Research On Super-resolution Direction Finding Algorithm For Improving The Resolution Of Spatial Spectrum Estimaiton

With the widely used of the passive direction finding technique in the field of electroniccounter measures, navigation, aviation, monitoring and control, et al, the direction findingtechnique based on spatial spectrum estimation had been developed much because of its highaccuracy and super resolution performance. After the deeply investigate of the scholars athome and abroad with decades, the spatial spectrum estimation theory and technique has beenmore mature, but a lot of issues are still deserve us to research. The objects of this paper is todiscuss several questions of passive detection finding technique based on spatial spectrumestimation, mainly in aspects of the source number detection method in colored noise,improving resolution performance of the direction of arrival (DOA) algorithm, the problem oftoo many array elements and the channel phase mismatch error in the actual system. The mainresearch work of this paper is summarized as followsThe colored noise will cause the divergence of the eigenvalues so as to make thedetection performance of the information theoretic criterion decease seriously, even fail. Inorder to solve this problem, a source number detection method used in colored noise isproposed. Though a straight line of the total least-squares (TLS) fitting with the eigenvaluesafter eigendecomposition, and using the characteristics of the miniumdistance sum of squarebetween the eigenvalues and its fitting straight line, a novel estimation function is constructed.By finding the minimum value of the target function, the source number estimation valuecould be acquired. The method can get a higher detection probability in colored noise, thesimulations and the experiment data proved it corretly.Focus on the problem that the resolution performance of multiple signal classification(MUSIC) algorithm decease seriously under the background of low signal-to-noise and smallsnapshots, the MUSIC algorithm based on improving subspace project is proposed. Thisalgorithm takes full use of information contained in subspace. By weighting noise subspacewith the corresponding eigenvalues after revised the result superimposes the weighted signalsubspace projectioin method, a kind of new spatial spectrum is obtained. Though searchingthe maximum of the spectrum function, the DOAs could be got. Simulations and experimentsresults confirm that the proposed estimation method can not only keep the high-resolution ofnoise subspace, but also improves the robustness of the algorithm.When the distance of two incident signals is close, the two spectral peaks will overlapeach other, which lead the resolution failure of MUSIC algorithm. As a result, a modified MUSIC algorithm using the second derivative of spectrum is proposed. The method isreprocesses the MUSIC algorithm spectral function. Using the characteristics that the secondderivative of the MUSIC function will cause negative peaks, some negative peaks are shownaround the original DOAs pointedly while MUISC spectrum does not make peaks there. Bysearching the minimum of the function, the DOAs could be acquired. This method can get abetter estimation accuracy and resolution under low SNR and small snapshots comparing withMUSIC algorithm. And it has no special requirements on the array form. Compared with theMUSIC algorithm, the amout of calculation increases slightly, the effect is little.In actual direction finding system, the size of the antenna dish is limited, the number ofantennas can not be too much, the array form is constrained of other components and themultiple antennas will cause the problem of channel phase mismatch error. Focus on theseproblems, a novel MUSIC method based on array rotating is proposed. Only two antennas areused in this method. By rotating the antenna dish and setting a regular time interval to receivedata, the two antennas can receive signal information at different location, which is equivalentto change a pair of antenna into multiple pairs of antennas on a uniform circular. As a result,this method can achieve the DOA estimation with only two antennas, which can increase theflexibility of the antenna displayed on the dish and the channel phase mismatch is greatlyreduced too. Simulations prove that the novel method has the similar accuracy and resolutionwith MUSIC algorithm and it is feasible.