Research On Key Technologies Of Integrated Sensing And Communication

As one of the key technologies to achieve efficient utilization of spectrum resources,spectrum sensing has been widely used in scenarios such as New Radio in Unlicensed Spectrum（NR-U）.The key to share the spectrum of multiple system signals in unlicensed frequency bands depends on the recognition and classification of signals.With the rapidly developing of emerging services such as unmanned driving and multi-vehicle interconnection,spectrum sensing and signal recognition based on cognitive radio（CR）are hardly able to meet the sensing requirements of current communication systems,and there is an urgent need for a deep collaborative integration of sensing and communication.Utilizing NR communication signals for environmental sensing can further improve the efficiency of resource utilization and achieve the dual functions of environmental sensing and data communication.This thesis focuses on the collaborative integration of sensing and communication,and combines technologies such as cognitive radio and deep learning to investigate the key technologies of sensing and communication collaboration from three aspects: spectrum sensing,signal sensing,and environment sensing.Firstly,this thesis designs and implements spectrum sensing.The second chapter of this thesis designs a multi-node collaborative spectrum sensing scheme based on USRP and GNU Radio in order to achieve a more efficient and accurate sensing of the spectrum environment.Spectrum sensing for 3GHz bandwidth in one second.The perceived time cost of the existing hardware are also analyzed.Then,in this thesis,a convolutional neural network model is constructed to identify and classify signals of multiple protocol for the scenario of co-existence of signals of multiple regimes in the unlicensed band.The training classification of WLAN,LTE LAA and 5G NR-U signals coexisting in the unlicensed band is performed,and the effects of data set size and channel on the classification accuracy are analyzed to compare the features of signals of different standards.The NR-U signals under WLAN,LTE LAA and Bluetooth interference are classified and identified to analyze the impact of different standard interference on NR-U signals.The signal initial access in the unlicensed band communication scenario may constantly switch frequency points.The parameters related to the initial access under the traditional authorized frequency band are closely related to the fixed frequency point where they are located,resulting in the way of initial access under the traditional authorized frequency band is not applicable in the unlicensed frequency band.Based on this,a deep learning-based NR-U cell search scheme is proposed.The 5G NR-U cell search is completed by identifying and classifying SSBs with different parameter types and determining the key parameters for cell search.The results show that the SSB classification accuracy can reach 100% under the condition of high signal-to-noise ratio,and can perfectly recover the MIB and SIB1 to complete the cell search.Finally,this thesis considers the capability of using NR communication signals to enhance the environment sensing.The ambiguity function performance of NR PSS,SSS,OFDM and NC-OFDM is analyzed.The communication sensing performance of OFDM with different parameter sets of NR is analyzed and compared,and the results show that as the NR subcarrier interval increases,the distance resolution is better,the discrimination ability of different targets is also better;however,the maximum detection distance is getting shorter.The SSB in 5G standard is studied and analyzed,and the reasons for its applicability to enhanced environmental sensing is given.Finally,a communication sensing integrated system model is built,and target detection simulations are performed using NR PSS,SSS,OFDM and NC-OFDM waveforms,and the results show that these types of signals are able to accomplish target detection in the environment.