Advanced High Resolution Methods for Radar Imaging and Micro-Doppler Signature Extraction

Lead Research Organisation: University of Strathclyde
Department Name: Electronic and Electrical Engineering

Abstract

Imaging radars are airborne or spaceborne radars which generate a reflectivity map of an illuminated area through transmission and reception of electromagnetic energy. Among many types of microwave sensors, special attention has been paid in the past to Synthetic Aperture Radar (SAR) because of its high spatial resolution, day or night all weather operational capabilities. With its fine two-dimensional resolution capability SAR has evolved to satisfy a variety of applications for both civilian and military users. These applications centre on target imaging and terrain mapping. A target is a specific object of interest that the radar illuminates. The typical target is man-made and consists of multiple scattering centres. An imaging radar system must distinguish between single and multiple scatters located in close proximity. Resolution is, nominally, the minimum distance needed between adjacent scatters to separate them in the image. Fine resolution provides the capability to image a complex object or scene as a number of separate scattering centres. This type of image provides detailed information to detect, characterize, and identify specific objects of interest. Because of the importance of object identification in military applications, much development effort has been directed at improving radar resolution. Military SAR applications include intelligence gathering, battlefield reconnaissance, and weapons guidance. Civilian applications include topographic mapping, oil spill monitoring, sea ice characterization and tracking, agricultural classification and assessment, lands use monitoring, and planetary or celestial investigations. Normally imaging radars provide a two-dimensional representation of a scatterer in the illuminated volume with no resolution or positioning of scatterer in the third dimension. Generally, we speak of monstatic (the transmitter and receiver are co-located) radar resolution in the range and cross-range or azimuth directions. Bistatic radars, where the transmitter and receiver are positioned in different physical positions have several operational advantages. In particular such bistatic systems help to increased receiver survivability while minimising receiver cost. Furthermore when one or both of the platforms are manoeuvring in an non linear planar path allows the resolution to be computed in the 3rd dimension thus facilitating the acquisition of target height information as well range and cross range resolution.When a radar interrogates a moving target it is traditional to exploit the target's Doppler for identification and characterisation. If the target possesses additional rotational, vibration or other internal motions then these induce additional spectral components separate from the main Doppler. These are termed microdoppler components and reside as additional sidebands around the main Doppler. A human walking or running will exhibit microdopplers due to swinging arms and leg movements. A military tank will exhibit microdopplers due to the wheel tracks while a helicopter and engine target will exhibit key microdoppler components. The use of time frequency signal representation such as the short time Fourier transform and wavelet analysis has been used to examine these microdopplers in the past. Good quality microdoppler signatures are important in new automatic target identification and recognition systems. Quality is directly related to the extracted microdoppler resolution extracted.The aim of this work is to explore new signal processing techniques which can be used to improve the resolution of the imaging radars algorithms and microdoppler signature extraction. The work will derive new mathematical relationships for bistatic spotlight SAR image formation and microdoppler signature extraction based on the Fractional Fourier transform and empirical mode decomposition. An FrFT compute engine will be realised and the algorithms will be tested on simulated and real data.

Publications

10 25 50
publication icon
Carmine Clemente (Co-Author) (2011) Characterization of Vibrating Targets

publication icon
Carmine Clemente (Co-Author) (2010) Fractional RDA and Enhanced FrCSAfor SAR Imaging

publication icon
Carmine Clemente (Co-Author) (2010) Fractional Range Doppler Algorithm for SAR Imaging

publication icon
Carmine Clemente (Co-Author) (2012) Vibrating Micro-Doppler signature extraction

publication icon
Clemente C (2012) Approximation of the Bistatic Slant Range Using Chebyshev Polynomials in IEEE Geoscience and Remote Sensing Letters

 
Description A bistatic (Radar Transmitter and Receiver not located at the same place) data simulator was also developed in order to provide bistatic raw data for different working configurations. A technique was developed to improve resolution (ability to separate two closely spaced targets) in monostatic (co-located Transmitter and Receiver) SAR images and was applied to both RDA (Range Doppler Algorithm) and the CSA (Chirp Scaling Algorithm) algorithm. The resulting novel methods called the FrFDA and the eFrCSA were tested revealing good results confirming the usefulness of the Fractional Fourier transform for radar imaging.

The bistatic SAR focusing problem was investigated and a new approach to obtain the polynomial approximation of the slant range function was derived using the Chebyshev approximation. The new polynomial replaces the conventional Taylor series approximation leading to a better approximation of the point target spectrum. The results show good performances in terms of point scatterer impulse response for different flight configurations. The Chebyshev polynomial approximation has been applied to the most accurate bistatic point target spectrum derived using the 2D principle of stationary phase. The Chebyshev approximation has been used to obtain a more accurate decoupling of the range-azimuth frequency components. The proposed approach replaces the coefficients of the polynomial Taylor series expansion of the azimuth-range frequency coupling term with the ones obtained using the Chebyshev polynomial approximation.

The obtained coefficients are then used to obtain an analytical formulation of the bistatic point target spectrum. The new approximation improves the accuracy of the bistatic point target spectrum by the 75%.



The Micro-Doppler literature in SAR was reviewed and a model for the bistatic SAR micro-Doppler signature for vibrating target was derived and tested on simulated data for millimeter Wave SAR and X-band SAR systems. A feasibility study on the application of super resolution techniques to the Space Surface Bistatic SAR (SS BSAR) with Global Navigation Satellite Systems illuminators was carried out with positive results. The different extraction techniques for the micro-Doppler signature from Radar signals were investigated. A new extraction method based on Singular Spectrum Analysis was developed and applied to simulated SAR data.



A feasibility study into the use of a Passive (Only the Transmitter has an Active element) Bistatic Radar (PBR) using Global Navigation Satellite Systems as an Illuminator of opportunity for micro-Doppler analysis of Helicopters signature was conducted. The studied system used one of the GPS illuminator and a passive receiver to perform micro-Doppler analysis of the helicopter rotor blades. The budget analysis for this kind of system was developed showing the feasibility of the exploitation of this PBR for micro-Doppler analysis. The results on simulated data show the possibility to use this cheap and innovative system to exploit the GNSS signal for defence purpose in the Automatic Target Recognition field.



The micro-Doppler effect introduced in SAR images from rotating wind turbine was investigated. The presence of active wind turbines can affect radar imaging systems that monitor ground and sea areas where windmills are often be present. The blade motion introduces Doppler effect on the returned echoes to the SAR platform that can introduce undesirable effects in the final image. The effect of active wind turbines on focussed SAR images was modelled and the effect for different aspect angles and rotating blade velocities were analysed through simulations.



The micro-Doppler extraction technique based on the Singular Spectrum Analysis was applied to the passive bistatic radar using an GPS illuminator for helicopter classification, showing the effectiveness of this technique also for this case of interest.



The research then focussed on the classification problem of micro-Doppler signatures. First the existing approaches for the feature extraction have been investigated. Second a quick and efficient technique for the extraction of features has been developed, tested and implemented on a real time DSP processor. The resulting system was shown to be computationally efficient and robust on radar data for a range of real world data sets.The analysis of the dataset and the application of the classification algorithm led to high confidence classification results.
Exploitation Route The outcome from this work has potential use in many aspect of commercial or non-commercial radar & sonar technologies. The main exploitation route is to further industrially engaged academic research and development. The work was used as basis for Workpackage elements of a successful bid to the Dstl-EPSRC UDRC Phase II : Signal Processing in a Networked Battlespace. A consortium comprising Loughborough, Surrey, Strathclyde and Cardiff (LSSC) began the £3.6m 5-year research program on April 1st 2013.



The relevance of the work and potential to specific applications include:



Fractional Fourier Based Range Doppler and Chirp Scaling high resolution SAR imaging algorithms has application in radar imaging systems improving the achievable minimum resolution and the signal to noise ratio. Specific applications are advanced target recognition based on radar images and target detection in noisy environment.



Bistatic SAR slang range approximation using Chebyshev polynomials has application in the field of bistatic SAR focussing where the accuracy and efficiency of the image focussing stage is fundamental. Specific application are frequent imaging, covert area monitoring and stealth target detection.



Micro-Doppler signature analysis from bistatic SAR is relevant for applications such as automatic target recognition.



Micro-Doppler signature extraction from SAR data using Singular Spectrum Analysis has relevant application for target classification in order to make the target features independent from the scene.



Passive bistatic radar for helicopter classification has interesting applications such as air traffic control, covert monitoring and target imaging.



Analysis of the effect of Wind Turbines in SAR images has important applications in the Electronic Counter Measures and Electronic Counter Counter Measures. i.e. it was demonstrated how small targets could be hiding in the ghost targets generated by the wind turbines rotor motion.



Extraction of micro-Doppler from passive Radar using Singular Spectrum Analysis has relevance in application for target classification in order to make the target features detectable.



Improvement of the bistatic point target spectrum based on the 2D PSP using Chebyshev polynomial approximation has application in the field of bistatic SAR focussing where the accuracy and efficiency of the image focussing stage is fundamental. Specific application are frequent imaging, covert area monitoring and stealth target detection.



Feature extraction technique for micro-Doppler classification purposes, embedded classifier with SVM on DSP, classification of micro-Doppler signatures has strong relevance in real time ATR of targets with micro-motions (i.e. human, animals, helicopters, etc. ).
Sectors Aerospace/ Defence and Marine,Agriculture/ Food and Drink,Digital/Communication/Information Technologies (including Software),Electronics,Environment,Security and Diplomacy

URL http://www.strath.ac.uk/eee/research/cesip/signalprocessingalgorithmsapplicationsgroup/