Semantic communication system with efficient integration of global and local context features

A communication system based on extended contextual semantic features was proposed by using an end-to-end integrated design method based on deep learning.Unlike existing research that focused only on local context while neglecting global context, the proposed system integrated both local and global contextual knowledge, semantic encoding and decoding was utilized by extended contextual knowledge, thereby enhancing the reliability of the semantic communication system.At the transmitter, efficient semantic representation was achieved through extended contextual semantic encoding.At the receiver, the accuracy of semantic inference was improved by combining mechanisms such as historical communication text mining, contextual semantic feature learning, and heuristic graph-based decoding strategy.When comparing with the traditional communication system and the existing semantic communication systems, simulation results demonstrate that the proposed system significantly improves the reliability of the communication system under low signal-to-noise ratio.