SHIPS MATCHING BASED ON AN ADAPTIVE ACOUSTIC SPECTRUM SIGNATURE DETECTION ALG...
Hm3114401445
1. Umoh, Gabriel Etim, Akpan, Aniefiok Otu / International Journal of Engineering Research
and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 1, January -February 2013, pp.1440-1445
Application Of Digital Signal Processing In Radar Signals
Umoh, Gabriel Etim, Akpan, Aniefiok Otu
Department Of Electrical/Electronics Engineering Faculty Of Engineering, Akwa Ibom State University
Ikot Akpaden, Mkpat Enin, Lga Akwa Ibom State, Nigeria
Department Of Physics Faculty Of Natural Andapplied Sciences, Akwa Ibom State University
Ikot Akpaden, Mkpat Enin, L.G.A. Akwa Ibom State, Nigeria
Abstract
There has been explosive growth in pattern while at the receiver, DSP performs many
Digital Signal Processing theory and applications complex tasks, including STAP (space time adaptive
over the years. This research explores the processing) – the removal of clutter, and beam
applications of digital signal processing in Radar. forming (electronic guidance of direction).
A survey on applications in digital signal The front end of the receiver for RADAR is
processing in Radar from a wide variety of areas still often analog due to the high frequencies
is carried out. A review is done on basic involved. With fast ADC convertors-often multiple
approaching models and techniques of signal channel, complex IF signals are digitized. However,
processing for different parameters and digital technology is coming closer to the antenna.
extracting information from the received signal. We may also require fast digital interfaces to detect
The various techniques adopted at different antenna position, or control other hardware.
stages of radar to obtain the target’s signature, is The main task of a radar‟s signal processor
also researched. is to make decisions. After a signal has been
transmitted, the receiver starts receiving return
Keywords: Radar, Doppler, Digital Signal signals, with those originating from near objects
Process (DSP), convolution, Clutter filters, arriving first because time of arrival translates into
Scanning, Synthetic aperture radar (SAR), target range. The signal processor places a raster of
cancellers range bins over the whole period of time, and now it
has to make a decision for each of the range bins as
INTRODUCTION to whether it contains an object or not.
Flexibility and versatility of digital This decision-making is severely hampered
techniques grew in the front-end signal processing by noise. Atmospheric noise enters into the system
and the advent of integrated digital circuitry, high through the antenna, and all the electronics in the
speed signal processors were developed and radar‟s signal path produces noise too.
realized. Radar continued to grow in the recent years
by keeping the future developments in mind and Major blocks of modern radar system
with better digital capability. Significant The major components of modern radar are the
contributions in DSP in Radar have been in MTI antenna, the tracking computer and the signal
processing, Automatic Detection and extraction of generator. The tracking computer in the modern
signal, image reconstruction, etc. A case study of radar does all the functions. By scheduling the
Radar Synthetic Vision System for Adverse appropriate antenna positions and transmitted
Weather Aircraft landing is discussed. In this report signals as a function of time, keeps track of targets
an effort is made to identify the contribution of DSP and running the display system.
in the advancement of Radars.
METHODOLOGIES
RADAR transmits radio signals at distant
objects and analyzes the reflections. Data gathered
can include the potion and movement of the object,
also radar can identify the object through its
“signature” the distinct reflection it generates. There
are many forms of RADAR – such as continuous,
CW, Doppler, ground penetrating or synthetic
aperture; and they are used in many applications,
from air traffic control to weather prediction.
In the modern Radar systems, digital signal
processing (DSP) is used extensively. At the
transmitter end, it generates and shapes the
transmission pulses, controls the antenna beam Fig 1. Block Diagram of a Modern Radar system
1440 | P a g e
2. Umoh, Gabriel Etim, Akpan, Aniefiok Otu / International Journal of Engineering Research
and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 1, January -February 2013, pp.1440-1445
Even if atmospheric attenuation can be neglected, position in azimuth and elevation. When the
the return from a distant object is incredibly weak. radar is tasked with surveying its sector and
Target returns often are no stronger than twice the tracking dozens of targets, there‟s a danger of
average noise level, sometimes even buried under it. either neglecting part of the search sector or
It is quite difficult to define a threshold for the losing a target if the corresponding trace record
decision whether a given peak is noise or a real isn‟t updated in time. Time management serves
target. If the threshold is too high then existing to maintain a priority queue of all the tasks and
targets are suppressed, that is, the probability of to produce a schedule for the beam steering
detection (PD) will drop. If the threshold is too low device. Power management is necessary if the
then noise peaks will be reported as targets, that is, transmitter circuitry runs the danger of
the probability of false alarms (PFA) will rise. A overheating. If there‟s no backup hardware then
common compromise is to have some 90% the only way of continuing regular operation is
probability of detection and a false alarm rate of 10 - to use less power when less power is required,
6
. say, for track confirmation.
It maintains a given PFA known as CFAR, for Countering interference: Interference can be a)
Constant False Alarm Rate. Rather than keeping the natural or b) man-made. Natural interference can
threshold at a fixed point, CFAR circuitry inspects be heavy rain or hail storms, but also varied
one range bin after the other and compares the propagation conditions. Man-made interference,
signal level found there with the signal levels found if created on purpose, is also called jamming and
in its neighboring bins. If the noise level is rather is one of the means of electronic
high in all these (e.g. because of precipitation) the countermeasures.
CFAR circuit will raise the threshold accordingly.
Further tasks of the signal processor are: Detection of Signals
Combining information: Secondary Detection is the process by which the
surveillance radars like those located on airports presence of the target is sensed in the presence of
can ask an aircraft‟s transponder for information competing indications which arise from background
like height, flight number or fuel state. Pilots echoes (clutter), atmospheric noise, or noise
may also issue a distress signal via the generated in the radar receiver. The noise power
transponder. The ground radar‟s signal processor present at the output of the radar receiver can be
combines this data with its own measurements of minimized by using filter, whose frequency
range and angular direction and plots them all response function maximizes the output peak-signal
together on the appropriate spot on the scope. to mean-noise (power) ratio is called matched filter.
Forming tracks: By correlating the data sets We shall discuss the application of digital filtering
which were obtained in successive scan cycles, to matched filters.
the radar can calculate a flight vector which
indicates an aircraft‟s speed and expected Fast Convolution Filter Implementation
position for the next scan period. Airport radars (a) Dual pipeline FFTs matched Filter
are capable of tracking hundreds of targets In this system, FFTs are pipelined and both
simultaneously, and flight safety depends heavily the forward and reverse radix-r FFTs are
on their reliability. Military tracking radars use implemented in hardware. Initial recording of the
this information for gun laying or guiding data is done using input buffer (IB) memory and it
missiles into some calculated collision point. takes „N/r‟ clock pulses to read N data points and „r‟
Resolving ambiguities in range or Doppler input rails. The amount of time „N/r‟ is called as one
measurements: Depending on the radar‟s pulse epoch. It requires three epochs for the first data to be
repetition frequency (PRF), the reading for completely filtered, and is delivered by one epoch
range, Doppler or even both are ambiguous. The thereafter. In the dual FFT systems arbitrary data is
signal processor is aware of this and selects a filtered sequentially with arbitrary reference
different PRF when the object in question is functions selected from reference memory.
measured again. With a suitable set of PRFs,
ambiguities can be eliminated and the true target Drawback
position can be determined. In many applications the same data set be
Ground clutter mapping: Clutter is a collective filtered with several different filters, in this case
term for all unwanted blips on a radar screen. only one forward transform is performed followed
Ground clutter originates from buildings, cars, by several inverse transforms, it is possible to
mountains etc, and a clutter map serves to raise eliminate one of the pipeline FFTs. This is desirable
the decision threshold in areas where known since it would save a large amount of hardware.
clutter sources are located. (b) Single forward FFT matched Filter
Time and power management: Within a The data is first transformed and the result stored in
window of some 60ox40o, phased array radars the temporary storage memory (TSM). The data is
can instantly switch their beam position to any then multiplied by the filter function and inverse
1441 | P a g e
3. Umoh, Gabriel Etim, Akpan, Aniefiok Otu / International Journal of Engineering Research
and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 1, January -February 2013, pp.1440-1445
transformed. This allows multiple readouts of the
forward transformed data for the TSM and multiple
filtering of the same data set; the output of each Input Delay
filter will appear sequentially. [IPP]
Drawback -
Output
The data at the output of the forward FFT
are in digit reverse order, it is then corrected by +
reading the data out of the TSM in digit reversed
order. The second FFT is performed the output is A two-pulse canceller is used if the clutter
placed into an output buffer, and to be read in a bit component (assuming DC only) remains constant in
reversed order from the output buffer. a given range bin and can be eliminated by
It requires five epochs for the first data to be subtracting the output from two successive pulses.
completely filtered, and is delivered by one epoch The transfer function of two-pulse canceller is equal
1
thereafter. to 1 Z . And is equivalent to FIR digital filter
with magnitude response sin ( / 2 ).
(c) Single inverse FFT matched Filter
A single inverse FFT is employed and the data is
In practice, the clutter has a power spectrum that
read from the input buffer in digit reverse order. The
covers frequencies above DC. The two-pulse
data is transformed and stored in TSM in normal
order, and then read out in bit reserve order. canceller will attenuate low frequency components
but may not totally reject clutter. A three-pulse
Drawback
canceller with its transform function equivalent to
Complex conjugation must be performed after each
1 2
transform and stored. The digit reversed access to FIR filter is 0.5 z 0.5 z . This attenuates
the IB is required and the IB maybe quite large further the components near DC.
compared to TSM and may be difficult to
implement. Implementation of Clutter Filter
(d) Reconfigurable FFT matched Filter The returns from the same range bin over
The FFT subsystem switches the inter-stage delay several pulses are linearly combined to form the
lines to realize both forward and reverse transforms. output per IPP, each delay of can be realized
Forward transform: by routing the data through the using shift register.
inter-stage delay memories (IDMs) in decreasing a. The direct implementation of the optimum
order. linear processor with N points requires N
Inverse transform: by sending data through the multiplications per output point. Since a
IDMs in the increased order of size. The total different optimum processor is designed for
memory of each stage is the number of delay lines each Doppler channel, the filter tap weights are
memories. different for each channel.
b. A simpler Suboptimum Processor is obtained
Comparison of 4 matched filter systems by cascading a three-pulse cancellor with a
The relative performance of dual pipeline bank of band-pass filters (implemented by a
matched filter is fastest but requires two complete sliding FFT). N-point FFT requires N log 2 N
pipeline FFTs hence more hardware is required.
Single inverse transform matched filter has better multiplications; only log 2 N multiplications
performance over single forward transform matched are required per each Doppler Channel. Thus
filter and also doesn‟t require a double buffered significant hardware simplifications are
output memory. Reconfigurable FFT matched filter possible with this scheme provided its
is preferred and chosen most of the times, it doesn‟t performance is adequate.
require digit reserved to the IB and also doesn‟t E.g. moving Target Detector
require digital recording of TSM.
MTI Signal Processing
Doppler processing A major task in moving target indicator
Doppler processing is used to filter out (MTI) radar is to obtain a time-domain filter, with
clutter and thereby reveal fast moving targets. Such the introduction of digital technology, these are
filters are implemented digitally, FFT or a set of achieved using digital transversal filters, recursive
transversal filters. Cancellers and few optimized filters and filter banks.
methods are some of the clutter rejection techniques. Adaptive Thresholding and Automatic Detection
Cancellers Digital processing permitted the reference
Clutter rejection filter amounts to the level to be generated/internally from the
design of FIR digital filter with stop-bands to reject observations themselves, thereby permitting more
the clutter frequency component. A simple filter is a sensitive and faster thresholds. Most of the Radars
two-pulse cancelor. employ automatic detection circuits to maintain,
1442 | P a g e
4. Umoh, Gabriel Etim, Akpan, Aniefiok Otu / International Journal of Engineering Research
and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 1, January -February 2013, pp.1440-1445
ideally, a constant false alarm rate (CFAR) by The adaptive threshold CFAR processors is
generating estimates of the receiver output. applicable to situations where the distribution of the
Automatic target detection for a search radar can be processed data (in the no-signal case) is known
achieved by comparing the processed voltage in generally and unknown parameters associated with
each cell to: the distribution can be estimated. It is often
1. A fixed threshold level. implemented as moving or sliding window through
2. Threshold levels based upon the mean amplitude which estimates of the unknown parameters of the
of the ambient interference. interference are formed.
3. A level computed on the basis of partial (a priori)
knowledge of the interference distribution. B. Distribution free CFAR Processors
4. A threshold level determined by distribution-free These provide CFAR characteristics when the
statistical hypothesis testing that assumes no background return has an unknown distribution.
prior knowledge of statistical distribution of the These processors remain insensitive to variations in
interference. the distribution, and generally experiences
In first case a detection decision is made if additional detection loss their CFAR properties
the processed signal ro , is equal or greater than a make their application advantageous.
1. Double Threshold Detector
present threshold. That is, if ro T p , a detection is 2. Modified double threshold Detector
declared. 3. Rank order Detector
The second and third cases represent adaptive 4. Rank-sum double quantizer Detector
threshold CFAR processors. In these processors,
estimates of the unknown parameters of the known C. Scanning Radar Application
distribution of the processed interference are The optimum processor for a pulsed, non-coherent
formed. In the second case can achieve CFAR when waveform on n pulses is a square law detector
the distribution of processed interference is followed by a n pulse non-coherent integrator that
completely described by its mean level. The third uses equal weighting of each detected pulse. The
case forms estimates of the unknown parameters of integrator must not only be realizable in practical
the known (a priori) distribution. sense but also
The fourth case are called nonparametric CFAR 1. provide a small detection loss as possible
processors. These distribution-free process form a 2. provide a means of minimizing losses
test variable whose statistics are independent of the associated with integration sample window
distribution of the input (non-processed) and scanning beam straddle of the target
interference. 3. In track-while scan applications, permit
accurate measurements of the target angular
A. Adaptive Threshold CFAR Processors position.
To adaptive Threshold
CFAR
Square law A/D Single Pulse
IF signal ADDER
Detector Convertor Delay
Scaling
Range gate Constant [kf]
Fig.3 Square law detector
Integrators that are typically configured are sliding Signal Processing in Synthetic Aperture Radar
window. And this requires the storage of data for n (SAR)
inter-pulse periods. The single-loop processor Digital processing has also permitted increased
requires storage of data for the single inter-pulse capability for extracting target information form the
period. Of course, if data memory is somehow radar signal. High resolution SAR provides an
restricted and if performance is acceptable the image of a scene. Radars are used to recognize one
feedback approach is preferred and single feedback type of target from another, with the aid of digital
loop is shown below. processing, inverse SAR (ISAR) produces an image
1443 | P a g e
5. Umoh, Gabriel Etim, Akpan, Aniefiok Otu / International Journal of Engineering Research
and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 1, January -February 2013, pp.1440-1445
of a target good enough to recognize from other o ( vt x1 ) o ( vt x2 )
2 2
j j
classes of targets by extracting the spectrum of a
signal r (t ) e e
cz1 cz1
is the
target echo signal. Interferometric SAR, which uses
two antennas spaced vertically with a common SAR sum of two chirps.
o [ 2 x1 x2 ) vt x2 x1 ]
2 2
system, can provide height information to obtain 3D j
image of a scene. Greater flexibility and real-time 1 e cz1
will be the signal of x1 is
operation suggest digital signal processing in SAR.
considered as a reference point and is called
SAR exploits the probability density of the clutter to
dechirping. The discrimination between the two
detect man-made features by modeling the clutter by
a family of densities and picking the density that targets are performed analyzing x1 at zero frequency
best describes the clutter on local basis. and target x2 analyzed as frequency
Fourier based methods are used for detection of
stationary and moving target detection and 2ov( x1 x2 ) / cz1 . Signal processing of the
identification in reconnaissance SAR. The multiplier of dechirping and FFT is carried out.
computational time of the time domain correlator Since this approximation is done near to x1 the FFT
(TDC) is overcome in the frequency domain. Here processing is limited to small region around the
digital spotlighting principle is used to extract the reference point. Also it is advantageous to band-
target‟s coherent SAR signature. limit the dechirped azimuth signal to the frequency
Step 1 range of interest and do FFT. To avoid aliasing low-
Coherent matched-filtered SAR reconstruction of pass filter is chosen whose impulse response is a
the scene in the presence of foliage is developed by rectangular window and the frequency response is
exploiting the angle and frequency from the target‟s N
coherent SAR signature. sin / N sin . The high frequency-domain
Step 2 sidelobes of the rectangular window is overcome by
With the help of Fourier-based method the three using a different window or an optimal low-pass
dimensional statistic representing the moving target filter.
coordinates and speed is used for moving target Thus this signal processing tends itself to a spotlight
detection. mode of SAR. Small areas of central reference point
Strip Mapping SAR are imaged individually and a large map is
On the incoming data, the pulse compression constructed by piecing together several sub-maps.
matched filtering is done to obtain Range resolution The key feature to be considered to discriminate the
and stored prior to azimuth processing. Data are man-made structure and the foliage is the
processed one row at a time to perform the azimuth characteristics of SAR signature of a man made
compression, since large amount of data are metallic cylinder are different from the nature
collected large memory space is required. For each objects such as tree with the same size of low
range resolution cell different azimuth compression frequencies in the UHF band.
filter is required. Why not shape information to distinguish
Assuming range azimuth resolutions are to be equal targets?
and total range is mapped to z then each memory The resolutions of the reconstructed SAR images are
poor at UHF frequencies. The SAR magnitude
column has M samples and 2z1 / L samples
2
reconstruction of both man-made targets of foliage
(niquist rate sampling) in each row. will appear as blobs. Thus, there is no much
The total storage requirement is discrimination about the information. By the process
2 z 2Rmax of Digital spotlighting the procedure to extract the
Memory coherent SAR signature of a target is done.
L L2 Target detection and identification
Total calculation is that required for the operation of a. Matched Filter Reconstruction
M azimuth pulse compression channels in parallel. Based on the SAR principles governing the
So the total computational time using direct variations of the gain and phase of target‟s coherent
convolution is upper- bounded by radar cross section with respect to the signal
2MRmax frequency and relative speed, angle, we have a
multiplications
L2 general expression for the two-dimensional matched
By using FFT for high speed convolution together filter that is capable of detecting a specified target in
with parallel parallelism reduces the speed the presence of foliage.
requirement significantly. The measured signal S ( ku ) is passed through a
Spotlight mode SAR bank of N, two-dimensional matched filters, and it is
Stretch processing is adopted, i.e., the azimuth passed through SAR wave front reconstruction
resolution is obtained from different Doppler shifts algorithm.
of the individual targets. Let x1 and x2 are the two
targets both at same range z1 then the received
1444 | P a g e
6. Umoh, Gabriel Etim, Akpan, Aniefiok Otu / International Journal of Engineering Research
and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 1, January -February 2013, pp.1440-1445
b. Digital spotlight detection and integration aperture, multiple images are formed. It
identification is then applied to two-dimensional adaptive filters
The image filtered reconstruction requires image used for change detection to these multiple images.
formation for all possible target types followed by a This had an advantage with the different responses
search algorithm to detect the targets. The targets of clutter and targets with aspect angle. Clutter
are identified by search method, extract the coherent returns were expected to show less aspect angle
signatures, and then perform matched filtering on dependency than targets, this property allows angle
the signature of these suspected targets. The dependency than targets, and clutter can be removed
extraction of coherent signature from the coherent by change detection algorithms. This technique has
reconstruction image is done and process is called two distinct advantages over traditional change
digital spotlighting. detection the multiple aperture targets are perfectly
Digital spotlighting is a computer based process that co-registered and only one flight pass is required
isolates the target or a target region in a SAR scene. over a given scene.
To detect and identify targets, digital spotlighting
the reconstructed image f ( x, y ) is the CONCLUSION
reconstruction domain. A stationary target‟s In this research a brief overview of application of
signature is fairly focused and once the target digital signal processing in Radar is presented.
signature is spotlighted, it is transformed into any Matched filter implementation, echo cancellers and
domain of the SAR signal for identification. automatic detection and tracking are discussed in
Detection of targets in foliage separate sections. In most of the modeling, Fast
Three different target detection techniques for UHF Fourier transform is a very commonly used
SAR were discussed in this research. These technique for analyzing and filtering digital signals.
techniques are brief out. Different techniques of detection of targets in
a. Baseline approach foliage are discussed for SAR. The recent advances
It consists of 3 stages with high resolution, Ka-band in signal processing are blended with many more
SAR DATA. The first stage, a simple two-parameter algorithms to present an up-to-date perspective and
constant false alarm rate (CFAR) detector, is a can be implemented in Digital Signal Processor
computationally simple prescreening algorithm that because of their flexibility and the ability to attain
rejects most natural clutter. A target detection is high precisions.
declared if the CFAR algorithm evaluation ratio
exceeds CFAR threshold. This algorithm is followed REFERENCES
by a clustering algorithm, which combines multiple 1. Alan V. Oppenheim, “Application of
detections occurring in a target-sized area into a Digital Signal Processing”, Prentice Hall,
single detection. The combination of these two Inc. Eaves Reedy, “Principles of Modern
algorithms is considered to be the baseline Radar”.
algorithm. 2. Firooz Sadjadi, Mike Helgenson, Jeff
b. Adaptive change detection Radke, Gunter Stein, “Radar Synthetic
The adaptive filters used are two-dimension Vision System for Adverse Weather
extensions of Widow Least mean Squares (LMS) Aircraft Landing”, IEEE Transactions on
filter, used as joint process estimators for noise Aerospace and Electronic Systems”, Jan
cancellation. Each makes use of two images, a 1999, Vol. 35, No. 1, pp. 2-15.
primary, D [presumed to have targets] and a 3. Jeffrey G, Nanis, “Adaptive Filters for
reference x [without targets]. The clutter of the Detection of targets in Foilage”, IEEE AES
reference image is convolved with the weight Systems Magazin, Aug 1995, pp. 34-37.
matrix, W, to predict the corresponding clutter in the 4. Lawrence R. Rabiner and Bernard Gold,
primary image. This primary image is subtracted “Theory and Applications of Digital Signal
from the primary to yield an error image, and the Processing”, Prentice Hall, Inc.
difference is used to adoptively tune each element of 5. Mehradad Soumekh, “Reconnaissance with
the weigh matrix by an approximation. At each Ultra UHF Synthetic Aperture Radar”,
iteration, the sign of the gradient is measured. After IEEE Signal Processing Magazine, July
consecutive sign changes, and scaling, the deviation 1995, Vol. 12, No. 4, pp. 21-39.
is found. By using two-pass change detection with 6. P.E. Blankenship and E.M. Hofstetter,
the adaptive filters, the target-to clutter ratio is “Digital Pulse Compression via Fast
increased. convolution”, IEEE Transactions Acoustic
c. Multiple aperture detection Speech and Signal Processing, Volume
It takes off both the UHF SAR and the angular ASSP-23, April 1975, pp. 1898-201.
diversity of target returns. In order to achieve a high 7. W.M. Brown and L.J. Porcello, “An
cross range resolution with a SAR operating at introduction to Synthetic Aperture Radar”,
FOPEAN frequencies, it is necessary to process the IEEE Spect., Sept. 1969, pp. 952-62.
data over a large integration angle. By splitting the
1445 | P a g e