Research On Freqeuency-Hopping Signals Parameters Estimation And Sorting Algorithms

Frequency Hopping(FH),as one of the main technical means of spread spectrum communication,has the advantages of strong anti-interference ability and low probability of interception.It is widely used in modern military communications.The anti-reconnaissance and anti-interference characteristics of FH technology pose a severe challenge to military communication reconnaissance missions.The reconnaissance tasks of FH signals mainly include the parameter estimation of FH signals and the sorting of FH signals.In recent years,many researchers pay more attention to the parameter estimation of FH signals and the sorting of FH signals in the field of communication countermeasures.According to the problems in the current FH signals reconnaissance technology,this thesis makes an in-depth study on the parameter estimation and sorting methods of FH signal.In this thesis,aiming at the problems of high complexity and low accuracy in estimating the parameters of asynchronous FH signals and synchronous orthogonal FH signals by existing algorithms,which lead to the reduction of sorting performance,the main research work of this thesis is including:(1)Detect the hop timings and pattern in the case of single array element,and use the detected hop timings and pattern to complete the splicing of asynchronous FH signal.Then,the asynchronous FH signals are sorted;(2)Underdetermined conditions are more in line with the actual reconnaissance environment in the case of multiple array elements.Underdetermined blind source separation technology is used to solve the sorting problem of synchronous orthogonal FH signals,fully considering the time-varying characteristic of the mixing matrix of FH signals,and the estimated mixing matrix and the recovered source signal are used to obtain the direction of arrival(DOA)of each segment of FH signals.Then,according to the difference of DOAs,the synchronous orthogonal FH signals are spliced,and the synchronous orthogonal FH signal are sorted.Aiming at the problems in the existing algorithms,effective methods are proposed in this thesis.The main work of the thesis is as follows:In the case of single array element,the existing algorithms complexity of detecting hop timings and pattern increases nonlinearly with the increase of receiving data length,which leads to the low timeliness of asynchronous FH signal splicing.A method of estimating hop timings and pattern based on hop timing segment detection is proposed.The method divides the received FH signals into several segments,and uses OMP method to determine whether the segment contains the hop timing.For fragments that do not contain hop timings,the frequency information is stored directly.For the segment containing hop timings,the sparse model of the segment is constructed by using the preset frequency set and the dual sparsity of the FH signal in Time-Frequency(TF)domain.Finally,the parameters are updated by using the method of alternated direction method of multipliers,and then the sparse matrix and hop timings are obtained.This method reduces the amount of data processed by subsequent sparse methods by detecting the segments containing hop timings,achieves the purpose of reducing the computational complexity,improves the timeliness of asynchronous frequency hopping signal splicing,and then improves the performance of asynchronous FH signals sorting.The simulation results verify the performance of the proposed algorithm.If the FH pattern does not belong to the preset frequency set,the existing methods for detecting the hop timings and the FH pattern are inaccurate in the case of single array element,which leads to the low accuracy of the asynchronous FH signals splicing.A method for detecting the hop timing and the FH pattern based on the frequency set correction is proposed.Firstly,the frequency bias vector is set and the modified frequency dictionary matrix is derived,and then the received FH signals are represented by the modified frequency dictionary matrix and the sparse matrix.Then,according to the smoothing characteristics of the FH signal,the probability distribution model of each parameter is constructed,and the iterative rules of each parameter are derived by using the variational Bayesian inference.After the algorithm converges,the sparse matrix and hop timings are obtained.The algorithm updates the frequency deviation during the iterative process and corrects the frequency dictionary matrix,which overcomes the shortcomings of the existing method to set a fixed frequency set,resulting in the performance degradation of the sparse matrix and hop timings.It improves the estimation accuracy of hop timings and FH pattern.It achieves the purpose of improving the accuracy of the asynchronous FH signal splicing,and then improves the performance of asynchronous FH sorting.The simulation results verify the performance of the proposed algorithm.In view of the poor performance of existing algorithms in estimating the mixing matrix of FH signals and recovering source signals,which leads to the degraded performance of synchronous orthogonal FH signals,two kinds of synchronous orthogonal FH signals sorting methods based on underdetermined blind source separation are proposed.The proposed method takes into account the time-varying characteristic of the mixing matrix of FH signals,and divides the received FH signals into segments by the detected hop timings,thus simplifying the time-varying mixing matrix problem of FH signals into a non-time-varying situation.The first method is to propose a single source detection(SSP)criterion and estimate the mixing matrix for FH signals when several SSPs exsit in the time-frequency domain.The second method is based on tensor decomposition for underdetermined mixed matrix estimation of FH signals,which relaxes the restriction on the sparsity of FH signals when the number of SSPs is small in TF domain.Under the condition of estimating the mixing matrix of FH signals,considering the redundant energy at TF point,the source signal recovery method based on subspace projection is improved by judging the number of FH signals at each TF point,which improves the recovery performance of FH signals.Using the frequency-azimuth information in the estimated mixing matrix and the recovered frequency hopping signal,the DOA of each FH segment is obtained,and the synchronous orthogonal FH signals are spliced by using the difference of DOAs of different segments.Since the estimation performance of DOA increases with the improvement of the estimation performance of mixing matrix and source signal,the aim of improving the accuracy of synchronous orthogonal FH signals splicing is achieved,and then the performance of sorting synchronous orthogonal FH signals is improved.The simulation results verify the performance of the proposed algorithms.