Parameter Estimation Of Spaceborne SAR Mainlobe Signal Base On Convex Optimization

In the field of electronic warfare, space-borne Synthetic Aperture Radar (Synthetic ApertureRadar, SAR) main lobe signal which can be obtained by passive electronic surveillance system isthe large time-bandwidth linear frequency modulation (LFM) signal. Taking American satellite"Lacrosse" as an example, when the space-borne SAR scans targets accurately, the pulse width ofthe main lobe signals can be hundreds of microseconds and the intra-pulse modulation bandwidthcan be several hundred megahertz. According to the Nyquist sampling theory, conventional signalparameter estimation methods for large time-bandwidth LFM signal require ultra-high samplingrate, which not only have a heavy calculation burden but also exceeds the capability of the existingAD converter. Thus, these methods cannot meet the actual demand of electronic warfare. Therefore,research on the problem of the parameters estimation of large time-bandwidth LFM signalsaccording to the characteristics of the space-borne SAR main lobe signal has a great theoretical andpractical application value. Previous work shows that the difficulty of solving problem of estimatingthe parameters of large time-bandwidth LFM signal is to reduce the sampling rate so as to reducethe amount of calculation. This paper proposes a method of temporal segmentation and parameterfitting based on the theory of fractional Fourier transform (FRFT) and convex optimization theoryto study large time-bandwidth LFM signal parameter estimation problem. The main works include:1. Introduce the basic definition of FRFT, focus on analyzing the steps of FRFT for processingLFM signal, which provides a theoretical basis for the estimation problem of solving largetime-bandwidth LFM signals based on fractional Fourier transform.2. Introduce the convex optimization theory, analyze the generalized inequality, Lagrange dualproblem and interior point method and focus on researching sparse representation and basis pursuittheory, which provides a theoretical basis for solving the estimation problem of largetime-bandwidth LFM signal parameters based on convex optimization theory.3. Conventional methods for parameter estimation of non-stationary signal perform poorly. To overcome this flaw, this paper uses the method of time-frequency domain equidistant segmentationand parameters fitting to estimate the parameters of the LFM signal on the basis of FRFT theory.Firstly, this method divides signal equidistantly in time domain. Then it estimates the parameters ofthese sub-pulse signals with FRFT. Finally, the estimated parameters of sub-pulse signals are fused.The simulation results show that this method has a good performance in the parameter estimation oflarge time-bandwidth LFM signal, and can avoid the problem of insufficient computer memorywhich is caused by the large amount of calculation when estimating the signal parameter directly,but it doesn’t reduce the amount of calculation.4. According to function fitting and interpolation theory in convex optimization，the methodanalyses the large time-bandwidth LFM signal with sparse representation and basis pursuit. Firstly,this method divides signal equidistantly in frequency domain by channelized filter bank, anddown-mix the sub-pulse signals into baseband signals. Then the appropriate Gabor atomicdictionary are selected for the sparse representation of the baseband signals. Finally, reconstruct thesub-baseband signals by using the least squares method and estimate their modulation slope. Thesimulation results show that this method can achieve high estimation accuracy and avoid the highsampling rate when estimating the parameters of the large time-bandwidth LFM signal.