Semantic Communication in Dynamic Channel Scenarios: Collaborative Optimization of Dual-Pipeline Joint Source-Channel Coding and Personalized Federated Learning

With the continuous advancement of wireless communication [3] introduced a attention feature model blended between network technologies and the widespread adoption of various feature extraction modules, enhancing adaptability to random data-intensive applications such as AR/VR multimedia, traditional channels but increasing complexity and latency. In contrast, communication systems are facing significant challenges [4] incorporated traditional modules like demodulation and in supporting massive data transmission. Concurrently, as the quantization into semantic communication, enabling adaptive developing of the sixth-generation (6G) network, the integration CSI optimization. of satellite internet into terrestrial communication systems In modern communication scenarios, network topologies becomes increasingly feasible. However, the satellite-to-ground typically feature multi-user access, where multiple client nodes transmission links are inherently constrained by limitations in connect to a central node, resembling noval network topologies bandwidth and latency. To address these existing and potential such as edge computing and self-organizing networks. However, challenges, deep learning-based joint source-channel coding in practical training and deployment, CSI exhibits continuous (Deep JSCC) has surfaced as a promising approach, serving as and dynamic variations, posing challenges for adaptive joint a method to realize Semantic Communication (SC).