Font Size: a A A

One-bit Compressed Sensing Radar Imaging Research On Key Technologies

Posted on:2022-04-26Degree:DoctorType:Dissertation
Country:ChinaCandidate:Z WangFull Text:PDF
GTID:1488306323962839Subject:Electromagnetic field and microwave technology
Abstract/Summary:PDF Full Text Request
Synthetic Aperture Radar(SAR)plays a very important role in the field of Radar observation.When signal processing is carried out according to the classical Nyquist the-orem,in order to achieve high resolution SAR imaging,the ADC in the system needs to collect a large amount of echo data.And these large quantities of data often bring great challenges to the receiving system.The theory of Compressive Sensing(CS)provides a feasible way for us to improve this problem.Therefore,CS-based SAR imaging has been widely concerned by the academic community.Furthermore,some literature points out that when the observation matrix meets some conditions,the signal can be recovered stably only through the symbolic information of the echo signal,which is the one-bit CS theory.When the theory is applied to SAR imaging,the ADC pressure can be greatly reduced without sacrificing range-direction resolution.At the same time,its quantiza-tion form only uses the comparator,which itself has the advantages of high speed,low cost and low power consumption.Therefore,the research on one-bit CS-SAR is of great significance.Although the application of one-bit CS theory to SAR radar imaging has the above advantages,there are also some new problems.Firstly,compared with the traditional matched filtering algorithm,the observation matrix dimension in the one-bit CS-SAR model is larger,which brings the problem of high computational complexity.In addition,due to the motion error of radar platform,the defocus problem of reconstructed scene still exists in one-bit CS-SAR.There are also off-grid problems caused by grid division and imaging problems of non-sparse scenes.In view of these problems,this dissertation gives the corresponding solutions.The first research content of this dissertation is weighted L1 norm one-bit CS-SAR imaging algorithm based on approximate observation operator.The algorithm combines the existing matched filtering algorithm with FFT operation and the one-bit CS-SAR model.By means of the existing approximate observation idea,the precise observation matrix with larger dimension is replaced with the approximate observation operator with lower complexity.In this fusion method,the advantages brought by CS can be retained to avoid the influence of the side lobe effect of matched filtering,and the advantages of low computational complexity of matched filtering algorithm can be retained.At the same time,in order to approximate the L0 norm regularization problem well and avoid NP difficult problem,we adopted the weighted L1 norm one-bit CS-SAR imaging model.The second research content of this dissertation is the research of one-bit CS-SAR imaging technology without phase transformation.Multi-bit SAR imaging focus-ing technology has been extensively studied.However,due to the particularity of one-bit SAR,the existing multi-bit algorithms cannot be applied directly.This reason is ex-plained below in this dissertation,and a phase-free strategy is proposed to achieve image focusing in single-bit CS-SAR imaging.The third research content of the dissertation is the one-bit CS-SAR imaging Off-grid problem.Radar echo data is the result of convolution between the signal emitted by radar and the scattering distribution in the radar observation area.When using CS theory to carry out sparse modeling for observation scenes,the common practice is to discretize the region first,that is,to divide the grid and form a discrete dictionary.Then,it is con-sidered that the target scene can be sparsely represented under the dictionary.However,in practice,the locations of the scattered points to be solved can be arbitrary,that is to say,they are not always on the preset grid points,thus there will be off-grid prob-lem.We also take the grid offset as an unknown quantity and estimate it together with the scattering distribution.Specifically,in the vicinity of the grid,the influence of grid offset is combined to do first-order Taylor expansion,the consistency constraint is com-bined with the one-bit nonlinear model for the data,and the sparsity constraint is added to construct the iterative solution algorithm.At the same time,we add a self-checking mechanism to reduce the error of adjacent point estimation for adjacent targets.The fourth part of the dissertation is the sparse representation of one-bit CS-SAR imaging.Compressed sensing theory requires that the reconstructed scene must be sparse.When the scene under consideration is not sparse,the sparse representation work should be done first to apply the compressed sensing theory.When considering sparse representation,this paper takes the amplitude and phase of the observed scene into consideration respectively by referring to the existing studies in the field of multi-bit.In practical applications,the phase of the observed scene may be random,but the corresponding scene amplitude can often be sparsely represented.In the simulation,dis-crete cosine transform(DCT)is used for the sparse representation of amplitude.The final experimental results verify the effectiveness of the proposed algorithm.
Keywords/Search Tags:One-bit Compressive Sensing, Synthetic Aperture Radar Imaging, One-Bit Quantization, Sparse regularization, Approximate Observation, Phase Error Correction, Off-Grid Problem, Amplitude Sparse Representation
PDF Full Text Request
Related items