Research On Blind Demodulation Technology Of OFDM Signal

The technology of OFDM realizes broadband communication signal,effectively improves the transmission rate of the system,and avoids the waste of spectrum resources.In this paper,the blind demodulation of signals in non-cooperative communication background is studied,including signal preprocessing,blind estimation of time parameters and time-frequency synchronization technology,and simulation and analysis are carried out.The following is the main content:Firstly,the complete process of blind demodulation for OFDM signals is given.Bandwidth estimation and signal-to-noise ratio?SNR?estimation in signal preprocessing module are mainly studied.Bandwidth estimation mainly studies algorithms based on Welch spectrum and wavelet transform.The performance of the algorithms under different bandwidth,different frequency offset,different wavelet decomposition layers and different channel conditions is simulated and compared.The main research of SNR estimation is based on AR model and cyclic prefix.The performance of the algorithm under different bandwidth,different frequency offset and different channel conditions is simulated and compared.According to the simulation results,the applicable scenarios and application combinations of different algorithms are given.Secondly,the estimation of time parameters of OFDM signals is studied,including the estimation algorithms based on cyclic prefix and cyclic autocorrelation,and the cyclic autocorrelation spectrum characteristics of OFDM signals of pilot-free and pilot-containing are analyzed.An improved parameter estimation algorithm based on cyclic autocorrelation is proposed to deal with the change of spectral characteristics caused by pilots.The improved algorithm can be applied to OFDM signals without pilot and OFDM signals with pilot.The simulation results show that the performance of time parameter estimation based on cyclic autocorrelation improved algorithm is the best for pilotless OFDM signals.At the same time,the improved algorithm effectively avoids the influence of the second-order periodic peak introduced by pilots on the estimation of time parameters of OFDM signals,and can accurately estimate the time parameters of OFDM signals with pilots.It has good robustness and application value.Finally,time-frequency blind synchronization of OFDM signals is studied,including carrier-frequency synchronization algorithm based on subspace,the algorithm based on cyclic prefix and the algorithm base on cyclic stationary characteristic.A subspace model is built based on subspace algorithm,and the cost function is constructed by the orthogonality between noise-subspace and signal-subspace.The frequency offset estimation process is divided into integer multiple frequency offset estimation and decimal multiple frequency offset estimation.Multilevel search is introduced to narrow the estimation range step by step and reduce the computational complexity effectively.The frequency offset estimation range is the whole system bandwidth.Maximum likelihood algorithm realizes time-frequency synchronization by using the autocorrelation of cyclic prefix.The frequency offset estimation ranges from positive to negative half of the subcarrier spacing.In the fading channel,the maximum time-delay is estimated,and the frequency offset is estimated by using the non-interference cyclic prefix.The algorithm based on cyclostationary characteristics utilizes the cyclostationary characteristics of OFDM signals to achieve synchronization.The frequency offset estimation ranges from positive to negative half of the subcarrier spacing.The simulation results show that the maximum likelihood algorithm can achieve high precision synchronization with fewer OFDM symbols,while the subspace algorithm has a wider range.In fading channels,the time-frequency synchronization performance of the three algorithms is reduced.Among them,the maximum likelihood algorithm is the best.When SNR is greater than 0 dB,the frequency offset estimation NRMSE can be reduced to less than 10^-2.