Semantic Communication
Semantic communication aims to transmit only the essential meaning of data, rather than raw bits, improving efficiency and bandwidth utilization. Current research focuses on integrating generative AI models, such as diffusion models and transformers, with techniques like federated learning and joint source-channel coding to achieve robust and efficient semantic encoding and decoding across various modalities (images, speech, video). This approach holds significant promise for enhancing communication in resource-constrained environments and enabling new applications in areas like autonomous driving, IoT, and satellite networks by prioritizing meaningful information and reducing transmission overhead.
Papers
Generative AI-aided Joint Training-free Secure Semantic Communications via Multi-modal Prompts
Hongyang Du, Guangyuan Liu, Dusit Niyato, Jiayi Zhang, Jiawen Kang, Zehui Xiong, Bo Ai, Dong In Kim
Enhancing Semantic Communication with Deep Generative Models -- An ICASSP Special Session Overview
Eleonora Grassucci, Yuki Mitsufuji, Ping Zhang, Danilo Comminiello