Study Of Algorithm On DOA Estimation Based On Polarization Sensitive Array

The history of array signal process can date back to the forties of 20th century when PLL (phase lock loop) was adopted in antenna tracking in the adaptive antenna technology. It is radar that first involves array processing and the usage of phased array radar in World War II is the precedent of military application. Array signal processing also plays an important role in radio astronomy, sonar, communications, direction finding, the medical field X-ray tomography and seismology field.Research on array signal processing consists of mainly two aspects, including adaptive filtering (adaptive array processing) and spatial spectrum estimation. Spatial spectrum estimation, also known as the DOA (direction-of-arrival) estimation, is primarily used to locate the source of space. Since the seventies of the 20th century, a large number of achievements on spatial spectrum estimation have been made, out of which, two methods are most outstanding. One is multiple signal classification (MUSIC) algorithms proposed by Schmidt. R and the other is Roy's ESPRIT (Estimating signal parameter via rotational invariance techniques) algorithm based on rotation invariance for signal parameter estimation. These two algorithms not only make the leap to modern super-resolution technology possible, but also promote the rise of Subspace Algorithm. Other algorithms can be seen as the deformation of the two algorithms.In the early years of study, array signal processing is started based on the scalar sensor array; however, spatial electromagnetic signals are featured by orientation and polarization. Electromagnetic wave polarization, which is the inherent properties of electromagnetic waves, describes motion characteristics of electromagnetic wave. For the scalar array, it can only get one-dimensional electromagnetic signal information, with outputs reflecting only the amplitude of the signal and not sensitive to polarization information. Application of electromagnetic vector sensor solves this problem; it can access not only the spatial but also the polarization of the signal information with sensor output also a vector. Therefore it can greatly improve the available characteristics of information, which enable the sources resolution processing from different perspectives, improve the capacity of the array for receiving information, and also lay the physical foundation for the development of array signal processing theory.Products of electromagnetic vector sensors have come into market and lots of achievements have been made due to the attention and research made by domestic and foreign scholars. However, most of the estimation algorithms are based on scalar sensors, therefore cannot avoid using complex long-vector matrix model for received data. Though those algorithms based on electromagnetic vector sensor array involve the polarization characteristics of electromagnetic signals, long-vector matrix models cannot express the orthogonal features of electric, magnetic and their propagation direction in the electromagnetic signal. Meanwhile, long-vector matrix in the parameter estimation process involves large amounts of computation, so the hardware requirements are higher.Tensor-based parallel factorization (PARAFAC) algorithm can combine tensor representation and the physical characteristics well; meanwhile, its advantages on computational efficiency and recognition have attracted wide concern from both domestic and foreign scholars. In recent years, signal processing techniques based on quaternion array have developed. That's mainly because through the multiple imaginary of quaternion, operations can be simplified and the efficiency of operation at the same time can be improved. Moreover, the performance of the array of parameter estimation can be greatly improved from stronger Quaternion orthogonal constraints between the vectors.