Blind Estimation Of Spreading Sequences For Non-cooperative Direct-spreading Signals

Due to its many advantages,such as high spectrum utilization rate,low probability of intercepting and good confidentiality,direct sequence spread spectrum(DSSS)has been widely used in the fields of mobile communications and electronic warfare in recent decades.In the case of cooperative communication,each part of correspondence member can perform down conversion,demodulation,de-spreading and other operations according to the agreed signal parameters.However,in the case of non-cooperative communication,there is only a small amount of priori information.If want to decipher the original information,we must estimate the PN sequence.Therefore,the research on the estimation of PN sequence is urgent and significant.In the basis of detection and parameter estimation of DSSS signals,this paper focuses on the blind estimation of PN sequences.The main work and innovation of this article are as follows:The method of PN sequence estimation based on matrix decomposition is studied.Firstly,the classical feature decomposition method is analyzed,and simulation is carried out to discuss its estimation effect and its advantages and disadvantages.Based on the discussion of the relationship between feature decomposition and singular value decomposition(SVD),a joint estimation method of PN sequence and information symbol based on SVD is proposed and applied to the recovery of PN sequence of long code signals.Aiming at the problem that the long code signal is concatenated after estimating each PN sequence,the sign is uncertain.According to the shift-add property of the m sequence,a third-order correlation function(TCF)is used to find a decision function.When the code rate is less than 20%,the peak value is accurately estimated to determine the symbol of the PN sequence.According to the matrix form of the direct-spread signal,which can be obtained by multiplying PN sequence and information symbol sequence,and the PN sequence and the information symbol sequence have the nature of low rank characteristics,we compose a low rank matrix decomposition(LRMF)estimation method.This method first derives a cost function according to the likelihood function,and then finds the minimum value of this cost function through the optimization method.When the cost function obtains the minimum value,the obtained matrix is the estimation result of the PN sequence and the information symbol sequence.It can be known from simulation experiments that the method based on low-rank matrix decomposition can effectively jointly estimate the information sequence and the PN sequence under the condition of low SNR.Aiming at the problem that PN sequence estimation is greatly affected by noise and interference in impulsive noise scenarios,after discussing the projection approximation subspace tracking(PAST)method,this paper proposes a new cost function based on the M estimation method,and On this basis,the recursive algorithm of projection approximation subspace tracking method based on minimum M estimation is derived.First,blind synchronization is performed on the received signal,then the synchronized signal is sampled in chip period and segmented in PN sequence period,then iterative method is used to solve the signal subspace when the cost function takes the minimum value,and finally the eigenvectors of the signal subspace can be used to estimate the symbol function to estimate the PN sequence.It can be seen from the simulation experiments that the method proposed in this paper has better performance in the impulsive noise environment than the SVD method and the subspace tracking method using recursive least squares estimation.