Research On Key Technologies Of Semantic Enhanced Intelligent Communication

With the challenges posed by the increasingly complex communication needs and the growing volume of data,semantic intelligent communication aims to deeply explore the semantic connotation of information and create a smart and concise new generation communication system.This thesis focuses on semantic enhanced intelligent communication,simulating scenarios with semantic network and transmission.Specifically,two main aspects are studied:the design and evaluation of intelligent network management methods enabled by text semantics,and the modeling and application of marine path loss based on environmental semantic awareness.For the future communication network,this thesis designs a semantic intelligent network management method using natural language as the agent interaction carrier.Specifically,for the future autonomous intelligent network composed of multiple agents,this thesis proposes a protocol mechanism based on natural language semantic interaction to realize autonomous cognition and swarm intelligence of the network.First,use the existing network management protocol instruction to build the agent interaction semantic knowledge base.Furthermore,a deep semantic understanding network based on bidirectional long and short-term memory network and a semantic generation rule based on template generation are designed to support the interaction of agents.Finally,based on NS3 simulation software to simulate typical network management scenarios,the performance of semantic intelligent network management system is verified and evaluated.The experimental results show that the network management method proposed in this thesis is efficient and robust.For the future communication transmission,this thesis designs a method of marine path loss modeling based on environmental semantic awareness,and applies it to the intelligent decision-making of UAV transmission mode.Marine wireless communication is faced with the main challenge of the complex characteristics of marine radio wave propagation.Extracting the semantic information about radio wave propagation in the marine meteorological environment will help to infer the marine channel state information and build a more accurate propagation loss model.Therefore,this thesis first proposes a model-data co-driven forecasting method of sea-surface duct parameters,which integrates the mesoscale model and the marine meteorological sounding data.Furthermore,this thesis proposes a marine propagation loss modeling method based on environmental semantic awareness.Finally,this thesis studies the application of the proposed marine propagation loss model in the selection of communication modes between unmanned aerial vehicles(UAVs),aiming at reducing the energy consumption of UAV data transmission.The simulation results show that the method proposed in this thesis improves the prediction accuracy of propagation loss,and has certain guiding significance for the optimization of communication between UAVs in marine environment.