Research On Localization Algorithm Of Mixed Far-Field And Near-Field Sources In Impulse Noise Environment

Signal source localization is one of the core technologies in array signal processing field,and is closely related to other technical fields,such as oceanography,medicine,etc.The continuous innovation of signal source location technology will promote the development and progress of related fields.According to the distance between the signal source and the sensor,the signal source can be divided into far-field signal source and near-field signal source.In many practical applications,there are both far-field signal source and near-field signal source,such as speaker localization using microphone arrays.At this time,the signal source is called mixed far-field and near-field sources.Most of the existing mixed far-field and near-field sources localization algorithms assume that the background noise is an ideal Gaussian noise.In fact,noise in the real environment has strong pulse characteristics,in such a noise environment,the performance of the mixed far-field and near-field source localization algorithms based on Gaussian noise as background noise will deteriorate or even completely fail.To solve this problem,from the perspective of noise suppression,this dissertation focuses on the estimation of mixed far-field and near-field sources localization parameters under alpha white noise and alpha colored noise.In this dissertation,we first study the localization of mixed far-field and near-field sources based on uniform linear array under alpha white noise,and prove that the fractional low-order covariance matrix and the compressed transform covariance matrix of the far field sources satisfy the Toeplitz Structure.Based on this characteristic,the matrix difference technique is introduced into the mixed far-field and near-field sources localization algorithm under alpha white noise,and the effective separation of mixed far-field and near-field sources is realized.And two revised algorithms based on matrix differencing of the fractional low-order covariance matrix and compressed transform covariance matrix are proposed.Simulation results show that the two algorithms proposed in this dissertation have more robust estimation performance in low GSNR and strong impulse noise environment than mixed far-field and near-field sources localization algorithm based on matrix difference of the second-order covariance matrix.Secondly,uniform linear array can only be used to estimate one-dimensional arrival angles.In this dissertation,uniform circular array is used as the array receiving model to estimate twodimensional arrival angles,and non-circular signals are used as incident signals to improve the estimation performance of the algorithm.And two mixed far-field and near-field non-circular signal localization algorithms based on subspace differencing of the fractional low-order covariance matrix and compressed transform covariance matrix under alpha white noise are proposed.Simulation results show that the two algorithms proposed in this dissertation have better positioning performance in low GSNR and strong impulse noise environment than the mixed far-field and near-field non-circular signal localization algorithms based on subspace differencing of the second-order covariance matrix.Finally,since complete albino alpha noise does not exist in the real environment,this dissertation studies the localization of mixed far-field and near-field sources in the environment of alpha colored noise.Using the characteristics that fractional cumulant and compressed transform cumulant can separate noise and signal and suppress alpha colored noise,a two-stage MUSIC algorithm based on fractional order cumulant and a two-stage MUSIC algorithm based on compressed transform cumulant are proposed under the condition of alpha color noise.Simulation results show that compared with the two-stage MUSIC algorithm based on high order cumulant,the two algorithms proposed in this dissertation have stronger robustness under alpha color noise.