Font Size: a A A

Study On Time Delay Estimation Methods And Applications In The Presence Of Impulsive Noise

Posted on:2008-03-01Degree:DoctorType:Dissertation
Country:ChinaCandidate:W H LiuFull Text:PDF
GTID:1118360218953629Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
Time Delay Estimation (TDE) is a popular issue in signal processing areas. Noise is one of the main problems in TDE. In practice, there is a kind of non-Gaussian noises with obvious impulsion in which cases the TDE algorithms based on second-order statistics of Gaussian model degenerate significantly. Therefore, robust methods in TDE need to be researched. A sort of generalized Gaussian distributions, Alpha-stable distribution, can more precisely describe the statistics of impulsive noise. Based on the impulsive noises of the Alpha-stable distribution model, this paper puts the emphases on the robust single path TDE methods and its applications in biological medical area, and also robust high-resolution multipaths TDE methods. The researches of the paper are helpful to the extension of the random signal analysesand processing theories from second-order statistics to FLOS statistics and upgrade the overall level and the robustness of the TDE algorithms and its applications. Therefore, the paper has great significances both in theories and applications.The main work of the paper is as follows:(1) A robust adaptive fractional TDE method, referred to as LMPFTDE, is proposed. By using impulsive noise modeled Alpha-stable distribution, ETDGE is generalized based on minimum dispersion (MD) criterion. LMPFTDE can work well both in Gaussian and impulsive noise environments. In order to avoid parameter selection, based on non-linear transformation, it proposes sigmoid correlation algorithm that fits for time delay invariance TDE; based on zero order statistics, it proposes the concepts of logarithmic wide sense stationary and logarithmic ergodic random process, and also an adaptive TDE method, referred to as NZOSTDE, with geometric power normalization step factor. These two methods are more suitable for much impulsive noise environments. Using NZOSTDE to detect EP latency delay, good results can be obtained.(2) With regard to TDE in impulsive noise multipath environment, based on FLOS, it proposes three robust and high resolution TDE algorithms which could break through the limit of correlation algorithm. They are robust multipath TDE algorithm in transformation domain referred to as FLOCCS-ESPRIT, p-order fractional correlation EM algorithm referred to as P-EM, and robust and high resolution multipath TDE referred to as RHMTDE. The FLOCCS-ESPRIT regards the fractional lower-order cross-covariance spectrum of multipath signal as an equivalent time sequence, then makes use of ESPRIT to obtain high resolution multipath TDE. P-EM and RHMTDE transfer a multi-dimension nonlinearity optimization problem to a multiple one-dimensional optimization problem from different angles. Based on EM frame, P-EM constructs complete data by signal decomposition, iterates respectively by p-order fractional correlation, and obtains the TDE of each path under least mean p-norm (LMP) criterion. Based on FLOS and MD, RHMTDE makes use of relax iteration to acquire TDE of each path. This research can improve position technology and detection robust in ground-penetrating radar under multipath impulsive environment.(3) Using Alpha-stable distribution to model impulsive noise, it proposes robust TDE algorithm based on LMP blind channel identification referred to as BCILMP and robust adaptive eigenvalue decomposition referred to as RAED. The main idea of BCILMP and RAED is to assemble two input signals so that the eigenvector corresponding to minimum eigenvalue of covariance matrix is the estimation of two channels. TDE is adaptive obtained under LMP criterion. RAED algorithm uses fractional order operation for error signal and input vector. Both of them improve robustness of AED algorithm based on Gaussian model.(4) It studies the characteristics of gastric electrical activity (GEA) with noise. Gastric slow wave decides the propagation of GEA. Due to the effects of spikes and/or motion artifacts, the slow wave often contains sharp impulsion; therefore Alpha-stable distribution is more suitable than Gaussian in GEA research. Based on BCILMP, the propagation velocities of simulative slow wave with noises are estimated. Compared with LMSTDE, BCILMP provides well TDE both in Gaussian and impulsive noise environments. And then, BCILMP is applied to estimate the propagation velocities of GEA in four gastroparetic patients, and the results are mostly consistent with the evaluations of the doctor. It can be the assistant judgments for doctors and also provides the references to select the parameters of gastric electrical stimulation to cure gastric disasters.
Keywords/Search Tags:Impulsive Noise, Alpha Stable Distribution, Fractional Lower Order Moments, Time Delay Estimation, Gastric Electrical Activity
PDF Full Text Request
Related items