Multidimensional Signal Spectrum Estimation Based On Compressive Sensing

Spatial spectrum estimation is one of the main problems in array signal processing,and it is also an important component of early warning detection system.When the direction of arrival of the radiation source signal is obtained,combined with other parameters of the target,it can provide a basis for accurate attack of enemy targets.With the improvement of the technology level,the receiving and processing of ultra wideband radio frequency signals,high-speed digital signal processing chip available using the traditional Nyquist sampling theorem,sampling is facing a huge amount of data,it is meet the bottleneck bandwidth of large sampling rate.At present,Compressive Sensing(CS)theory has become a new research direction of signal sampling and compression,in order to alleviate the hardware and software pressure.In this paper,the compressed sensing theory is applied to the processing of narrowband signals and wideband signals,and the corresponding algorithm is proposed by using the sparse characteristics of signals in many dimensions,thus achieving high resolution DOA estimation.On this basis,the traditional array model is improved to optimize the array structure,while reducing the sampling rate,reducing the number of front-end link channels,and effectively reducing the amount of signal processing data.The main works are as follows:1.The basic content of array signal processing is introduced,and the traditional algorithm is analyzed and simulated.The three basic problems of compressed sensing theory are explained.MP algorithm and OMP algorithm are mainly studied.2.In the aspect of narrowband signal processing,the traditional array structure is improved,and the space-time joint compression sampling array(STCS)is proposed.The array can reduce the dimensionality,reduce the rate and compress the sampling in the space and time domain.On the basis of the array structure,2D MVDR algorithm is used to realize the accurate estimation of the space frequency two-dimensional spectrum.The application of space-time joint compression sampling array can reduce the sampling rate and reduce the link path at the front of the array,thus effectively ensuring the real-time performance of the digital signal processing.3.In the aspect of wideband signal processing,in order to solve the DOA estimation problem,a sparse model of joint array covariance data stacked into tensor is improved.In this paper,we construct a over complete basis by using the sparsity of the spatial angle,and we proposed a MMS-OMP algorithm.The sparse representation is used to support the high resolution estimation of the DOA.Experimental results show that the proposed algorithm has better resolution and stability,and avoids the disadvantages of focusing algorithm.Moreover,the use of greedy algorithm can reduce the computational complexity.4.Furthermore,Considering the optimization of the array structure,a single receiver array model is proposed.The array model adopts a time preprocessing technique to sum up the output weights of the array elements and output them to the receiver,which effectively reduces the receiver and received data.In this array structure,a SF-FOCUSS algorithm is proposed for wideband sparse source DOA estimation.The DOA estimation is achieved by using the sparsity of the signal in both spatial and frequency domain.The experimental results show that,under certain conditions,the random weight vector satisfies the Gauss distribution and the Bernoulli distribution can achieve the accurate estimation of DOA,and the performance of the Gauss distribution is better than the Bernoulli distribution.