Towards Intelligent Communications: Large Model Empowered Semantic Communications

Deep learning enabled semantic communications have shown great potential tosignificantly improve transmission efficiency and alleviate spectrum scarcity,by effectively exchanging the semantics behind the data. Recently, theemergence of large models, boasting billions of parameters, has unveiledremarkable human-like intelligence, offering a promising avenue for advancingsemantic communication by enhancing semantic understanding and contextualunderstanding. This article systematically investigates the largemodel-empowered semantic communication systems from potential applications tosystem design. First, we propose a new semantic communication architecture thatseamlessly integrates large models into semantic communication through theintroduction of a memory module. Then, the typical applications are illustratedto show the benefits of the new architecture. Besides, we discuss the keydesigns in implementing the new semantic communication systems from moduledesign to system training. Finally, the potential research directions areidentified to boost the large model-empowered semantic communications.