Beidou Enabled Semantic Edge Aware Smart Grid Transmission

Conventional satellite communication systems often suffer from limited bandwidth, posing significant challenges for efficient image transmission in remote mountainous regions within smart grids. To address this issue, we propose a novel end-to-end framework called Beidou-enabled semantic and edge-aware smart grid transmission (BSE-SG), which integrates the Beidou-3 (BD3) short message protocol with semantic communication (SC). In the considered system, image data collected at the edge is first encoded to extract semantic features, which are then compressed via vector quantization (VQ) and transmitted as semantic indices through BD3 satellite links to the central server. Considering the bit-level transmission constraints of BD3, an entropy-coding-based codebook reordering strategy is proposed to reduce transmission overhead by adaptively reorganizing the codebook according to index usage frequency, assigning short bit representations to high-frequency indices. The semantic encoder-decoder is further fine-tuned to prioritize the selection of these indices. In addition, the compactness of the reordered codebook structure enhances the system’s robustness against channel-induced disturbances. Experimental results based on real-world power grid image datasets demonstrate the proposed BSE-SG can achieve up to 30.3%, 15.5%, and 30.5% MS-SSIM gains compared with WITT, JPEG, and BPG. Meanwhile, the codebook reordering further reduces the transmission data volume by 19.1%, verifying the practicality and effectiveness of the proposed strategy in bandwidth-constrained satellite communication system.