Research On Wideband Signal Detection Techniques Under Sub-nyquist Sampling

With the popularization and application of cognitive radio,communication reconnaissance systems must be able to detect signals quickly and efficiently.Compared with Nyquist sampling,sub-Nyquist sampling which can solve the problems of high sampling rate in wideband sparse signal,gradually become a research hotspot of broadband signal acquisition.Based on sub-Nyquist sampling,existing wideband signal detection algorithms have the disadvantages of weak robustness and poor timeliness,and rarely involve Direct-Sequence Spread-Spectrum signals.In view of the current situation,this paper makes a research on the mentioned problems by constructing detection statistics with sub-Nyquist samples.For wideband signal robust blind detection under sub-Nyquist sampling,an algorithm based on eigenvalues energy with compressive samples is proposed.Firstly,cyclic autocorrelation matrix is analyzed and operated eigenvalue decomposition to construct detection statistics on the condition that cyclic frequency equals zero.Secondly,the relation between compressive samples and cyclic autocorrelation matrix is established,and the latter is reconstructed by the former with the sparsity of cyclic spectrum density.Threshold is finally acquired by analyzing the distribution of detection statistics.Simulation results show that the proposed algorithm has better detection robustness and relatively low computational complexity.For wideband signal speedy blind detection under sub-Nyquist sampling,two algorithms with coprime samples are put forward,which are based on coprime sampling and improved generalized coprime sampling respectively.In the first algorithm,cyclic autocorrelation matrix which cyclic frequency equals zero is firstly estimated to construct detection statistics.With the threshold acquired by coprime samples,detection is finally implemented.Furthermore,the feasibility of other detection methods and practicability of generalized coprime sampling are delved.Simulation results demonstrate that the proposed algorithm is a better solution to speedy blind detection with lower sampling rate and less complexity.On the basis of the first one,the second algorithm makes improvement on the following aspects.Firstly,improved generalized coprime sampling is proposed,which can increase average number to elevate the estimation accuracy of cyclic autocorrelation matrix without adding data length.Secondly,a fast coordinate combination computation method is presented,which effectively decreases computational complexity by recurrence.Simulation results show that the algorithm performs well when data length is small.For the feasibility analysis of known specifications Direct-Sequence Spread-Spectrum signal detection,a carrier frequency estimation algorithm based on cyclic spectral density section envelope is firstly proposed with Nyquist sampling,as well as a fast detection scheme which consists of three parts.Simulation results verify the effectiveness of the algorithm and scheme.In the cases of sub-Nyquist sampling,reasons why current studies have no breakthrough and feasibility analysis are given.