Semantic communication systems in mobile networks necessitate real-time collaborative updates of semantic encoders as a result of node mobility that induces semantic extraction drift. However, diversity within modal transmission content and heterogeneous encoder architectures, along with challenges such as imbalanced training requirements and pseudo-label noise, limit the effectiveness of general collaborative update approaches. In this paper, we propose Fed-MoSeC, a novel federated learning framework for updating cross-modal semantic encoders. Our framework trains only a newly designed graph neural network-based adapter while freezing heterogeneous encoders for various modalities, converting heterogeneous cross-modal updates into a homogeneous aggregation task, and significantly reducing communication overhead. By combining confidence-based filtering with similarity-matrix distillation, the novel integrated Pseudo-label Noise Counteracting Component (PNCC) is designed to be robust to noisy data. The Training Optimization Component (TOC) based on a bi-level Cournot–Stackelberg game theoretical algorithm achieves near-optimal Subgame-Perfect Nash Equilibrium (SPNE) to incentivize across nodes and maximize update nodes’ utilities with different training levels. Experimental results highlight the advantages of Fed-MoSeC over existing potential application algorithms, reducing communication by 95–97% and improving RSUM by 18% at 70% pseudo-label noise.
Yushi Wang, Zheng-Xin Yu, Gu-Han Zheng et al.· IEEE Transactions on Mobile...· 0 citations
Acquiring channel state information (CSI) with manageable overhead has been essential to provide high-performance communication services, which is extremely challenging in the emerging sixth generation (6G) mobile network. Channel extrapolation has been proposed to infer complete CSI using a small portion of known CSI, its performance can be dramatically enhanced by artificial intelligence (AI). However, AI-driven channel extrapolation suffers from poor generalization across scenarios and high computational complexity, which is common in the broad research of AI and large language models. Inspired by the modular function of human brain, we propose a configurable AI-driven framework to achieve generalizable and computational efficient channel extrapolation from a modular perspective. We propose a three-stage framework, consisting of experts emergent, experts construction and experts selection. This framework assumes that CSI correlations can be captured by a small number of specialized functional modules (experts) that are activated differently across scenarios. Such modularity emerges in the experts emergent stage via pre-training using CSI data covering comprehensive scenarios. The neurons with similar weight-space patterns are grouped as experts in the experts construction stage. A lightweight gating function is added to control the routing of experts and is fine-tuned for each scenario in the experts selection stage. Simulation results demonstrate that the proposed three-stage framework reduce the channel extrapolation error and computational complexities dramatically by $1.1-19.1$ db and $38$ \%, respectively. In addition, attributed to the proposed experts emergent and section modules, the proposed framework outperforms its counterpart mix-of-expert model dramatically in terms of channel extrapolation performance.
Yuan Gao, Xinyi Wu, J. Jun et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
