Study On Theory And Method Of Electromagnetic Vector Sensor Array Signal Processing In Noise Environments

The electromagnetic vector sensor array signal processing is a new emerging subdiscipline in array signal processing domain. In recent years, DOA and polarization parameters estimation of signals using array of electromagnetic (EM) vector sensors(VS's) has become a hotspot, which is widely used in radar, sonar and communication etc.The research on DOA and polarization parameters simultaneous estimation using an array of electromagnetic vector sensor is of great significance in practical application and academy. People have used traditional scalar sensor array on DOA estimation for a long time, and have massive academic achievements. Through this system we cannot obtain the integrated electric and magnetic fields information but only the information on one field component. The array element output only reflected the receive signal intensity and the absolute phase. The lack of information will eventually affects the algorithm performance. However, EMVS can provide complete electric and magnetic fields, which is composed of three orthogonal electric dipoles and three corresponding orthogonal magnetic dipoles. Not only it can use its array geometry structure to acquire the spatial information of the signal, but also can acquire all polarization information. So it has higher information acquire ability than the common arrays due to the additional polarization information. Compared with traditional scalar sensors, electromagnetic vector sensor array has a better performance: steady detective ability, stronger anti-jamming ability as well as higher space resolution.At present the mass of method used in electromagnetic vector sensor array signal processing domain is based on supposition of the white Gaussian noise and the steady signal. But in practical environment, the white Gaussian noise supposition may not always be true, and many signals exited are not steady. For example, many man-made signals , such as BPSK, FSK, AM signals, exhibit in the cyclostationarity, and LFM signals used in radar applications. As a result, the performance of subspace-based DOA and polarization estimation techniques may degrade when dealing with non-stationary signals.The electromagnetic vector sensor array has rich redundant information. Using this characteristic and with the concrete characteristic of the signal, we research the estimations of DOA and polarization in colored noise.Firstly, considering the non-stationary signals such as LFM signal, were widely used in many signal processing applications, and one of the most important application is on radar and sonar. About the non-stationary signals, time-frequency analysis is combined with EMVS array processing to develop a DOA and polarization estimation method. Spatial polar time-frequency(TF)distribution (SPTFD) is generalized by using one or two EMVS and the TF distribution. We debelop TF-MUSIC based on Wigner-Ville Spectral Distribution and TF-MUSIC baced on Wigner-Hough Translation for estimation the DOA and polarization parameters. The proposed methods have the ability to select the signal of interesting, and are suitable for non-stationary signals.Then, considering a kind of special non-stationary signal-cyclostaionarity signal, we have analysised the DOA and Polarization parameters estimation of the coherent signal source. The existence of coherent signal source make the performance of many metheods serious decline or even failure. By exploiting the polarization information and the cyclostationarity of the signal, we propose the following methods: (1) Polarization smoothing method for the 2-D DOA estimation of cyclostationary coherent signals; (2) Polarization smoothing method for the DOA estimation of cyclostationary coherent signals. All of the above approaches not only solve the DOA estimation of coherent sources, but also have the capability of suppressing noise,and have signal-selectivity.