Semantic Forwarding and Codebook-Enhanced Model Division Multiple Access for Satellite-Terrestrial Networks

A research team from institutions including Beijing University of Posts and Telecommunications has published a groundbreaking paper on arXiv, introducing a new AI-driven framework called Semantic Forwarding-based Semantic Communication (SFSC) for satellite-terrestrial networks. The core innovation addresses the severe limitations of conventional bit-level transmission in space, which suffers from high path loss and limited spectrum. Instead of trying to perfectly recover every bit of data, the SFSC framework prioritizes transmitting only the task-relevant semantic information. This paradigm shift is enhanced by a novel satellite semantic forwarding mechanism, allowing relay satellites to forward processed signals at the semantic level without the computational burden of full decoding and re-encoding.

The technical architecture employs a vector-quantized joint semantic coding and modulation scheme, where a semantic encoder and codebook are co-optimized to shape constellation symbols for better channel adaptability. For robustness, a channel-aware semantic reconstruction scheme uses Feature-wise Linear Modulation (FiLM) to fuse the received signal-to-noise ratio (SNR) with semantic features. To handle multiple users, the team also developed a Codebook Split-enhanced Model Division Multiple Access (CS-MDMA) method. Simulation results are impressive, showing the proposed framework achieves a peak signal-to-noise ratio (PSNR) gain of approximately 7.9 dB over existing benchmarks in challenging low-SNR environments. This represents a major leap toward making satellite internet and IoT communications significantly more efficient and reliable.