Research On Theories And Key Technologies Of Integrated Communication And Sensing In Mmwave Systems

As wireless communication gradually moves towards the era of intelligent connection of everything,the future wireless communication may not only achieve extreme communication performance such as ultra-high speed,ultra-low latency,and ultra-strong reliability,but also have highly fine-grained sensing capabilities.The combination of millimeter-wave(mm Wave)communication with massive multiple input multiple output(MIMO),and ultra-dense networking technologies can not only overcome high path-loss and achieve high transmission rates,but also greatly enhance the space-time-frequency sensing resolution.As the development trend of communication and sensing in frequency band,hardware architecture,and signal processing is getting closer,to save spectrum resources and reduce equipment cost,the integration of communication and sensing has become an urgent need.Therefore,this dissertation develops research on theories and key technologies of integrated communication and sensing in mm Wave systems,by drawing on multi-domain information fusion theories,and artificial intelligence and robot sensing technologies.First of all,for the joint design paradigm of communication and sensing,we elaborate on the basic principles of mm Wave communication and wireless sensing,as well as the relationship between multipath channel parameters and electromagnetic propagation environment,including multipath channel models,classic positioning principles and radio frequency environment maps,to consolidate the foundation for the research on mm Wave-based integration of communication and sensing.We propose to effectively integrate multi-domain soft information from different sensing methods in the same device,from different users in the same network,and from different communication networks,through probability theories,to comprehensively improve the integrated performance of communication and sensing.Then,for mm Wave communication systems with hardware impairments,we propose a low-cost fully-digital beamforming module-assisted hybrid beamforming architecture,and realize the integration of beam training and localization.We analyze the impact of low-bit phase shifters,low-bit analog-digital-converters,and non-ideal radio frequency chains,on communication and sensing theoretically.We propose a fast beam training algorithm,first,radio frequency impairments are obtained by an internal calibration method,then,beam directions are estimated by compressed sensing algorithms,finally,the angle information of beams are used to localize the scatterers and users.The proposed system architecture and algorithm can achieve high beam matching and localization performance with low cost and time consumption.Subsequently,for extremely-large antenna array systems,we realize the integration of communication and localization by extremely large lens antenna array.For the near-field scenarios with spherical wave-front,we propose two electromagnetic lens phase design schemes,and establish a closed-form expression of the response of extremely-large lens antenna array.We reveal that the “sinc” function of the array response in the far field is a special case of the nearfield results.We further analyze the energy focusing window effect of the lens antenna array for integrated multi-user communication and localization.Finally,a channel estimation algorithm is designed to effectively extract the channel parameters,including location parameters.After that,for the mm Wave cloud radio access network architecture in complex multipath propagation environment,we establish a model-driven neural network algorithm to realize the information fusion of multiple base-stations and different types of measurements.First,a simultaneous location and velocity estimation model is established in a three-dimensional mobile scenario with mixed time,angle,and Doppler measurements.Then,we propose a neural network-assisted algorithm to further improve the performance,which retains the robustness of model and powerful learning ability of neural network.Finally,a public data set is adopted to verify the performance,and the experiment results show the superior performance of the proposed algorithm in localization accuracy,robustness,time consumption to the benchmarks.Moreover,for the fusion of different sensing types in the communication system,we propose a hybrid active and passive sensing mechanism and realize the mutual enhancement for integrated communication and sensing.The beam scanning process and the downlink positioning reference signal are used to realize the active and passive sensing processes,respectively.We unify active and passive sensing results by establishing the theoretical model of active sensing uncertainty.Belief propagation(BP)-based simultaneous localization and mapping(SLAM)algorithm is extended to realize the mutual assistance of active and passive sensing.The proposed mechanism can work in scenarios where no prior information about users,base-stations,and environments is available,and achieve higher sensing accuracy than benchmarks.Finally,for the fusion of biased measurements and multi-user information in communication systems,we propose a measurements plug-and-play and crowdsourcing mechanism,and realize the fusion of measurements with different quality and from multiple users.According to the attribute of measurement bias,different quality measurements are classified.Different types of measurement are implanted to the BP SLAM algorithm by different ways,and the measurement bias can be estimated during the SLAM process.Furthermore,a crowdsourcing mechanism is designed to solve the problems of “double counting”,“data reliability”,and“privacy protection”,and realize the environment-awareness communication of new users.