Font Size: a A A

Research On Algorithms Of Source Localization Based On Compressed Sensing Theory

Posted on:2018-12-08Degree:MasterType:Thesis
Country:ChinaCandidate:J YangFull Text:PDF
GTID:2348330512484877Subject:Electronic and communication engineering
Abstract/Summary:PDF Full Text Request
As one of the most important issues in signal processing,array signal processing is widely applied in military and civilian field.This thesis focuses on the algorithms of source localization.According to the distance between source and receiver array,a source can be classified into far-field(FF)source or near-field(NF)source.FF source is just parameterized by deirection of arrival(DOA),while a NF source is characterized by both its DOA and range.It can reduce the number of required snapshots and estimate parameter more precisely by applying compressed sensing(CS)to source localization.Main works of this thesis are as follws:(1)Research on model of FF and NF source localiazation and classical methods.The author studies the model of array receive data and carries out simulation of classical source localization algorithms and simulates Cramer-Rao bound in three source localization scenarios.(2)Research on the model and algorithms of source localiazation based on CS.The performace of one convex optimization and two greedy algorithms for DOA estimation is validated in this thesis and simulation of one mix-field source localization method with comination of cumulant and CS of is conducted.Main innovations of this thesis are as follws:(1)One modified of DOA estimation method based on OMP algorithm is achieved.The author analyses the uncertainy of application of CS in DOA estimation through studying the correlation among steering vectors,and researches the influence of covariance on sparse reconstruction.And the author also raises an OMP-based algorithm,which adaptively exploits prior information and obtains obvious advantage with low SNR and a few snapshots.(2)A new localization method for mix-field multi-band signals is proposed.Two classical CS-algorithms are modified to improve spatial spectrum and a new method for mix-field and multi-band source localization based CS is proposed,which obtains more accurate estimates in joint sparse model.The source number it locates can equal to the sensor number and it still works for pure NF sources.
Keywords/Search Tags:source localization, compressed sensing, far field, near field
PDF Full Text Request
Related items