| In the 5th generation mobile communication(5G)New Radio(NR)systems,after user equipment(UE)is powered on,it accesses the cell through the process of cell search and random access.Cell search is the first step when establishing communication between UE and base station(BS),its performance directly affects the subsequent signal transmission.In the initial access stage of the standalone(SA)architecture,UE hardly knows any information about BS and transmission channel,including time and frequency information of the system as well as cell information,but the orthogonal frequency-division multiplexing(OFDM)system is sensitive to timing offset and frequency offset.Therefore,cell search algorithms need to be designed to achieve downlink time and frequency synchronization between UE and BS,and for UE to obtain the physical cell identification(PCID)and master information block(MIB)messages of the corresponding cell.In 5G systems,complex channel conditions,low signal-to-noise ratio(SNR)and large carrier frequency offset(CFO)bring greater challenges to cell search.To address these challenges,5G introduces the design of synchronization signal block(SSB),which allows flexible configuration of time and frequency domain locations for SSB transmissions and also supports beam scanning transmissions for coverage enhancement.These new features further increase the complexity of cell search,while many application scenarios require short processing time and high implementation accuracy,and for power saving requirements,UEs have higher requirements for low power consumption,requiring more efficient algorithms to adapt to the new performance requirements.In view of these,this thesis investigates cell search and downlink time frequency synchronization methods for 5G NR mobile communication systems with unknown SSB frequency location and large CFO under standalone networking architecture.First,the transmission specifications of downlink synchronization and broadcast information are studied in conjunction with the 5G NR technical protocols and standards.The impacts of symbol timing error and CFO on the performance of OFDM systems are investigated,so as to reveal the importance of time and frequency synchronization for normal operation of the system.Subsequently,the structure of SSB and the design of its signals and channels are studied,and time frequency positions of SSB transmissions is explained in combination with the frame structure and synchronization raster.For the standalone networking architecture,the process of 5G NR cell search is studied,including primary synchronization signal(PSS)search,secondary synchronization signal(SSS)detection and physical broadcast channel(PBCH)demodulation,and the physical layer process of SSB generation and transmission at BS side is studied.Research on the transmission specifications of downlink synchronization information shows that large CFO can have a serious impact on the accuracy and effectiveness of the time and frequency synchronization algorithms,and it is necessary to use the properties and characteristics of the NR synchronization signals to design downlink time and frequency synchronization methods that support large CFO.In addition,the flexible configuration of SSB frequency location increases the complexity of frequency search,and thus more efficient methods for frequency information acquisition need to be designed.Next,cell search and downlink time frequency synchronization algorithms under 5G NR standalone networking are studied.Based on the cross-correlation characteristics of the synchronization signal,search and detection algorithm of PSS and symbol timing algorithm are investigated.The above algorithms acquire timing metrics through segmented cross-correlation to adapt to the influence of large CFO on signal correlation,and moves SSB to the vicinity of the center frequency through a series of possible spectrum shifts,to search for SSB at each frequency point of the synchronization raster in the operating band.Simulation results show that according to the reasonable decision thresholds set by timing metrics under different channel conditions,the segmented cross-correlation based PSS search and coarse timing estimation algorithms can adapt to the effects of large noise and large frequency offset,and achieves good results in synchronization signal detection and coarse timing estimation.Then,CFO estimation algorithms are investigated,in which firstly integral part of frequency offset(IFO)is estimated based on time-domain correlation or frequency-domain correlation of synchronization signal,and then fractional part of frequency offset(FFO)is estimated based on synchronization signal or cyclic prefix(CP).Simulation results show that the estimation algorithm based on CP achieves better performance than that based on PSS,and fine estimation results can be achieved under medium-high SNR conditions.After that,detection algorithms of SSS are studied.Compared with the frequency-domain direct correlation based detection algorithm,the fast Walsh-Hadamard transform(FWHT)based detection algorithm achieves equivalent results,but with significantly reduced computation amounts.Further,downlink timing and frequency synchronization methods in 5G NR massive MIMO systems are studied.For UE reception with multiple antennas,theoretical model of downlink timing and frequency offset estimation is established.Based on the conjugate symmetry property of synchronization signal,a symmetric autocorrelation algorithm for PSS search and coarse timing estimation is proposed.The symmetric autocorrelation algorithm can adapt to any size of frequency offset,and does not need to perform spectrum shifts on received signal,which can eliminate the process of SSB frequency search.Simulation results show that the symmetric autocorrelation algorithm can achieve high detection probability under medium-high SNR conditions,and its timing estimation error is within the allowed range of the system.Subsequently,based on the maximum likelihood(ML)criterion,a combined frequency estimation algorithm of SSB frequency shift and CFO is proposed.The ML joint frequency estimation algorithm can tolerate certain timing estimation error,and can estimate the frequency location of SSB within a certain range together,eliminating the need of SSB frequency search within a certain frequency band,thereby simplifying and accelerating the cell search process.Simulation results show that the ML joint frequency estimation algorithm can well estimate the frequency location of SSB and CFO under the medium-high SNR conditions.Finally,deep learning(DL)based synchronization signal search,downlink timing estimation and CFO estimation methods are proposed,to address the problem of difficulty in setting judgment thresholds in timing metric algorithms.The deep learning based synchronization signal search and downlink timing estimation methods do not require manual setting of decision thresholds,and can adapt to new channel conditions as well as transmission conditions through learning from sample data,thus with better adaptability and scalability than the mathematical model methods.Simulation results show that the deep learning based methods can achieve high detection probability of synchronization signal,small timing estimation error and small CFO estimation error under medium-high SNR conditions,and have high robustness to channel conditions such as CFO. |
PDF Full Text Request