| The reason why frequency hopping signal(FHS)is widely used in military communications is because of its good confidentiality and strong anti-interference.However,the high sampling rate and the huge amount of data brought by high bandwidth have brought great challenges to the FHS detection.In order to solve this problem,this thesis introduces compressive sensing(CS)theory to study the parameter estimation algorithm of FHS in the compressed domain.The work mainly includes:(1)Because most of the existing single FHS parameter estimation algorithms have high complexity,so that they are not suitable for the situation where the amount of data is large,and some algorithms have low parameter estimation accuracy and poor adaptability under low signal-to-noise ratio(SNR),besides,the influence of duty ratio on parameter estimation is not considered.Therefore,in this thesis,the received FHS is first processed with equal intervals,and the carrier frequency is estimated by the maximum cosine method after compressing,then the time-frequency(TF)pattern is obtained through the TF matrix.Finally,the atomic decomposition(AD)method is used to locally estimate the initial estimated hopping time,and then the accurate hopping speed is estimated.By segmenting the received FHS and adopting the idea of overall initial estimation and then local accurate estimation,the complexity of the algorithm is reduced.And the advantage of the AD method with high parameter estimation accuracy is fully utilized,improving the parameter estimation accuracy.Besides,it is suitable for the situation where the duty ratio exists and has good performance under low SNR,and solves the effect of the inverted ? phenomenon generated by some modulation methods such as BPSP and QPSK on the AD method.(2)In view of the problems that low stability and poor adaptability under low SNR of the traditional sparse linear regression(SLR)algorithm,this thesis improves the SLR algorithm.Firstly,by rearranging the coefficient matrix,the TF matrix containing TF information is obtained,then the maximum value of the TF matrix column vector is extracted and the difference is adopted to estimate the hopping time.The simulation results show that the improved SLR algorithm has better stability and SNR adaptability,and the parameter estimation accuracy is significantly improved.(3)Because most of existing estimation algorithms are only applicable to single FHS scenarios,little research has been done on estimation of multiple FHS,and the existing algorithms have high complexity or poor accuracy.This thesis proposes a parameter estimation algorithm for multiple FHS in the compression domain.Firstly,the problem of sparsity estimation of the multiple FHS in the compression domain is studied,and the idea of the spectrum sensing is used to estimate the number of each segmented signal without reconstructing the source signal.Secondly,the improved OMP algorithm and TF matrix estimation are used to obtain initial estimation parameters.Finally,based on the initial estimated TF pattern,an adaptive sliding window(SD)algorithm is proposed to accurately estimate the hopping time of FHS.Simulation experiments show that,compared with the traditional sparsity estimation algorithm,the proposed sparsity algorithm has higher estimation accuracy,and the algorithm complexity is much lower than other algorithms,and the sparsity of multiple FHS can be accurately estimated at a low SNR.Besides,in order to reduce the calculation amount of the algorithm,the OMP algorithm is improved while ensuring the estimation accuracy of the algorithm.In addition,compared with the traditional SD method,the accuracy of parameter estimation of the adaptive SD method proposed is significantly improved,and with the reducing of the number of iterations,the calculation amount of the algorithm is decreased,which solves the problem that the traditional SD method cannot take into account both the parameter estimation accuracy and the algorithm complexity. |
PDF Full Text Request