Welcome to TFTSA
for Radar Applications




General Information on Radar Signal


Radar transient signals are usually transient in time domain and wide-band in frequency domain. Since classical methods ( such as Fourier transform ) have difficulty to handle transient signal analysis, time-frequency or time-scale analysis is more suitable for preserving high-frequency contents of the information carried by transient signals. Comparing with the Fourier transform, the time-frequency/time-scale transform of a transient signal may detect and locate rapid changes of the signal better than Fourier transform. One of the important applications of time-frequency/time-scale transform is the detection and extraction of unknown radar signals in noise. The localization of the time and frequency by time-frequency/time-scale transform makes it possible for de-noising, signal detection and extraction in the time-frequency domain. Feature extraction in the time-frequency domain refers to the identification of specific feature attributes from the transform domain. To recover and isolate specific signal features, the transform pairs must be not only effective, but also computationally efficient.


General Information on Radar Images


Radar transmits electromagnetic waves to a target which consists of a number of point scatterers and receives the scattered waves from the target. The scattering properties of the target describe the features of the target. The integrated effect of the scattered fields can be measured directly by the radar, thus, the spatial distribution of the reflectivity corresponding to the target can be reconstructed by a radar processor. The distribution of the reflectivity is referred to as the radar image of the target. The target's reflectivity is usually mapped onto a range (or down-range) and cross-range plane and is viewed as a radar image of the target. The range is the dimension along the radar line-of-sight to the target. The cross-range is the dimension transverse to the line-of-sight.

High resolution radar images are always demanded. The range resolution is directly related to the bandwidth of the transmitted radar signal. Stepped-frequency waveforms and linear frequency-modulated chirp waveforms are examples of wide-band radar signals which are commonly used in radar imaging systems. The cross-range resolution is determined by the antenna beamwidth, which is inversely proportional to the length of the antenna aperture. To achieve high cross-range resolution without using a large antenna aperture, synthetic array processing is widely employed. Synthetic array radar processing coherently combines signals obtained from sequences of small apertures to emulate the result from a large aperture.

Synthetic array radar includes both synthetic aperture radar (SAR) and inverse synthetic aperture radar (ISAR). Traditionally, SAR refers to the situation where the radar is moving and the target is stationary; ISAR refers to the geometrical inverse in which the target is moving and the radar is stationary. For ISAR, the synthetic aperture is formed by coherently combining signals obtained from a single aperture as it observes a rotating target. The rotation of the target emulates the result from a larger circular aperture focusing at the rotation center of the target.

ISAR uses Doppler information to obtain the cross-range resolution. Due to the target's rotation, which can be characterized as a superposition of pitch, roll, and yaw motions, different parts of the target have slightly different velocities relative to the radar and, hence, produce slightly different Doppler frequencies in the radar receiver. The differential Doppler shift of adjacent point scatterers can be observed in the receiver; therefore, the distribution of the target's reflectivity can be measured by the Doppler spectrum. The conventional method to retrieve Doppler information is the Fourier transform.


Radar Range-Doppler Image


The objective of radar processing is to estimate the target's reflective density function from received baseband signal samples, the so-called frequency signature. If the target's range is known exactly and the velocity and acceleration of the target's motion are constant and known exactly over the imaging time duration, then the extraneous phase term of the motion can be exactly removed. Therefore, the reflective density function of the target can be obtained simply by taking the inverse Fourier transform of the phase compensated frequency signature. The process of estimating the target's range and removing the extraneous phase term is called focusing or gross translational motion compensation. Then, the inverse Fourier transform can be used to construct the reflective density function of the target. For SAR, the motion compensation is facilitated by measuring the actual motion of the radar platform. In ISAR, the actual motion can be measured by a range-tracker, or estimated by a motion compensation algorithm which estimates motion parameters and compensates motion with respect to the target's range, velocity, acceleration and other higher order terms.

The radar processor uses the frequency signatures as the raw data to perform range processing and Doppler processing. Range processing functions as a matched filter for use with pulse compression, which removes frequency or phase modulation and resolves range. For each range cell, the range profiles constitute a new complex time history series. Then, Doppler processing takes Fourier transform for the time history series and generates its Doppler spectrum, or profile. By combining the Doppler spectra for all range cells, finally, the range-Doppler image is formed. Therefore, the radar image is the target's reflectivities mapped onto the range-Doppler plane.


Motion Compensation


Motion compensation is a very important step to achieve the requirements of using Fourier processing and to have a clear radar image.

Conventional motion compensation is a gross compensation for the whole target. It performs mainly range and the Doppler tracking. While a target is moving smoothly, conventional motion compensation is good enough to produce a clear image of the target.

However, when a target exhibits complex motion, such as pitching, yawing, rolling, or maneuvering, conventional motion compensation for the whole target is not sufficient to produce an acceptable image for viewing and analysis. In this case, more sophisticated motion compensation procedures for each individual scatterer such as polar reformatting and sub-aperture approach are needed. It keeps each scatterer within its range cell and maintains constant Doppler frequency shift for each of them. Thus, the Fourier transform may be applied properly to construct a clear image of the target.

In case a target exhibits significant maneuvering, even sophisticated motion compensation is still not sufficient, and the residual of the motion is still large. With large motion residuals or phase errors, individual scatterers may still drift through their range cells; thus, the Doppler spectrum may still be time-varying and image is still blurred.


Time-Frequency Processing


Conventional approach to retrieve Doppler spectrum is the Fourier transform. To use the Fourier transform adequately, some restrictions must be applied: the Doppler frequency contents of the data should not change within the time duration of the data. If the Doppler contents do change with time, the Doppler spectrum by using the Fourier transform becomes smeared and, thus, the cross-range resolution is degraded. However, the restrictions can be lifted if the Doppler information can be retrieved with time-frequency transforms which do not require stationary Doppler spectrum. Replacing the conventional Fourier transform with the joint time-frequency transform, a 2-D range-Doppler Fourier frame becomes a 3-D time-range-Doppler cube. By sampling in time, a time sequence of 2-D range-Doppler images can be viewed. Each individual time-sampled image from the cube provides not only superior resolution but also excellent noise performance with enhanced signal-to-noise ratio.


Data Files and Formats


The data files are written as a complex 2-D matrix. Each row of the matrix corresponds to a range cell, and each column corresponds to a pulse in a burst. The data files are pulse compressed and motion compensated. By taking 1-D FFT of the time history pulses for every range cell, range-Doppler image can be reconstructed. These data files are saved as two formats: To represent complex data, the odd columns represent the real-part and the even columns represent the imagienary-part of the data. The total number of rows equals to the number of burst.

Points of Contact:

The radar data may be made available to individual researchers and organizations through the WWW.


File Parameters:


For radar pulse data:

The data consists of 10 complex (inphase and quadrature) signals each. There are 3 tests from each of 4 sources called A2, CCC2, F2, and H2. Each ASCII file contains a sequence of 20 concatenated signals of length 180 samples. The first 10 signals are the real-part (inphase) and the last 10 signals are the imaginary-part (quadrature). There is a total of 20x180 = 3600 samples in the data. The SNR of this data is higher. The purpose of analysis is to classify the source of the signals.

For noisy radar pulse data:

The data consists of a complex (inphase and quadrature) signal with 3120 samples. The SNR of this data is lower. The purpose of analysis is to detect and to classify the source of the signals.

For simulated noisy chirp data:

The data consists of 15,000 samples. The SNR of this data is about -5dB. The purpose of analysis is to detect and to extract the signal embedded in noise.

For simulated B-727 data:

The Stepped Frequency Radar operates at 9GHz and has a bandwidth of 150 MHz. For each pulse, 64 complex range samples were saved. The file contains 256 successive pulses. The Pulse repetition frequency is 20KHz. Motion compensation and range processing have been applied to the data. Radar image can be reconstructed by taking 1-D FFT of 256 pulses for each range sample. The B727s is high SNR with fluctuation in velocity which causes the image blurring. The B727s0 is high SNR without blurring. The B727sn is low SNR without blurring.

For simulated MIG-25 data:

The Stepped Frequency Radar operates at 9GHz and has a bandwidth of 512MHz. For each pulse, 64 complex range samples were saved. The file contains 512 successive pulses. The Pulse repetition frequency is 15KHz. Basic motion compensation processing without polar reformation has been applied to the data without pulse compression. Radar image can be reconstructed by taking 2-D FFT of the data.

For real B-727 data:

The Stepped Frequency Radar operates at 9GHz and has a bandwidth of 150 MHz. For each pulse, 128 complex range samples were saved. The file contains 128 successive pulses. Motion compensation and range processing has been applied to the data. Radar image can be reconstructed by taking 1-D FFT of 128 pulses for each range sample.


File Access
The files listed below can be obtained by your browser, the links in the table below can be used to download the desired files.

Files at http://airborne.nrl.navy.mil/~vchen/data
File NameFormat SizeDimensionsDescription
PULSE01.MAT MATLAB34 KB180 x 10 Radar pulses (complex)
PULSE01.DAT ASCII 60 KB180 x 20 Radar pulses (real+imag.)
PULSE02.MAT MATLAB34 KB180 x 10 Radar pulses (complex)
PULSE02.DAT ASCII 60 KB180 x 20 Radar pulses (real+imag.)
PULSE03.MAT MATLAB34 KB180 x 10 Radar pulses (complex)
PULSE03.DAT ASCII 60 KB180 x 20 Radar pulses (real+imag.)
PULSE04.MAT MATLAB34 KB180 x 10 Radar pulses (complex)
PULSE04.DAT ASCII 60 KB180 x 20 Radar pulses (real+imag.)
PULSE05.MAT MATLAB51 KB3120 x 1 Noisy radar pulses (complex)
PULSE05.DAT ASCII102 KB3120 x 2 Noisy radar pulses (real+imag.)
CHIRPM5DB.MAT MATLAB119 KB1 x 15000 Radar chirp signal in noise
CHIRPM5DB.DAT ASCII238 KB1 x 15000 Radar chirp signal in noise
B727S.MAT MATLAB264 KB64 x 256 Simulated ISAR data with blurring (complex)
B727S.DAT ASCII519 KB64 x 512 Simulated ISAR data with blurring (real+imag.)
B727S0.MAT MATLAB264 KB64 x 256 Simulated ISAR data without blurring (complex)
B727S0.DAT ASCII519 KB64 x 512 Simulated ISAR data without blurring (real+imag.)
B727SN.MAT MATLAB264 KB64 x 256 Simulated ISAR data without blurring (complex)
B727SN.DAT ASCII519 KB64 x 512 Simulated ISAR data without blurring (real+imag.)
RECON1D.M Matlab M-file for 1-D processing
MIG25.MAT MATLAB519 KB64 x 512 Simulated MIG-25 with blurring (complex)
MIG25.DAT ASCII1 MB64 x 1024 Simulated MIG-25 with blurring (real+imag.)
RECON2D.M Matlab M-file for 2-D processing
B727R.MAT MATLAB264 KB128 x 128 Real B-727 data (complex)
B727R.DAT ASCII519 KB128 x 256 Real B-727 data (real+imag.)


Last update 8 Nov. 1999
U.S. Naval Research Laboratory
Radar Division
This is an official U.S. Navy web site
webmaster
Policy: Terms of Use, Privacy Policy