基于内部实体信息增强的语义通信

Semantic Communication Based on Internal Entity Information Enhancement

近年来,深度学习的快速发展为语义通信的实现提供了强有力的支持。一个值得探究的问题是知识库的应用。部署和生成知识库需要占用大量的缓存资源,而频繁地请求知识库可能会增加通信开销。考虑了如何在小规模知识库以及不增加传输数据量的情况下提高文本语义通信系统的性能,设计了一个基于内部实体信息增强的语义通信系统。具体地说,设计了一个实体识别器来识别文本内部的实体信息,并通过二次编码的方式来挖掘和增强实体信息,在接收端将实体信息和文本的整体语义信息融合解码。仿真结果表明,该系统可以有效利用实体信息提高接收端的性能,在低信噪比下有3%以上的提升。

Recently,the rapid development of deep learning has provided strong support for the implementation of semantic communication.A noteworthy issue is the application of the knowledge base.Deploying and generating the knowledge base requires substantial caching resources,and frequent requests to the knowledge base may increase communication overhead.We consider the issue of improving the performance of text semantic communication systems in the context of a small-scale knowledge base without increasing data traffic,and a semantic communication system is designed based on internal entity information enhancement.Specifically,an entity recognizer is designed to identify the entity information within the text,and the entity information is mined and enhanced through the secondary coding.The the entity information and the overall semantic information of the text are integrated to be decoded at the receiver.Simulation results demonstrate that the proposed system can effectively use the entity information to improve the performance at the receiver and achieve 3%performance improvement under the low signal-to-noise ratio regime.