Research On Wideband Spectrum Sensing In Electronic Reconnaissance

Spectrum sensing, one of the key technologies of electronic surveillance, is aimed at perceiving active bands of electromagnetic signals efficiently and accurately, which lays the foundation for follow-up parameter estimation in electronic surveillance. Whereas, with the rapid development of military information technology, electronic reconnaissance receivers are forced to implement efficient, reliable and real-time perception for more complex electronic signals. According to Nyquist sampling theorem, it requires much higher speed collectors and much larger storage overhead, so that, traditional narrowband spectrum sensing technologies have already can't meet the demand of perception. Compressive sensing acting as a revolutionary technology brings the dawn for this bottleneck with its perfect knowledge system and broad industry applications. In view of the electromagnetic signal in the frequency domain performance of sparse, it is possible to combine compressive sensing and spectrum sensing to accomplish the effective perception of wideband spectrum sensing in electronic reconnaissance, which has a strong practical significance.At present, wideband spectrum sensing based on compressive sensing is to reconstruct the original time-domain signal at first, and then do Fourier transform for its spectrum. In this paper, a different approach, from the perspective of direct perception spectrum to do in-depth research. The main contents as follows:1. It analyzes the traditional narrowband spectrum sensing technology and its challenges. On the basis of a detailed introduction of compressive sensing theory, it also discusses the wideband compressed spectrum sensing characteristics. For common hidden terminal problem, multi-node cooperative spectrum sensing methods is introduced, and its model and data fusion method is described in brief to set the stage for expansion of research work in this paper.2. It does in detail the research on reconstruction algorithms of compressive sensing. Especially for four classic GP algorithms including OMP, SOMP, Co Sa MP and SP, this paper carries on the analysis and simulations to compare their performance of advantages and disadvantages.3. It presents three wideband spectrum sensing methods based on multi-coset sampling model. The first one is based on autocorrelation matrix reconstruction. In the case of unknown the original signal sparse degree, it greatly simplifies the reconstruction process using the linear relationship between original signal autocorrelation vector and the observed sample one, thus overcomes the non-real time shortcoming caused by iteration in traditional wideband spectrum sensing. Besides, sparse-ruler sampling not only ensures the uniqueness of LS solution, but also improves the accuracy of perception. The second one is a singlenode spectrum sensing based on subspace method. The core idea of the method is converting the spectrum sensing problem into the subspace parameter estimation problem, which can realize accurate wideband spectrum sensing under low SNR by first computing the number of active channels to reduce the computation of estimating the active band set. The third one is a cooperative wideband spectrum sensing based on subspace method. It firstly obtains spatial diversity gain by DPAST algorithm, and then completes wideband spectrum sensing by using the orthogonality relations of the subspace, which can solve the hidden terminal problem and achieve more reliable and accurate wideband spectrum sensing.