Research On Non-data Aided Technologies In Demodulation For OFDM/OQAM Systems

There are some deficiencies in the traditional OFDM systems.For example,the cyclic prefix reduces power efficiency,and OFDM systems have large out-of-band radiation.OFDM/OQAM systems with well time-frequency localization avoid the above deficiencies.Non-data aided demodulation refers that without the prior knowledge of transmitting signals,the receiver processes recieved signal and recovers the useful information.Non-data aided demodulation is of great significance in practical application scenarios such as the radio monitoring and management,the fault detection of communication systems,and the communication countermeasure.Synchronization and channel estimation are both key technologies in demodulation,which can greatly reduce the bit error rate over the multi-path fading channel environment.The research on non-data aided synchronization and channel estimation is of great significance in promoting OFDM/OQAM schemes into practical applications.This paper proposes a joint estimation method of symbol timing offset and carrier frequency offset based on the second-order cyclic cummulants of OFDM/OQAM signals,a simple and effective channel order estimation method based on the cross relation error function and a channel identification method based on the frequency domain cross relation error function.The major achievements and contributions are outlined as follows:(1)The current non-data aided synchronization methods have poor performance over multipath channel environments,or need the prior knowledge of transmitting signal power,multipath fading channel parameters,prototype filter function,and subcarrier weights.Avoiding the above deficiencies,this paper proposes a joint time-frequency synchronization method,which is applicable to the multi-path fading channel environments,and need the prior knowledge of prototype filter and subcarrier weights.In the proposed method the multipath fading channel is modeled as multiple flat Rayleigh fading sub-channels.The ratio of cyclic-cumulants at different cyclic frequencies eliminates the effect of carrier frequency offset and multipath fading channel parameters.The ratio of cyclic-cumulants at different time delay eliminates the effects of symbol timing offset and multipath fading channel parameters.Based on the above characteristic,the symbol timing offset and carrier frequency offset are estimated.For completeness,the modified Cramer-Rao Bound is derived.Simulation results show that the proposed method is effective and feasible,and improves the performance under the condition of low signal to noise ratio.(2)The current non-data aided channel order estimation methods are almost computationally complex and have poor performance under the condition of low signal to noise ratio.Avoiding the above deficiencies,this paper proposes a simple and effective channel order estimation method.In the proposed method,the matrix composed of received signals is operated on eigenvalue decomposition.Then these eigenvalues are arranged in descending order to construct the eigenvalue function.The eigenvalue function has larger attenuation when the argument is less than the real channel order,and smaller attenuation when the argument is more than the real channel order.According to the above characteristic,the eigenvalue function is operated on the logaritmim and the second-order difference to construct the cost function.The cost function has local minimum in the real channel order.Simulation results show that the proposed method is effective and feasible,and improves the performance under the condition of low and middle signal to noise ratio.(3)It is complex to get the channel parameters from the recieved signals in time domain.Therefore,this paper proposes a channel identification method based on error function of the cross relation algorithm in frequency domain.The proposed method regards the error function of the cross relation algorithm in frequency domain as the objective function.Then the objective function is modelled as an unconstrained optimization model,in which the steepest descent method is used to get the optimal value.Simulation results show that the proposed channel identification method is effective and feasible,and improves the performance under the condition of middle and high signal to noise ratio.