Search papers, labs, and topics across Lattice.
This paper explores a semantic-aware neural joint source-channel coding (JSCC) framework for robust video transmission over block erasure channels, focusing on both spatial and feature domains. By partitioning video frames and latent features, the proposed methods enable localized handling of erasures and semantic recovery of missing information, respectively. Experimental results demonstrate that spatial-domain JSCC is effective against random localized losses, while feature-domain JSCC outperforms in scenarios with distributed erasures, revealing critical trade-offs between spatial continuity and semantic redundancy.
Spatial-domain JSCC excels at localized loss recovery, but feature-domain JSCC outshines in maintaining fidelity during distributed erasures.
This paper investigates semantic-aware neural joint source-channel coding (JSCC) for robust video transmission over block erasure channels. We propose a neural video compression framework exploring both spatial-domain and feature-domain designs. In the spatial domain, video frames are partitioned into blocks, enabling localized erasure handling and fine-grained robustness control via uniform erasure and two-level, semantic-guided non-uniform erasure strategies. In the feature domain, latent features are partitioned, enabling missing features to be semantically recovered while maintaining overall spatial consistency. Comprehensive experiments quantify reconstruction quality under varying uniform and non-uniform erasure probabilities. Our results show that spatial-domain JSCC excels at handling random localized losses, whereas feature-domain JSCC provides superior robustness to distributed erasures and maintains fidelity under low-loss scenarios. The analysis highlights the trade-offs between spatial continuity and semantic redundancy, offering insights for designing robust, task-aware video communication systems.
