Multi-Task Semantic Communication for Image Transmission over MIMO Channel

Semantic communication has emerged as a promising paradigm through the transmission of semantic features to improve spectrum efficiency. This paper proposes an end-to-end semantic communication system for multi-task image transmission based on multi-input multi-output (MIMO) channel, which benefits from space multiplexing and higher channel capacity. First, we design a deep joint source-channel coding (DeepJSCC) scheme, which introduces an image cluster method at the transmitter based on the common semantic information among multiple images. Furthermore, to achieve semantic transmission of different clusters in the MIMO channel, a precoding scheme based on transformer is proposed, guaranteeing the receiver can correctly recover the semantic features. Finally, numerical results demonstrate that our proposed framework significantly outperforms other frameworks in terms of improving transmission efficiency.