Mini Project ON Air Borne Radar: Department of Electronics and Communication
Mini Project ON Air Borne Radar: Department of Electronics and Communication
Mini Project ON Air Borne Radar: Department of Electronics and Communication
ON
AIR BORNE RADAR
DEPARTMENT OF ELECTRONICS AND COMMUNICATION
NOIDA
INSTITUTE OF ENGENERING AND TECHNOLOGY
SUBMITTED TO SUBMITTED BY
PANKAJ KUMAR
UTSAV BAJPAI
ASHISH MALIK
CERTIFICATE
V.K.Pandey
(Head of Department)
E.C.E
ACKNOWLEDGEMENT
We owe a great many thanks to a great many people who helped and
supported us during this project.
One of the earliest radar simulators was built by R. K. Moore at the University of Kansas
in the late 1950’s.1 It was an acoustic simulation that used a water tank and piezoelectric
transducers to replicate the transmission, delay, reflection, and reception of the radar
signal. A representative radar range trace could be achieved. This device was useful for
engineering research and as an aid in radar systems design.
By the mid-1960’s, glass plates with flying spot scanners were routinely used to
generate reasonably good radar displays. The glass plate was a photographic positive
of the desired radar image of the gaming area. It was backlit with a uniform light source
while a photodiode was rapidly scanned on the front side, producing a voltage
proportional to the radar return. Unfortunately, this system had numerous shortcomings,
including lack of compensation for occulting with changes in aircraft altitude.
Terrain boards were used in the 1970’s.A vertical video probe was placed at the
appropriate location and a directional light source was positioned according to the
encounter geometry. The light source illuminated the monochrome-colored terrain in a
manner that created pseudo-radar imagery with fairly accurate radar shadows. The
system worked, but was somewhat cumbersome and inflexible.
Neither the flying spot scanner nor the terrain board produced an accurate depiction of
angle resolution effects.
Another difficulty with early DRLMS was the lack of adequate source data from which to
build the landmass database. This problem was abated in the early 1970’s with the
introduction of the Defence Mapping Agency (DMA) Digital Landmass System (DLMS),
the forerunner of the modern Digital Terrain Elevation Data (DTED) and Digital Feature
Analysis Data (DFAD). Soon this database was finding application. 2 Several DRLMS
were built, and by 1980 their construction was routine. These were hardware-intensive
implementations of the radar models comprised of 800–900 printed circuits.
By the mid-1980’s, several DRLMS were built with minicomputer front-ends and array
processor back-ends. Examples include the B-1 DRLMS built by Cubic and Boeing.
These systems shared many of the undesirable characteristics of the predecessors.
In 1985, we developed a new approach to building DRLMS: a software-only solution
using a high-speed, general-purpose computer constructed of monoboard computers on
a VME bus. This proved to be successful, and as computers, memory, disk drives, etc.
have improved, so has this approach to building the DRLMS. Thus Camber has become
the leading DRLMS manufacturer.
Boards stuffed to the gills with small-scale integrated circuits (e.g., the GE F-16
DRLMS). Suffice it to say, they were big, expensive, and nontrivial to maintain.
By 1990, general-purpose workstations became a powerful enough computing resource
to replace the VME system, and UNIX SVR4 arrived, providing a means for running
real-time UNIX. So we migrated to Silicon Graphics, Incorporated (SGI) computers and
are building DRLMS with them today. The Indian Navy Sea Harrier flight simulator is
currently being upgraded and will soon be fitted with an SGI-based DRLMS for the Blue
Fox Radar.
A Typical Radar
Figure 1 provides a block diagram of an A/G radar. Table 1 lists the basic radar
characteristics. Note that this information is generic and not specific to any particular
radar, but is typical of a modern A/G radar.
COCKPIT
AVIONICS CONTROLS
COMPUTER
Table 2 lists the three primary radar modes. The RBGM mode is a conventional radar
mode. The only distinction is that with modern technology it is possible to match the
radar resolution to the display resolution by variable pulse compression, and thus
eliminate the collapsing losses present in earlier radars. The DBS mode is a scanning
mode, providing constant azimuth resolution throughout the field of regard. It is
generated by sequential batch processing of short, fixed-length FFTs per-formed at a
variable PRF and combined (as adjacent segments) to give the continuous scan
display. The SAR mode is a spotlight mode, providing constant cross-range resolution
at any designated range/azimuth location. It is generated by a single, long FFT that is
performed with motion compensation at a constant PRF. (In reality, several FFTs are
used to provide adequate azimuth coverage and several looks, performed at different
RF frequencies, are noncoherently combined to improve image quality.)
Both DBS and SAR modes require motion compensation. The aircraft motion is
acquired from the inertial navigation system. Then the receiver Local Oscillator (LO) is
offset by appropriate frequency to remove the instantaneous Line-Of-Sight doppler from
the radar signal that is due to the aircraft motion.
] Table 2. Typical A/G Radar Mode Parameters.
Azimuth
Range
Miscellaneous
Display Format
The radar described above possesses typical radar parameters and processing and
provides a basis for discussion. However, an actual production radar will have dozens of
modes, submodes, and functions. It is not unusual to have ten or more display formats
with significant radar-generated symbology and overlays for each display. It is not
unusual to have 100 or more user-selectable radar parameter values.
Most of this complexity is due to the great amount of design flexibility afforded by the introduction of
the RDP and PSP into the radar about 20 years ago. All of the above represents added complexity
to the DRLMS, but is not truly a technical challenge. The major technical challenges are:
Figure 2 provides a functional block diagram of the radar model employed by the
DRLMS. The modelling technique employed by a radar simulator is an important
component in producing a realis-tic radar display. Only through a comprehensive
understanding of the underlying principles and theories of radar systems engineering is
it possible to provide realistic simulation suitable for train-ing. The modeling approach
incorporates a detailed database and a realistic treatment of the interaction of the
simulated radar model with the terrain, water, weather, and targets. All appropriate radar
phenomena and effects are accounted for. Figure 2 indicates the primary functional
elements of the simulated radar and environment. The environment model provides the
gaming area database to which the radar model is applied. The rest of the radar model
can be partitioned in much the same way as the actual radar. It consists of an antenna,
transmitter, receiver, a signal processor, and inter-faces to other systems such as the
display system.
Environment
Terrain and weather are represented by three-dimensional and four-dimensional
databases. The ter-rain is described by topography, radar reflectivity, and surface
attributes. The topography consists of a 2-D grid of elevations or a list of polygon
vertices; the grid is best for applying the radar model, the polygons are best for storage
and for representing varying scene densities. The radar reflectivity is provided for each
element in the topography and is the radar backscatter coefficient (s 0) for homog-enous
surfaces and Radar Cross Section (RCS) for specular returns, such as man-made
objects. Surface attributes are provided for each element in the topography and include
indications of trans-
Environment Transmitter Antenna Receiver
Landmass
Generator Current
Range
Trace
Jammer
Generator
Spatial Processing
The first step in modeling any of the A/G radar modes is to gather the terrain data that is
within the radar antenna azimuth beamwidth. This is accomplished by periodically
forming range traces at closely spaced azimuth angles.
The range trace azimuth spacing is based on the actual radar antenna beamwidth, scan
rate, and pulse repetition frequency.
The range samples are spaced according to the radar range resolution or the radar
display resolution. Note that the formation of range traces is area sampling: all data
points between range traces are processed to form the range samples on a given range
trace Once range trace is computed, the occulting process is applied to the trace to
determine shadowing and the grazing angle; Earth’s curvature effects are included. The
occulting process for targets is essentially the same except the target altitude is
introduced. This calculation is performed sequen-tially, beginning with the near range
and proceeding to far range for efficiency and to reduce redundant calculations.
For each sample, the depression angle to the terrain is computed. The grazing angle
and reflectivity computed and are used to determine the backscatter coefficient.
to prevent blinking and other simulator anomalies.
Signal Strength
The total power for each radar range/azimuth bin is determined as the simulated
antenna sweeps through the scan volume. Signal power due to terrain, water, weather,
jammers, and targets are calculated separately using the radar equation and then.
combined to derive the received signal strength. These equations determine the radar
signal attenuation due to both free-space and atmospheric losses
The received power due to the terrain is calculated using the radar equation. It is based
upon the combination of the backscatter coefficient for diffuse clutter and RCS for
discretes. Each terrain sample is computed as a random variable with appropriate
distribution (e.g., Rician) and with the computed value of average power.
For each range bin that encounters a non-zero rainfall value along the line-of-sight, both
rainfall attenuation and backscatter values are determined by accessing a look-up table
indexed on rainfall rate. The volumetric backscatter coefficient is used to calculate the
magnitude of the resulting radar return using the appropriate radar equation. The
attenuation is applied to the current range bin and cumulatively to following range bins.
The received power due to target returns is calculated using a radar equation. The
target’s RCS is treated as a discrete and is independent of range.
RCS tables are maintained in the DRLMS for each target type. These tables contain
RCS values for each target as a function of aspect angle. In the case of distributed
targets, the contributions of all scatterers are summed to form a composite return.
Effects of target scintillation are included in the calculations.
The received power due to water is modeled in a manner similar to landmass returns.
Water tends to be homogenous and is modeled as a random process. The backscatter
coefficient is modeled as a log-normally distributed random variable. The mean
backscatter coefficient is a function of the sea state.
Antenna
The antenna model consists of two main components: an antenna servomechanism
model and a model of the antenna beam pattern. The servomechanism model
simulates the positioning of the antenna in azimuth and elevation. It accounts for the
scan mode, antenna dynamics including the interactions of the drive motors, gimbals,
resolvers, and the required coordinate transformations. For each radar mode there is at
least one scan pattern. This pattern is specified by the scan generator, which uses an
internal real-time clock to periodically determine the commanded antenna position. For
instance, the RBGM mode is an azimuth scan. This position is converted to the proper
coordinate system that is applied to a control system model for the antenna. The control
system model accounts for the antenna drives and inertia. A gimbal model is used to
determine the antenna direction of motion and the gimbal limits. A resolver model
records the antenna position which forms the feedback loop for the control system. Of
course, in some modes the antenna position is commanded by the host.
The antenna beam pattern is modeled in both elevation and azimuth. After the
individual returns generated by the environment model and the transmitter model are
complete, they are modified by computing the attenuation relative to the antenna peak
gain at boresight as a function of elevation angle off boresight.
The antenna azimuth beamwidth is accounted for by convolution of the azimuth antenna
pattern with the adjacent range traces. This provides the reduction in cross-range
resolution with range for RBGM and DBS modes. The azimuth spacing is selected to
provide adequate fidelity and can be as small as the product of the radar Pulse
Repetition Interval (PRI) and scan rate and as coarse as the radar dwell (approximately
the antenna beamwidth). The azimuth antenna pattern can be adequately modeled with
5 to 7 samples.
Receiver Model
The antenna model provides an output array of signal powers indexed by range at the
dwell rate. In the receiver model, the transmit power, range bin integration, and receiver
characteristics are modeled. The range bins are formed by convolving the range
ambiguity function with the range samples of each range array. The range ambiguity
function is derived from the receiver characteristics and can be adequately modeled with
5 to 7 samples.
Receiver functions such as Sensitivity Time Constant (STC), the Automatic Gain Control
(AGC) are also modeled. The STC waveform is modeled as a polynomial in range.The
AGC is modeled with a classic AGC control loop carefully adjusted to match the
characteristics of the actual radar. In addition, any Constant False Alarm Rate (CFAR)
techniques such as Fast Time Constant (FTC) are accounted for. A noise signal is
summed with the radar returns to model the background interference in the receiver
frontend. The strength of the noise signal is dependent on the receiver bandwidth, noise
figure, etc. Note that the receiver gain affects the signal and may affect the noise
depending on where the gain is applied. Random noise signals at the proper strength
are computed for each range sample and added to the composite radar signal.
Scan Conversion
If the DRLMS provides a raster scan output, then scan conversion is performed to
convert the polar format range/azimuth signal into a Cartesian format x,y signal. (Scan
conversion is essentially an R/ q-to-XY conversion.) Significant data
collapsing/expanding occurs at short/long ranges.
Hardware Architecture
Figure 3 describes a flight simulator and indicates the three interfaces that are important
to the DRLMS:
• Cockpit/host interface
• Host/DRLMS interface
• DRLMS/cockpit interface
Normally, all radar controls and indicators are interfaced through some sort of linkage to
the host computer and the host computer relays this information to the DRLMS on the
host/DRLMS inter-face. The host computer provides, via host/DRLMS interface,
simulator data (e.g., initialize, run, freeze), ownship and target position and attitude, and
environmental data. The DRLMS generates the radar video and transmits this to the
cockpit via the radar/cockpit interface.
Figure 4 is a typical DRLMS hardware block diagram. We routinely build DRLMS using
an SGI Challenge workstation configured with four 250MHz R4400 processors and
128Mbytes of RAM. The VMEbus provides the channel for the host interface (shown
here as an HSD interface) and the cockpit interface (shown here as a Synergy frame
buffer.) This system is usually delivered as a
rackmounted version and requires no special power or cooling, and has a footprint of
less than 4 sq ft. Contrast this with the previous generation of hardware-intensive
DRLMS that occupied 8 racks and several hundred square feet, required forced-air
cooling, and required special power of 300-500 amps at 5 volts.
WIDE ANGLE
COLLIMATED
DISPLAY OR
PROJECTION
RADAR SYSTEM.
MODEL FOCUSED AT
INFINITY.
FLIR
MODEL
CHANNEL 1 2 3
COCKPIT
DATA VISUAL
BASE MODEL
VISUAL VIDEO
HOST HOST/WS/I/F GRAPHICS RADAR VIDEO
COMPUTER WORK-
STATION FLIR VIDEO
TRAINER
INTERFACE
LINKAGE
COCKPIT CONTROLS/INDICATORS
The high-level software data flow diagram shown in Figure 5 illustrates the information
flow between major software processes. The system runs in realtime and is usually
driven by 30Hz interrupts from the host computer.
The system runs the IRIX 5.3 (32-bit) or IRIX 6.2 (64-bit) operating system, both of
which are Unix SVR4. We maintain POSIX.4 compliance. All software is written in C++
using Object Oriented Design (OOD) methodologies.
DRLMS
SHARED
MEMORY
SHARED MSG Q
MEMORY
SHARED
RADAR MEMORY
(1)
MODEL HOST
(1) HOST INPUTS
INTERFACE
WEATHER
(1) MODELER TO DRLMS. DRLMS
SHARED OUTPUTS TO HOST
MEMORY
SHARED
SHARED MSG Q MEMORY
EXPAND MEMORY
PAGER
MSG Q HOST
INPUTS
HOST
INPUTS
PACKED
INTO A
MSG
(1
) DISPLAY
MSG Q
HOST
INPUTS
DISPLAY HOST
COMPUTER
)
) S
S
IT CPU IT
B BOARD B
(4 PROCESSOR) (40
(256 S
S U
U B
B S
A S
T E
A MEMORY R
D
D BOARD D
A
I/O BOARD
INTERFACE
MODULES
DISK
TERMINAL
DRIVE
SYNERGY
HSD FRAME BUFFER
INTERFACE
TO HOST TO COCKPIT
DISPLAYS
Today, the use of commercial workstations and modular software allows for the creation
of high-fidelity, realistic Airborne Radar Simulators. These systems are less expensive,
easier to modify and more maintainable and reliable than their predecess
REFRENCES
1. Moore, R. K.: “Acoustic Simulation Of Radar Returns,” Microwaves, vol.1, no. 7, pp.
20-35, December, 1962.
3. Bair, G. L. and Smith, J.: “The AN/APS-137(V) ISAR Derivative System for the P-
3Aircraft,” Proceedings Of The 31st Tri-Service Radar Symposium, 1985.