Study On Theory And Method Of Signal Parameters Estimation With Polarization Sensitive Array In Complex Noise

Array signal processing which is an important branch of signal processing focuses onthe way of extraction of both characteristics of received signals and features of sourceswhich generate signals. It utilizes sensors array to receive and process the correspondingdata from received signal. Polarization Sensitive Array (PSA) signal processing is anewly emerging subdiscipline in array signal processing, which has been widely usedin field related in radar, sonar, communication, navigation, land and ocean exploring,seismic, ECM(Electronic Counter Measures) and biomedical signal processing etc. PSAhas been paid more and more attention since several past years and now is the hotspotin Array signal processing.Estimation techniques of the direction-of-arrival (DOA) of signals impinging uponan antenna array have been the subject of intensive study. Traditional scalar sensorarray (TSSA) on DOA estimation has two main disadvantages as shown below althoughit has a long research literature and also has made massive e?ort in academic region:first, the array element cannot obtain the complete electric and magnetic fields infor-mation but only the one field component information; second the array element outputcan only re?ect received signal intensity and its absolute phase. These faults eventuallydegenerates the algorithm performance. Compared with TSSA, however, Electromag-netic Vector Sensor (EMVS) can provide complete electric and magnetic fields, whichare composed of three orthogonal electric dipoles and three corresponding orthogonalmagnetic dipoles. The array formed by EMVS or trimmed vector sensors is called vectorsensors array (VSA). In other hand, as the VSA can acquire the polarization state of theimpinging signals, it is often called polarization-sensitive array (PSA). Not only it canuse its array geometry structure to acquire the spatial information of the signal, but alsocan acquire all polarization information. It has higher information acquire ability thanthe (TSSA) due to the additional polarization information. The PSA performs much better than the TSSA as shown followed: steady detective ability, stronger anti-jammingability as well as higher space resolution. The research on joint estimation of DOA andpolarization parameters using PSA is of great significance in practical application andacademy.The signal parameters, such as DOA, polarization, temporal frequency etc, playimportant roles in many applications, and their estimations have been and continue tobe a topic of great research interest. Based on narrative above, this paper mainly exploresthe corresponding theory and method of multi-parameters estimation of PSA signals, andalso discusses the two-dimensional(2-D) DOA and polarization parameters estimation ofcyclostationary signals in complex noise, multi-parameters estimation based on airbornevector sensor array, 2-D DOA and polarization parameters estimation of coherent sources,multi-parameters estimation based on the PSA in alpha-stable distribution noise, and2-D DOA and polarization parameters estimation of LFM signals.It is worth noting that many modulated signals exhibit a cyclostationarity propertyin practical application such as radar, sonar and telecommunications, correspondingto the underlying periodicity arising from the carrier frequencies and baud rates. Re-cently, signal cyclostationarity has been widely considered for DOA, resulting in manytechniques that inherently exhibit a signal-selective property. For the problem of cy-clostationary signals parameters estimation, the third chapter proposes the followingmethods:(1)Based on EMVS, this dissertation proposes HOCS-MUSIC, HOCS-ESPRITand HOCS-TLS-ESPRIT method for estimating the DOA and polarization parame-ters. These proposed methods are suitable in the occasion where the aperture is lim-ited. (2)Fourth-order cumulant based method using polarization-sensitive uniform lineararray(PS-ULA). This dissertation proposes HOCS-MUSIC and HOCS-ESPRIT method.By su?ciently utilizing the cyclostationarity of the signals and array manifold of PS-ULA, these proposed methods can e?ectively suppress the additive stationary noises withany distribution and interfering signals with di?erent cyclic frequencies.(3)Based on anarbitrary array of EMVS, a new method is proposed for multi-parameters estimationof the cyclostationary signals. The algorithm can be used not only for arbitrary arraygeometries but also separate the signals by su?ciently utilizing the cyclostationarity ofthe signals. Meanwhile, they can suppress the colored noise and interference.Higher-order statistics and higher-order cyclic statistics as the mathematical toolshave many excellent properties. The fifth chapter introduces fourth-order cumulant andfourth-order cyclic cumulant approaches to the airborne electromagnetic vector sensorarray (EMVSA) signal processing. Based on airborne EMVSA, the proposed methods as follows:(1)Fourth-order cumulant based method, which gives the joint estimation of the 2-D DOA, frequency and polarization parameters of the non-Gaussian narrowband signals,and can suppress the white Gauss noise without any frequency information.(2)Fourth-order cyclic cumulant based method, which gives the joint estimation of the 2-D DOAand polarization parameters. Take full advantage of the algorithm is the cumulativevolume of high-cycle characteristics of the signal not only has more choices, but alsomore additive noise.For Coherent Source DOA and polarization parameters of the estimation analysis,the existence of coherent sources seriously decline in the performance of algorithm andsometimes even leads to failure. Based on the existing TSSA processing algorithmsmake use of coherent signals in the airspace of the information, but the polarization ofthe signal information and characteristics of cyclostationary signals are not fully utilized.Cyclostationary signal properties can not only e?ectively suppress interference and ef-fects of stationary noise, but also has the ability to select desired signals. In addition,the polarization of the signal information for the e?ective coherent decoupling solutionprovides a totally new means. Chapter V of this article proposed that, with the help ofa uniform rectangular array on the basis of EMVS, one can get a decoupling solutionthrough processing the information of the polarization combined with characteristicsof cyclostationary signals. The approach details are shown below: (1)Spatial smooth-ing(SS) method for coherent 2-D DOA and polarization estimation;(2)Polarimetric an-gular smoothing(PSA) method for coherent 2-D DOA estimation. All of the aboveapproaches not only solve the DOA estimation of coherent sources, but also have thecapability of suppressing interference and noise, and have signal-selectivity.In many practical applications, there are many non-Gaussian signals and noises,such as environmental noise,atmospheric noise, wireless channel noise, radar clutter andso on. It will be very inaccurate if they are treated as just Gaussian model.αstabledistribution is a very important non-Gaussian random distribution., which has extensiveuse in submarine sound signal, atmosphere and biomedicine signal processing. In re-cent years,αstable distribution as well as relevant fraction low-order statistic are usedabroad in acoustic and radar signal processing, speech signal processing and medicinesignal processing and so on. In the sixth chapter, it has been discussed that the multi-parameters estimation method which applies the polarization-sensitive uniform circulararray (PS-UCA) based onαstable distribution noise, and propose the method for 2-DDOA and polarization estimation using PS-UCA based on fraction lower-order statis-tic, has the capability of suppressingαstable distribution noise with the background of αstable distribution noise.Time-frequency analysis (TFA) methods are well suited for non-stationary time-varying signals and hence yield good performance superior to common subspace methods.The seventh chapter studies time-frequency analysis methods and their application inPSA processing to estimate DOA and polarization parameters of non-stationary signals.Spatial polar time-frequency distribution (SPTFD) is generalized by using PSA and thetime-frequency (TF) distribution. it has been developed as two principal approaches.One approach based on wigner-viller (WV) spectrum MUSIC uses a single EMVS. Theother based on cross WV spectrum MUSIC adopts dual-EMVSs for estimating the DOAand polarization parameters. The proposed methods have the ability to select the desiredsignal, suppress additive noise and interference, and are also suitable for stationarysignals.At last, a summarization of whole dissertation is concluded in the final chapter.