Array Signal Parameter Estimation Based On Fourth-Order Cumulants

Array signal parameter estimation is an important research subject of array signal processing. It has a wide-ranging application vision in such as radar, communication, sonar, etc..With the advantages of blindness to Guassian noise and array extending capability, fourth-order cumulant has been widely applied to array signal processing in recent years. Focusing on the applications of fourth-order cumulants in array signal parameter estimation, I studied direction of arrival estimation, joint frequency, two-dimensional arrival angles and polarization estimation, joint frequency, two-dimensional arrival angles and range estimation based on fourth-order cumulants in this dissertation. The main work and contributions of the dissertation are in the several aspects as follows:1. MUSIC algorithm and ESPRIT algorithm for direction of arrival estimation are studied.The definition and properties of high-order cumulant are given. The relation between fourth-order cumulant and second-order statistics in array signal processing is analized. The array extending capability of fourth-order cumulant is discussed. MUSIC-like algorithm and virtual-ESPRIT algorithm based on fourth-order cumulant for direction of arrival estimation are studied. Simulations of MUSIC-like algorithm are provided to illustrate the array extending property of fourth-order cumulant.2. An improved algorithm based on fourth-order cumulant is presented for joint frequencies, two-dimensional arrival angles and polarizations estimation of incoming multiple independent non-Gaussian narrow-band signals. The algorithm needs only a subarray consisting of four short dipole antennas configured in a certain fashion, other elements of the array may have arbitrary and unknown response and geometry. Compared with those algorithms based on fourth-order cumulant published, it can release the constrains on the array and offer a saving in hardware. Computer simulations that confirm its effectiveness are provided.3. A computationally efficient algorithm based on fourth-order cumulant is presented for joint frequencies, two-dimensional arrival angles and ranges estimation of incoming multiple independent non-Gaussian narrow-bandnear-field sources. It can give the the signal parameter estimation without any spectral searching and only needs five guiding sensors configured in a certain fashion without any constrains on other elements of the array. The algorithm can be employed to arbitray Guassian noise inviorement since fourth-order cumulants are employed. Simulation results show that the proposed algorithm works well in different conditions.