareness into SSM parameterization via SpatialMix-SSD, combining quad-directional scanning, deformable depthwise convolution, and hidden-state computation. The operator is instantiated as a hierarchical Local–Global–Local mixing block for efficient feature integration. Meanwhile, to enhance adaptability under varying SNR/CBR conditions, we introduce a lightweight dual-stream adaptation mechanism, enabling a single model to adapt to a range of simulated channel and bandwidth conditions without retraining. On AFHQ at <inline-formula> <tex-math notation="LaTeX">$CBR=1/48$ </tex-math></inline-formula> and <inline-formula> <tex-math notation="LaTeX">$\mathrm {SNR}=15$ </tex-math></inline-formula> dB, SpatialMix-Basic achieves 30.7 dB PSNR, yielding 0.4 dB higher PSNR and 16.4% fewer MACs than SwinJSCC w/o SA&RA.