Professional Documents
Culture Documents
Gauthier Sylvain
Surface Radar Section
Riseborough, Edwin
Surface Radar Section
Tim J. Nohara
Sicom Systems Ltd
Graeme Jones
Sicom Systems Ltd
This report describes the major capabilities, features, models and applications of MFRSIM. The first
section provides background information on the development of the DRDC Ottawa simulator over the
years. The second section provides an overview of MFRSIM, including its modules and capabilities.
The radar models of MFRSIM are examined in detail in the third section. This will provide a clear
understanding of the capabilities and limitations of this simulator. The fourth section explains how to
use MFRSIM and shows some examples using it. The last section discusses the future R&D plans for
development of MFRSIM and its main applications.
Résumé
RDDC Ottawa a développé un simulateur de radar multifonction a travers plusieurs contrats avec
Atlantis Scientific Inc et Sicom System Ltd. La dernière version s’appelle MFRSIM et est codée en
MATLAB version 6.1 (MathWorks Inc). MFRSIM fonctionne sur des ordinateurs personnels
compatibles avec IBM. MFRSIM a été développé spécifiquement pour évaluer la capacité de détection
des radars multifonctions contre les missiles anti-navires opérant en région côtière. Des radars
multifonctions avec des antennes rotatives où fixes, à réseau de phase, peuvent être simuler par
MFRSIM. Ce simulateur peut aussi simuler des antennes rotatives conventionnelles comme celle
utiliser dans les radars de surveillance de volume. La simulation peut inclure du fouillis de sol, de
précipitations, d’insectes, de plaquettes de brouillage, et de brouillage. MFRSIM produit des sorties de
détection radar d’une manière causale de faisceau en faisceau Les sorties du détecteur sont envoyées
au contrôleur de faisceau et au système de poursuite. L’algorithme de poursuite initie automatiquement
les nouvelles traces de poursuite et communique avec le contrôleur de faisceaux. Les données
produites par le contrôleur de faisceaux et du système de poursuite sont utilisées par le contrôleur de
faisceaux pour planifiée la transmission des faisceaux de surveillance, de confirmation, et de poursuite.
Les paramètres de la forme d’onde transmise peuvent être changer de faisceau en faisceau. Une
procédure e été développé qui permet au contrôleur de faisceaux de choisir automatiquement les
paramètres de la forme d’onde parmi un ensemble de forme d’onde prédéterminé. Cette capacité est
encore rudimentaire. Cependant l’intention est éventuellement de choisir une forme d’onde en fonction
de l’écho radar et des données d’environnement.
In the late 1990’s, it became apparent to DRDC Ottawa scientists that APAR couldn’t be fitted on the
Halifax class frigates as originally intended. In fact, APAR is much too heavy to be installed on the
Canadian patrol frigate (CPF) without major modifications to the ship’s structure. Hence, a new
version of the MFR simulator was developed to investigate economic alternatives for the mid-life
upgrade to the CPF radars. The simulation capabilities were significantly enhanced over the original
version. For example, various configuration of MFR with multiple antenna faces, that can be rotating
or fixed, can be simulated. Rotating MFRs such as ARABEL and MESAR with single or back-to-back
faces, similar to APAR, have been studied in detail using this simulator. The simulator has also been
used to evaluate the detection capability of the CPF radars against the most likely ASM missile to be
encountered by the Canadian Navy.
In 2000, MFR R&D requirements shifted toward developing a simulator that can be used to study
adaptive techniques, including sensor integration, to optimize the MFR performance. MFR’s have a
high level of agility in the transmitted waveform and antenna beam steering. This high beam flexibility
makes it possible to change scan strategies or scheduling on the fly. Scheduling can be optimized as a
function of the mission, detected threat and environment. MFRs can modify the transmitted waveform
from pulse to pulse or burst to burst over a wide range of parameters. Depending on the environment
and the threat, certain waveforms will provide a better performance than others. What is the best
waveform or beam scheduling for a given scenario or environment? Most of the time, a radar operator
would not know how to adequately exploit the versatility of the MFR. Current emphasis on studying
littoral regions reduces the radar reaction time, making adaptive techniques more important. Sensor
integration as suggested by the NAAWS study is also becoming much more important to win back
reaction time [1]. The newest version of the MFR simulator, called MFRSIM, has been developed to
study adaptive control of MFR. The simulated MFR can change the parameters of the transmitted
waveforms from dwell to dwell, conduct random beam scan, and receive cues from separate radars. It
is planned that other sensors will be added to cue the MFR.
The present simulator, MFRSIM is a very capable MFR software simulator. The major capabilities,
features, models and applications of MFRSIM are described in this report. The first section provides
background information on the development of the DRDC Ottawa simulator over the years. The second
section provides an overview of MFRSIM, including its modules and capabilities. The radar models of
MFRSIM are examined in detail in the third section. This will provide a clear understanding of the
capabilities and limitations of this simulator. The fourth section explains how to use MFRSIM and
shows some examples using it. The last section discusses the future R&D plans for development of
MFRSIM and its main applications.
Gauthier, S., Riseborough, E., Nohara T.J., Jones. G. 2002, Multifunction Radar
Simulator (MFRSIM). DRDC Ottawa TM 2002-165. Defence R&D Canada - Ottawa
A la fin des années 1990, il est devenu évident aux scientifiques du RDDC Ottawa que APAR ne
pourrait pas être installée sur les frégates canadiennes comme originalement voulu. En faits APAR est
beaucoup trop lourd pour être installée sur les frégates canadiennes. Cela demanderait des
modifications considérables à la structure du navire. On a donc développé une nouvelle version de
simulateur de RMF pour étudier des alternatives économiques au sujet de la mise à jour des
composants radars des frégates. Les capacités de simulation ont été améliorées considérablement
comparée à la version précédente. Par exemple, différentes configurations de RMF ayant plusieurs
antennes, rotatives ou fixes, peuvent être simulées. Des RMFs avec une où deux antennes rotatives ont
été étudiées en détails à l’aide de ce simulateur. ARABEL et MESAR sont des exemples de RMFs
utilisant des antennes rotatives. Ces RMFs, cependant, requièrent l’utilisation de missiles actifs plutôt
que semi-actifs. Le simulateur a aussi été utilisé pour évaluer les capacités de détection des radars
installée sur les frégates canadiennes contre des missiles anti-navires que pourrait affronter la marine
canadienne.
En 2000, les besoins de R & D sur les RMFs ont nécessité un simulateur qui permettrait d’étudier des
techniques de contrôle adaptatives incluant l’intégration de récepteurs pour optimiser les performances
radars. Les RMFs ont un grand degré de flexibilité dans la forme d’onde qui est transmise et le
balayage du faisceau électronique. La grande flexibilité du faisceau rend possible un changement de
stratégie de balayage électronique en temps réel. Le contrôle des faisceaux peut être optimiser en
fonction de la mission, de la menace qui est détectée et de l’environnement. Les RMFs peuvent
changer la forme d’onde transmise d’impulsion en impulsion. Dépendant de l’environnement et de la
menace, certaines formes d’ondes fourniront une meilleure performance que d’autres. La présente
emphase sur les régions côtières réduit considérablement le temps de réaction rendant les techniques
adaptatives plus importantes. L’intégration de récepteurs comme suggérer par une étude de l’OTAN
(NAAWS study) est aussi en train de devenir plus important pour augmenter le temps de réaction. La
dernière version du simulateur s’appelle MFRSIM et a été développée pour étudier des techniques de
contrôle adaptatif de RMFs.
Aujourd’hui, MFRSIM est un logiciel de simulation qui est très capable. Ce rapport décrit les capacités
principales, les caractéristiques, les modèles et les applications de MFRSIM. La première section
fournit des informations sur le développement de MFRSIM au cours des années. La deuxième section
fournit une vue d’ensemble sur MFRSIM incluant ses modules et capacités. Les modèles radars de
MFRSIM sont examinés en détails dans la troisième section ce qui devrait donner une bonne
compréhension des capacités et limitations de ce simulateur. La quatrième section explique comment
utiliser MFRSIM avec quelques exemples. La dernière section décrit les futures directions de R&D
pour MFRSIM ainsi que ses applications.
Gauthier, S., Riseborough, E., Nohara T.J., Jones, G. 2002. Multifunction Radar Simulator
(MFRSIM). DRDC Ottawa TM 2002-165. R & D pour la défense Canada - Ottawa
Abstract........................................................................................................................................ i
Sommaire................................................................................................................................... iv
1. INTRODUCTION......................................................................................................... 1
2. BACKGROUND........................................................................................................... 2
2.1 Multifunction Array Radar Simulator .............................................................. 2
2.1.1 APAR a Naval MFR............................................................................ 2
2.1.2 MFARSIM........................................................................................... 3
2.2 Rotating Multifunction Array Radar Simulator................................................ 4
2.2.1 Economical alternatives to APAR....................................................... 4
2.2.2 RMFARSIM........................................................................................ 5
4. RADAR MODELS...................................................................................................... 14
4.1 Radar Range Equation.................................................................................... 14
4.1.1 Missile Models .................................................................................. 14
4.1.2 Antenna Models ................................................................................ 14
4.1.3 Propagation Models........................................................................... 15
4.1.4 Clutter Models................................................................................... 16
4.1.4.1 Sea Clutter.......................................................................... 16
4.1.4.2 Land Clutter ....................................................................... 18
4.1.4.3 Rain and Chaff Clutter ....................................................... 18
7. References ................................................................................................................... 37
Figure 15. The starting window with the File menu highlighted ........................................................ 28
Figure 16. SIR plot from calculator mode (default parameters + Doppler processing) ........................ 28
Figure 20. PPI views of missile, Sea Giraffe, APAR and SPS-49 detections (clockwise from top left) . 33
Figure 21. Adjusting the Sea Giraffe's search parameters for tracker interfacing............................... 34
Figure 23. PPI display of the APAR face's tracking beams ............................................................. 35
DRDC Ottawa has developed a MFR software simulator through several contracts with Atlantis
Scientific Inc and Sicom System Ltd. The latest version, called MFRSIM, is coded in MATLAB
version 6.1 (by MathWorks Inc), running on an IBM compatible PC. MFRSIM has been developed
specifically to evaluate the detection capability of an MFR against anti-ship missiles (ASM) operating
in littoral environment. Both rotating and non-rotating phased array MFR can be simulated, as well as,
conventional rotating antennas such as volume search radars. The simulation can include land clutter,
sea clutter, chaff clutter, rain clutter and angel clutter with jamming. MFRSIM produces detection
outputs in a causal manner for one dwell at a time. The detector outputs are sent to the beam scheduler
and the associated tracker. The tracker automatically initiates new tracks and interfaces with the beam
scheduler to request tracking dwells. The beam scheduler controls the scheduling of surveillance;
confirmation, cued search and tracking beams based on the tracker and detector outputs. The
parameters of the transmitted dwell can be changed from dwell to dwell. A procedure has been
developed to automatically select the waveform parameters for the scheduled dwell from a library of
predefined waveforms. Although this capability is still rudimentary, the intention is eventually to select
a given waveform as a function of the radar returns and environmental data.
This report describes the major capabilities, features, models and applications of MFRSIM. The first
section provides background information on the development of the DRDC Ottawa simulator over the
years. The second section provides an overview of MFRSIM, including its modules and capabilities.
The radar models of MFRSIM are examined in detail in the third section. This will provide a clear
understanding of the capabilities and limitations of this simulator. The fourth section explains how to
use MFRSIM and shows some examples using it. The last section discusses the future R&D plans for
development of MFRSIM and its main applications.
MFRs have a high level of agility in the transmitted waveform and antenna beam steering. MFRs are
usually phased array radars that can be electronically scanned in both azimuth and elevation. Hence,
they can look in a specific direction almost instantaneously. There is no need to wait for the next
revisit, as in conventional rotating radar. The high beam flexibility makes it possible to perform several
functions in a single system. Typically, the duties of a naval MFR include, horizon search and track,
surface search and track, own missile tracking with uplink data transmission and target illumination,
high diver tracking, cued acquisition as from adjunct sensors, burnthrough in self-screening jamming,
track-on-jammer, non-cooperative target recognition, and possibly kill assessments [1]. The various
duties of MFR are shown diagrammatically in Figure 2. MFR perform all these functions
sequentially. Hence, time to be spent on each function has to be optimized.
APAR is not intended to be the only radar onboard, but would be part of a suite of sensors as
recommended by a NATO Anti-Air Warfare System (NAAWS) study conducted in early 1990’s. The
NAAWS study recommended suite of sensors that would be able to defeat the emerging threats of new
ASMs [1]. The preferred configuration consisted of a suite of four sensors including a precision
electronic support measures sensor, an infrared search and track sensor dedicated to horizon
surveillance, a volume surveillance radar, and a MFR [1]. The keys functions of the MFR under the
NAAWS configuration were horizon search and support to target engagement. Horizon search is
2.1.2 MFARSIM
In early 1990’s, DRDC Ottawa developed a software program called MFARSIM, which stands for
multifunction array radar simulator [1]. MFARSIM has been developed specifically to evaluate the
detection capability of APAR-like radars against ASMs. This simulator can only simulate a fixed
antenna array face scanning electronically the horizon against one ASM that flies in a straight line and
at constant velocity (Figure 3). MFARSIM has three major modules, which are the scenario
generation, the radar range equation and the post processing modules. The scenario generation module
generates the missile trajectories and all antenna beam positions in one pass and determines when they
intersect. Only these specific dwells are passed to the next module of MFARSIM, the radar range
equation module. This approach considerably reduces the computer processing requirements, since the
full range equation is run for much fewer dwells positions. The radar range equation module calculates
the signal-to-noise ratio (SNR) that will be produced at the output of the radar receiver. This module
uses the usual form of the radar range equation for this calculation. The effects of anomalous
propagation are included in MFARSIM. Data files of propagation loss factors are calculated using the
commercial software called PCPEM [15] and read into MFARSIM. The post-processing module
calculates the probability of detecting the missile using the resulting SNR. The first version of
MFARSIM had no models of clutter. It assumed that clutter has been suppressed successfully. Curves
of signal-to-noise ratio and the probability of detection are the main outputs of this simulator.
MFARSIM was used to investigate the impact of beam spacing and anomalous propagation on the
performance of MFR. Beam spacing has a significant impact on the time spent on horizon search and
T a rg e te d s h ip
dx
M F R rad ar
dy
S E A S K IM M E R
In late 1990’s, the Surface Radar Section started studying economic MFR alternatives to APAR for the
mid life CPF upgrade. At that time, it became apparent to DRDC Ottawa scientists that APAR couldn’t
be fitted on the Halifax class frigates as originally intended. APAR is much too heavy to be installed
on the CPF without major modifications of the ship’s structure [16]. Extensive studies have
investigated potential solutions to fitting APAR on the to Halifax class frigate, but none were approved.
One of the options considered, cutting the ship in two sections and adding a ten-meter plug to
accommodate the new systems. The idea was abandoned due to its prohibitive cost. All other options
have also been discarded. Alternative MFRs of interest included rotating MFR with one face or two
back-to-back faces. ARABEL and SAMPSON are good examples of naval rotating MFRs. ARABEL
is a passive phased array rotating MFR operating at I/J band. SAMPSON is a dual rotating active
phased array MFR operating at E/F band. These MFRs, however, requires the use of active missiles,
rather than semi-active, or separate illuminators. Semi-active missiles, such as the Sea Sparrow and
Evolved Sea Sparrow will be the missiles used in the CPF. The plan would be to keep the STIRs
already on the CPF to perform the target illumination for the semi-active missiles.
The simulation capabilities have been significantly enhanced over the older version. RMFARSIM can
simulate scenarios with up to five ASM (sea skimmers or high divers) operating in littoral regions
(Figure 4). Each ASM can manoeuvre in both azimuth and elevation. Littoral land clutter models and
jamming were added. The simulated MFRs can have multiple antenna faces that can be rotating or
fixed. RMFARSIM can also simulate conventional rotating search radars. Modeled MFR functions
include horizon search (or volume search), tracking, cueing, and confirmation. For the rotating MFR,
the missiles are only visible for a fraction of the time when the antenna faces are oriented in their
direction. Therefore, confirmation, tracking, and cueing must be scheduled when the missiles are
visible. This is one of the main differences between fixed and rotating antenna faces. For the non-
rotating case, after MFR functions, the antenna beam returns to its prior azimuth position relative to the
boresight. The propagation-modeling software PCPEM used to calculate the propagation loss path was
replaced by TERPEM and integrated directly into RMFARSIM [17].
Monte Carlo simulation capabilities have also been added to RMFARSIM. In fluctuating systems, a
Monte Carlo algorithm generates fluctuating quantities based on their statistics. In RMFARSIM,
Monte Carlo simulation is used to produce radar-like detection events. Full Monte Carlo simulation
runs, to ascertain system performance, would be extremely time consuming, as the complete simulator
would have to be run many times. RMFARSIM circumvents this to generate feasible results by taking a
post-detection Bayesian approach. We generate a uniform random number between 0 and 1 for each
range bin where the Pd has been calculated. This is then repeated for 1000 equivalent Monte Carlo
runs. Based on the random number generated for each range bin, detection is declared if this number is
lower than the probability of detection for that range. Otherwise there is no detection. We define the
earliest detection range (or tracking range) as the first range bin (in descending order) that has a
successful detection. This procedure is repeated for all 1000 Monte Carlo events. These results are then
“histogrammed” to show the detection or tracking range distribution. An example of a Monte-Carlo
histogram generated by RMFARSIM is shown in Figure 5. There are only a few detections that occur
around 15 km where the sea skimmers crossed the horizon. Later, hundreds of detections occur around
13.7 km, and so on. The low spot around 13.5 km is due to blind zones caused by eclipsing loss. The
Monte Carlo histograms may prove very useful to show the best case and worst-case scenarios. For
example, for the worst case around 12.4 km, there might be not enough time to intercept the missile.
As a result, the user would experiment with the radar sensitivity to move the worst-case result to a
desired range. PPI displays showing where detections occur have also been added to RMFARSIM.
Figure 6 shows example of RMFARSIM displays for different scenarios.
JAMMER
- MULTIFUNCTION RADAR
- VOLUME SEARCH RADAR
RMFRSIM has been used to study the optimization of various configurations of an MFR against the
most likely ASM scenarios to be encountered by the Canadian Navy [6, 7]. During this study, the
detection capabilities of existing radar components installed on the CPF were also evaluated against
these ASM scenarios. The study includes the Sea Giraffe and the SPS-49 as well as one and two
rotating MFR faces each similar to an APAR face. A special emphasis of the study was put on
detection of sea skimmers in littoral waters.
Rotating MFRs with single or back-to-back faces, similar to those of APAR, have been studied in detail
using the simulator. The simulator has also been used to evaluate the detection capability of the CPF
radars against the most likely ASM missile threat to be encountered by the Canadian Navy.
As expected, the Canadian Navy is now looking to alternative radar solutions for the CPF mid-life
upgrade (FELEX program) and our DRDC Simulator is being used to aid in this assessment.
In order to study the adaptive control of an MFR, the Surface Radar Section has developed a new
version of MFR simulator called MFRSIM [8, 9]. MFRSIM has first been made completely causal or
event driven, as required for the studying. In contrast with RMFARSIM, which generates a pre-
defined matrix in batch, Monte Carlo capabilities have not be implemented with MFRSIM. MFRSIM
was further modified to enable the transmission of dynamic waveforms, random beam scans, and
cueing from separate radars.
MFRSIM produces detection outputs in a causal manner for one beam (position) at a time. The
detector outputs are sent to the beam scheduler function and the associated tracker. The tracker
automatically initiates new tracks and interfaces with the beam scheduler to request tracking dwells.
The beam scheduler controls the scheduling of surveillance, confirmation, cued search and tracking
beams based on the tracker and detector outputs. The parameters of the transmitted dwell can be
changed from dwell to dwell. A procedure has been developed to automatically select the waveform
parameters for the scheduled dwell from a library of predefined waveforms. Although this capability is
still rudimentary, the intention is to eventually select a given waveform as a function of the radar
returns and environmental data. The associated tracker and PPI display can receive and display
detection outputs time-sequentially (Figure 9).
BEAM GENERATION
DETECTION MODULE
MHMT
- DETECTION EVENT?
- RANGE, ANGLE, VELOCITY TRACKER
R, AZ, EL, V
BEAM SCHEDULER
- WHAT, WHERE, WHEN
PREDICTION
The beam scheduler generates the position of each antenna beam in horizon search mode until a
detection occurs. The beam position for the horizon search can be generated sequentially, i.e., from a
minimum to a maximum value. The search scan pattern can also be generated randomly as is done by
the MESAR and SPY-1 radars. MFRSIM can be set to conduct volume search rather than horizon
search. The volume search can also be done sequentially according to a given pattern (row by row) or
randomly. MESAR not only conducts random volume scan but also changes the parameters of the
waveforms as a function of desired range. MFRSIM generates the scenario one-time increment at a
time.
A pre-detection check is done for each antenna position to determine if there is a possibility of a
detection and if the full radar range equation calculation has to be performed. For this purpose, two
thresholds have been set; the antenna gain in the target direction and a simple form of SNR. If the
antenna gain in the direction of the illuminated missile is superior to an input threshold, then there is a
possibility of having detection. This triggers looking at the next threshold. The second threshold is a
simple function of signal to noise ratio, which does not include the propagation factor and clutter level.
If this SNR is above a given threshold detection is possible and then the whole detection algorithm is
run.
The detection process includes the calculation of the mean power using the standard form of the radar
range equation and the associated spectrum that will be produced at the receiver output. In effect,
MFRSIM does not generate a signal that propagates to the target and returns. The propagation factor
includes the propagation effects from multipath and atmospheric conditions. The propagation factor F
is calculated by the commercial program TERPEM developed by Signal Sciences Limited in the United
Kingdom [17]. For all scenarios, including littoral regions, MFRSIM can read DTED files to
automatically extract the land profile to be used in TERPEM to determine the propagation factor.
The detection process includes the calculation of the mean power using the usual form of radar range
equation and associated spectrum that will be produced at the receiver output. In effect, MFRSIM does
not generate a signal that propagates to the targets and comes back. It simply calculates a mean power
using the usual form of the radar range equation. The propagation factor has to be calculated to include
the propagation effects from multipath and atmospheric conditions, which are very important for low-
level propagation. The commercial program TERPEM calculates the propagation factor F. For
scenarios, including littoral regions, MFARSIM can read DTED files and automatically extract the land
profile to be used in TERPEM for the propagation factor calculation. TERPEM is directly integrated
into MFRSIM and is completely transparent to the user.
The simulation program considers four types of processing; no processing, n-pulse canceller, DFB, and
an n-pulse canceller with a DFB. In order to determine the output of the filter, one needs to know their
power spectral responses, which are given in several radar books [4]. The resulting signal to residual
clutter ratio is then used to calculate the probability of detection Pd . The mathematical engine used to
calculate the probability of detection is based on the Neuvy equation [1, 23]. These probabilities of
detections are then used to produce radar-like detection statistics, i.e., to show where detection occurs.
A uniform random number between 0 and 1 is generated, and a detection is declared if this number is
inferior to the probability of detection. Otherwise there is no detection. When detection is declared,
radar measurements similar to those produce by real radar will be generated and passed to the beam
scheduler. Standard accuracy formulas given by reference [8] are used to generate the radar
measurements.
The beam scheduler will update its detection list and communicate the radar measurements to an
external tracker, proprietary to Sicom System ltd [8]. The tracker will initiate tracking and return the
predicted location of the targets. The associated tracker and PPI display can time-sequentially receive
and display detection outputs (Figure 10). The beam scheduler controls the scheduling of surveillance,
confirmation, cued search and tracking beams based on the tracker and detector outputs. The
scheduler includes advanced logic for rotating MFRs to ensure that the target can be illuminated at the
desired time. The parameters of the transmitted signal can be changed from dwell to dwell.
In the search mode, the waveform can be changed sequentially or randomly from dwell to dwell or scan
to scan. One hopes that by changing the waveform from dwell to dwell, one would be able to detect the
target faster. For the confirmation dwell, the user has the option of using the same waveform as in the
detection process. For the tracking dwell, the waveform is adjusted automatically as a function of the
target velocity and range. The pulse repetition frequency (PRF) is first adjusted to minimize eclipsed
ranges. Then if necessary the target is moved out of the blind zone by moving the radar frequency
upward by 500MHz.
MFRSIM takes into account false alarms and generates false radar measurements when they occur.
Using the probability of false alarms per burst input by the user, the probability of false alarm per dwell
can be calculated. Using a process similar to the one used in generating the detection event, MFRSIM
generates a random number and if this number is inferior to the probability of false alarm then a false
alarm is declared. Once false alarms are declared, MFRSIM will generate random radar measurements
over the whole range of possible parameters. For example, the value of false measured range is
selected randomly between the minimum and maximum possible range. A similar process is done for
the generation of speed and direction measurements. The false measurements are sent to the scheduler,
which acts upon them as real detections, i.e., they are send to the tracker for processing.
MFRSIM gives the option of using non-adaptive or adaptive scans. In non-adaptive search both dwell
time and azimuth are kept constant over the whole scan. In an adaptive scan, both the dwell and
The propagation factor F is defined as the ratio of the electric field that exists in the presence of the
earth's atmosphere and surface divided by the electric field that would exist in free space. The
propagation factor includes the effects of reflection, refraction, and diffraction. The propagation factor
F is calculated using the commercial program TERPEM developed by Signal Sciences Limited, U.K.
[17]. TERPEM is based on the parabolic equation,
∂ 2u ∂ 2u
+ 2ik + k 2 (m 2 ( x, y ) − 1) u = 0 (6)
∂z 2
∂x 2
where u is the electric or magnetic field component, k is the wave number, and m is the modified
refractive index.
The parabolic equation is a very popular and powerful tool for solving radiowave propagation
problems. It outputs a very good approximation of the full electromagnetic wave equation [20, 21].
This equation is solved numerically through marching techniques. The user specifies the initial values
of the electric field at or close to the antenna. A solution is found for the first iterative cell by satisfying
the boundary conditions at the bottom and the top of the cell. This solution is then passed to the next
cell and so on. An advantage of the marching technique is that it can handle changing conditions along
the range coordinates. The solution u(x,z) represents the electric field that would be measured. The
electric field that would exist in free space can be calculated using the radar range equation. The
propagation factor F can then be calculated from these two values of the electric fields [17]. Figure
11 shows an example of the output produced by TERPEM. The usual lobed structure caused by
multipath interference is easily recognizable. We can also see that electromagnetic energy propagates
beyond the horizon due to ducting effects and to a lesser extent diffraction. For our simulator,
TERPEM is directly integrated into MFRSIM and is completely transparent to the user.
The angular limits of integration extend several beamwidths (n) past the maximum beam centerline to
include the contribution from the whole antenna pattern [4]. This is equivalent to integrating from zero
to n(BW) and then multiplying the result by 2 as is done in Equation (7).
The model used to calculate the mean reflectivity of the sea is taken from the NATO Above-Air
Warfare System (NAAWS) study initiated in late 1980's [18]. This study was driven by the growing
threat of sea-skimming and high diving missiles [1]. The NAAWS sea clutter model is an empirical
model in which the mean reflectivity (in dBm2/m2) is calculated as
x sea = xref + K g + K s + K p + K d (8)
where x ref is the mean reference reflectivity and the K terms are correction factors.
The reference value x ref is the mean reflectivity measured at a given frequency for sea state 5, 0.1
degree grazing angle, vertical polarization, and zero degree wind direction. The K correction factors
calculate the mean reflectivity correction for different conditions of grazing angle, sea state,
polarization, and wind direction [18]. The frequency spectrum of the sea clutter Ssea(f) is assumed to be
Gaussian [4] as in Equation (9).
− ( f − f sea ) 2
Psea
S sea ( f ) = e 2σ sea 2
(9)
(2π )σ
sea
The mean Doppler frequency of the sea, fsea, and the standard deviation αsea, are calculated as
0.2ν wind cos(φ a − φ w )
f sea = (10)
λ
0.2 (0.125) ν wind cos(φ a − φ w )
α sea = (11)
λ
where ν wind is the mean wind speed, φ w is the wind direction and φa is the beam azimuth angle.
These equations define the sea clutter spectrum.
where Krain is a constant, and Wc is the chaff density in g/km3 [4]. The frequency spectrum of the rain
and chaff clutter is assumed to be Gaussian as in Equation (9). The mean frequency "f" and standard
deviation "α" are calculated as
0.2 σ shear + σ turb
2 2
0.2 ν wind cos(φ a − φ w )
f volume = ; α volume = (16)
λ λ
where σshear and σturb describe the effects of wind shear and turbulence. Their values are given in
reference [0]. These models are all describe in radar books and based on experimental data [4].
For birds, a model based on density of the flock (ρr) and relative RCS gain variations (β1, β2) in
decibels is used to determine the reflectivity [5]:
Sensitivity time control can be used to reduce reflected power clutter at short-ranges. In MFRSIM, the
receiver gain is modified according to Equation (21).
n
R mod Runamb
g rx = ( g − g 0 ) + g0 (21)
Runamb
where g and g0 are the maximum gain and the gain offset at zero range respectively [4]. The user
chooses the value of the range exponent n. Usually, n=3 for land and sea clutter reduction and n=2 for
rain clutter rejection. MFRSIM also includes simple models for receiver non-linearity and power due
to thermal, tropospheric, and phase noises [5]. From these models, we can calculate the spectrum of
the mean power at the output of the receiver, which is a combination of the missile, clutter, jammer,
and noise power.
1
ln (2) β
α log
P fa
Pd = 10 −u ; where u = 2 (25)
nnc3 SNR
The value of the parameters α and β are given in Table 1. The equations of Neuvy are based on the
theory of Swerling and Marcum [24, 25] and give Pd when nnc pulses are incoherently integrated.
0 1 + 2 exp(-nnc/3) 0.167
1 0.67(1 + 0.67 exp(-nnc/3) 1
2 1 0.167 + exp(-nnc/3)
3 0.75(1 + 0.67 exp(-nnc/3) 0.67
4 1 0.67 exp(-nnc/3)
Although we use the equation of Neuvy to model the square law detector, this is not strictly correct.
The equations of Neuvy have been developed for a SNR that has a constant mean and standard
deviation from pulse-to-pulse. In our case, the frequency can change from burst-to-burst and
consequently the mean and standard deviation will change with each burst. For example, the impact of
multipath and ducting are very dependent on frequency. In order to use the equation of Neuvy, we
calculate a mean SNR as follows.
#bursts
1
S NR = ∑ SNRn
# bursts n =1
(26)
The equations of Neuvy were developed for sinusoidal signals competing against white Gaussian noise.
In our model, the values used for the SNR are the outputs from the Doppler filters, which may contain
residual clutter or jamming power and hence provide different statistics to white gaussian noise. In any
case, MFRSIM currently uses the equations of Neuvy to calculate the probability of detection. In the
future, however, more accurate expressions for these calculations may be available.
Usually, the probability to exceed the first threshold is the same for all events. In this case, the
probability that the first threshold will be exceeded at least M times in N events is calculated from
Equation (27).
N
N!
Pd = ∑C
i=M
iN ( p d ) i (1 − p d ) N −i ; ciN =
i!( N − i )!
(27)
Pd is the probability of detection of the dwell and pd the probability of detection of the bursts [26].
However, in our simulation, the SNR changes from burst-to-bust and so the probability of detection per
Equation (28) might seem difficult to implement on a computer since we need to generate all
combinations of M-of-N events. Speed is an important consideration, due to the large number of
combinations that must be calculated for each missile intercept. There is a simple solution to
implement Equation (28). One simply counts in base 2 from zero to the largest number with N digits.
For example, let us generate all of the combinations of three binary events. Each event may either have
a value of 0 or 1. The maximum value of a three-digit number in base two is 111, which is 7 in base
10. Counting from zero to 7 in base then generates all the combinations. In our model, the zeros
represent the misses while the ones are the hits. A matrix c N
I ≥M of all combinations having at least M
hits on N events can be built by removing all other combinations. The zeros and ones have to be
r r
replaced by the appropriate probability of detection. Two vectors Pd and Pnd are built using the
probability of detection and non-detection of each burst within the dwell.
Pd of burst one 1 − Pd of burst one
r Pd of burst two r 1 − Pd of burst two (29)
Pd = ; Pnd =
" " " " " " " "
P 1− P
d of burst N d of burst N
The matrix of appropriate probability for all combinations of at least M hits is then obtained from
r r
M iN> M = C iN≥ M Pd + δ (C iN≥ M ) Pnd (30)
where δ represents the Kronecker delta function and operates on each element of the matrix.
1 ; if x = 0
δ ( x) = (31)
0 ; if x = 1
The Pd for the M-out-of-N detector can now be calculated from Equation (28) by multiplying the
element of each row of the matrix probability, M iN> M , and then summing all the rows together.
There are many plotting options available within MFRSIM that serve as analytical, diagnostic and
output display tools. Diagnostic information can be gleaned from many plots, which can display a
separation of the power levels of targets, clutter and noise, as well as the individual performance of the
various processing options (Doppler, MTI filtering, non-coherent integration). MFRSIM also provides
pre-run analytical plotting tools, such as blind zone and (DTED) terrain displays, which allow the user
to determine appropriate radar parameter settings. A post-simulation output display is also provided,
which displays the actual recorded and processed plots on a simulated PPI screen.
- calculator mode
- simulation mode without tracker
- simulation mode with (MT) tracker
To aid in obtaining quick results and for preliminary radar system evaluation, a radar calculator mode
has been included in the package. In this mode, a stripped, non-causal version of the simulator is run,
with a constant velocity missile approaching the radar’s (locked) azimuth boresight, either at a constant
height or along the antenna’s elevation boresight. In the presence of only system noise and surface
back-scatter interference, results illustrating properties such as propagation factors, SNRs, SIRs and
detection probabilities are produced. TERPEM may be used in the calculator mode.
The other two modes (simulation mode with and without tracker) are causal in nature and provide a
complete simulation run, making available to the user all the functionality of MFRSIM.
Figure 16. SIR plot from calculator mode (default parameters + Doppler processing)
The beam pattern plots of the type shown in Figure 5-6 are a unique feature of MFRSIM. The entire
history of the radar beam’s scan pattern and the missile trajectory, or a portion thereof, can be
displayed, in both azimuth and elevation, for three different antenna coordinate systems (electrical
centre, mechanical centre and absolute). These plots provide a rich source of information, and it is
recommended that users become familiar with their form and interpretation. The two angles of interest
are of course the elevation angle and the azimuth angle of the radar beam. Beams (surveillance,
confirmation, tracking and cueing) are represented by different coloured circles in Figure 5-6 and the
missile trajectory by coloured lines. For the current example, both surveillance (blue) and confirmation
(green) beams are displayed. When intersections occur at a particular time, for example around 7.8
seconds and 8.7 seconds, one or more beams (indicated by circles) will of course appear on or near the
corresponding missile trajectory. These intersections are easily seen in Figure 5-6. Whenever we have
intersections, the actual antenna gains at the intersection points are calculated; these are displayed in the
bottom graph of Figure 5-6. Before detailed radar calculations (such as SIR and detection) are
performed, the antenna gains must be large enough to make detection possible,
The causal nature of a basic simulation (Full Mode) arises because beams are directed and radar
processing (detection and estimation) is performed from one dwell to the next, dictated by a resource
manager that decides when and where to paint a particular beam, as would be the case for an
operational radar.
Much more complex radar systems can be modelled than the default radar example just considered. The
following example presents a much more complex situation. A simulation is created that uses multiple
radars in a single simulation. In particular, three radars are modelled: the Sea Giraffe (a rotating C-
Defining the environmental settings is a straight forward operation. By selecting the environment
option from the parameter menu, five parameter windows are accessible, organized as follows: general,
volume clutter, surface clutter, angel clutter, anomalous propagation. Again, the user walks through
each of these screens and makes the appropriate settings. In the example that follows, only surface
clutter is modelled; all other sources of clutter are turned off.
Lastly, the trajectory and dynamics of the incoming missile(s) must be specified. The window, selected
with Parameters >> Missile, is shown below.
As seen in Figure 5-7, a very simple missile trajectory is chosen. It travels in a straight line, starting 30
km out from the target towards the target, from due North at a speed of about Mach 1.5. The missile
travels 10 m above the sea surface, and has an RCS of 0.75 m2. Figure 5-7 reveals that much more
complicated flight paths are possible. The parameters are divided into two sections; the top three rows
establsih initial conditions and the bottom three allow specific characteristics to be specified for any
number of trajectory “legs”. Within this framework, the user is able to define complex flight paths by
Having defined our three radars, environment and missile trajectory, the user is now in a position to run
the simulation. The simulation runs in a similar manner as during the default run performed earlier.
Timing information (with the occasional diagnostic) appears in the MATLAB window, and, depending
on the speed of the user’s computer, a progress bar may be spotted when beam intersections are being
processed. After the simulation has been run, the user would like to know if it has been successful, or,
more particularly, whether the target missile was detected. Probably the best way to check this is to
open a PPI display, available from the pull-down plot menu (Plot>> Open PPI Display…). The
following four plots show different views of the PPI display for this example.
Figure 5-8 clearly shows that detections were received and processed by all radars. With knowledge of
radars involved, the results seem plausible. The APAR face has clearly performed the best, with “rock
solid” detections, which are hardly surprising since it possesses the highest resolution and power-
aperture product and fastest revisit time (there is no rotation). The Sea Giraffe has detected the target
most of the time, with the dropouts probably due to the rotation. The SPS-49 made some detection at
the further ranges but seems to have had dropouts at nearer ranges. This may be explained by the
change in waveform the SPS-49 applies for ranges under 20 km, which is for chaff suppression.
It is a simple matter to use the tracker; the File >> Run Simulation option (Figure 5-3) does it by
default. It is, however, important to ensure that the radar parameters are appropriate for tracker
operation and interaction. They can be checked and adjusted through the SEARCH RADAR REGION
PARAMETER GROUP window. For this example, we shall only interface the Sea Giraffe and APAR
systems to the tracker. The portion relevant to the new Sea Giraffe settings is reproduced below.
Figure 5-9 shows that we are now interfacing with the tracker. This produces a two-fold effect – it
enables the radar to send confirmed detections to the tracker for track association and update, as well as
letting the radar request predictions for utilisation of tracking. As shown above, we have chosen not to
utilise any tracking beams for the Sea Giraffe, since it is a fixed antenna radar without monopulse; the
track request interface retrieves 3-D information which is more suited to beam agile systems such as
APAR. We have already set up APAR’s tracking parameters and track time loading, thus it is only
necessary to ensure that it is selected (“yes”) to interface with the radar.
The tracker appears once the run has been started, and remains after the conclusion of the simulation
until the user presses a key (while in MATLAB’s main window). A snapshot of part of the tracker’s
display is shown below, where, for illustrative purposes, we have turned the persistence up so the
complete track history is available. The result has produced a solid track at the expected location.
Figure 5-11 below also shows the PPI display of the tracking beams from APAR, which consistently
has painted the missile throughout its trajectory.
In the future, MFRSIM will be used to study adaptive techniques using MFRs for different missions,
environments, and threats. MFRs have a high level of agility in the transmitted waveform and antenna
beam steering, making it possible to change scan strategies or scheduling “on the fly”. Depending on
the environment and the threat, certain waveforms should provide better performance over others. The
naval AN/SPY-1 radar is a good example of an advanced MFR that can modify its waveform and beam
scheduling (as well as signal processing) to adapt to different missions, environments, and threats.
Current emphasis on studying littoral regions where reaction time is short, make adaptive techniques
more vital. The modified simulator will be used to evaluate the impact of dynamic waveforms, random
scan search on the MFR performance and to optimize the transmitted waveforms for selected scenarios.
The objective is to develop adaptive algorithms to continuously optimize the performance of naval
radars in the ever-changing environment. The real problem is how to implement the logic to apply
adaptability to the MFRs. This problem is tackled in collaboration with TTCP SEN TP 3 “Radar
systems”, which is developing an adaptive MFR simulator. The intention is to develop intelligent
controller of an MFR through advanced algorithms and processing. MFRSIM is one of the key
components considered for the development of this adaptive MFR simulator. The intention is to use
the modified simulator to optimize the parameters of the transmitted waveforms and beam scheduling
as a function of the threat and environment. The results of the studies should provide a basic reference
for the development of potential schemes for future MFRs.
MFRSIM will also be used to evaluate the impact of sensor integration and various “hand-off”
techniques and cued search methods for the overall air defence system. This research will be done in
support of the Technology Demonstration Project on “Shipborne Integration Sensors and Weapons
Systems” (SISWS). Sensor integration, as suggested by the NAAWS study, is also becoming much
more important to win back reaction time. Presently, the modeled MFR in MFRSIM can be cued from
adjunct radars such as a volume search radar. New models are required for additional sensors such as
infrared search and track sensors or ESM systems (such as CANEWS 2). MFRs are able to look in any
direction almost instantaneously. An attractive “hand-off” technique would use this capability to
confirm detections by adjunct radars immediately rather than to wait confirmation from these slow
rotating radars.
Eventually, MFRSIM will be further modified to evaluate other concepts such the Cooperative
Engagement Capability, and netcentric multiplatform naval surveillance. The simulation architecture
of MFRSIM has been designed to allow for such modifications.
1. Martin, R., Toulgoat, M., (Jan 1997), Multifunction Radars in the NAAWS
tradition, Defence Research Establishment Ottawa, report No. 1309
2. Lightstone L. Vrzic S., (Nov. 1992), Simulation Studies for the horizon search
performance of a multifunction array radar, Atlantis scientific systems group inc.,
Contract number W7714-2-0043, Report number 288
3. Lightstone L., Vrzic S., (March 1993), Further Simulation Studies for the horizon
search performance of a multi function array radar, Atlantis Scientific systems
group inc, contract number W7714-2-0078, Report number 302
4. Lightstone L., Enweani B., (March 1995), New simulation studies for the horizon
search performance of a multi function radar in a clutter environment, Atlantis
scientific systems group inc., Contract number W7714-4-0162, Report number
363
5. Jones, G., and al, (March 1999), Rotating Multifunction Radar System Simulation
and Analysis, VolI and II, Sicom systems Ltd, Contract number W7714-7-
0104/001/SV
6. Jones, G., and al, (24 July 1994), Optimisation of multifunction radars through
simulation, Vol I: Software, experiment descriptions and unclassified simulation
results, Sicom System ltd, Contract number W7714-8-0220, Sept 200Lok J.J,
Canada and Germany Join Dutch for APAR, Jane’s Defence Weekly, pp8
7. Jones, G., and al, (Sept. 2000), Optimisation of multifunction radars through
simulation, Vol II: Software, experiment descriptions and classified simulation
results, Sicom System ltd, Contract number W7714-8-0220
8. Jones, G., and al, (March 2001), Modification to an adaptive Multifunction radar
simulator, Sicom systems Ltd, Contract number W7714-000370/001/SV
9. Jones, G., and al, (March 2002) Dynamic Multifunction radar simulator, Sicom
systems Ltd, Contract number W7714-000370/001/SV
10. Lok J.J., (21 Aug 1993), Contract Signing Advances APAR, Jane’s Defence
Weekly, pp21-22
11. Hobson S, (18 July 1992), APAR builds a cleaner image, Jane’s Defence Weekly,
18July 1992, pp17-18 Lok, J.J., Joint Naval Radar Project Planned, Jane’s Defence
Weekly, pp14-18
14. Lok J.J., (21 Aug 1993), Contract Signing Advances APAR, Jane’s Defence
Weekly, pp21-22
16. Hosbon S, (Jan 1999), Dutch contract pushes APAR radar sales to seven while
Canada waits in the wings, Jane’s International Review, pp7
18. Reilly, J.P., (June 1998), Clutter Models for shipboard radar application, NATO
AAW System Program Office naval sea Command
19. Huizing, A.G., Theil, A., (1993) Carpet Computer-aided radar performance
evaluation tool: User’s Manual, Artech House
20. Barrios, (Jan 1994), A terrain parabolic equation model for propagation in the
troposphere, IEEE Transactions on Antennas and propagation, Vol. 42, No. 1
21. Craig, K.H., Levy, M.F., (Apr 1991), Paraboloic Equatiuon modelling of the
effects of multipath and ducting on radar systems, IEE Proceedings-F, Vol. 138,
No. 2
22. Skolnik, M. I., (1991), Radar handbook 2nd Edition, McGraw-Hill, Toronto
23. Neuvy, J., (July 1970), An aspect of determining the range of the radar detection,
IEEE Trans Aero Elec Syst, Vol AES-6, No 4
24. Marcum, J.I., (Apr 1960), A statistical theory of targets detection by pulsed radar,
IRE Trans information Theory, Vol IT-6, pp59-145
25. Swerling, P., (Apr 1960), Probability of detection of fluctuating targets, IRE Trans
information Theory, VOL IT-6, pp269-308
26. Bird, J.S., (Nov 1985), Low Doppler target detection in ground clutter, CRC report
No 13977
1. ORIGINATOR (the name and address of the organization preparing the document. 2. SECURITY CLASSIFICATION
Organizations for whom the document was prepared, e.g. Establishment sponsoring a (overall security classification of the document,
contractor’s report, or tasking agency, are entered in section 8.) including special warning terms if applicable)
3. TITLE (the complete document title as indicated on the title page. Its classification should be indicated by the appropriate
abbreviation (S,C or U) in parentheses after the title.)
5. DATE OF PUBLICATION (month and year of publication of 6a. NO. OF PAGES (total 6b. NO. OF REFS (total cited in
document) containing information. Include document)
Annexes, Appendices, etc.)
December 2002 53 26
7. DESCRIPTIVE NOTES (the category of the document, e.g. technical report, technical note or memorandum. If appropriate, enter the type of
report, e.g. interim, progress, summary, annual or final. Give the inclusive dates when a specific reporting period is covered.)
Technical Memorandum
8. SPONSORING ACTIVITY (the name of the department project office or laboratory sponsoring the research and development. Include the
address.)
9a. PROJECT OR GRANT NO. (if appropriate, the applicable research 9b. CONTRACT NO. (if appropriate, the applicable number under
and development project or grant number under which the which the document was written)
document was written. Please specify whether project or grant)
11ar11
10a. ORIGINATOR’S DOCUMENT NUMBER (the official document 10b. OTHER DOCUMENT NOS. (Any other numbers which may
number by which the document is identified by the originating be assigned this document either by the originator or by the
activity. This number must be unique to this document.) sponsor)
11. DOCUMENT AVAILABILITY (any limitations on further dissemination of the document, other than those imposed by security classification)
12. DOCUMENT ANNOUNCEMENT (any limitation to the bibliographic announcement of this document. This will normally correspond to
the Document Availability (11). However, where further distribution (beyond the audience specified in 11) is possible, a wider
announcement audience may be selected.)
UNCLASSIFIED
SECURITY CLASSIFICATION OF FORM DCD03 2/06/87
UNCLASSIFIED
SECURITY CLASSIFICATION OF FORM
13. ABSTRACT ( a brief and factual summary of the document. It may also appear elsewhere in the body of the document itself. It is highly
desirable that the abstract of classified documents be unclassified. Each paragraph of the abstract shall begin with an indication of the
security classification of the information in the paragraph (unless the document itself is unclassified) represented as (S), (C), or (U).
It is not necessary to include here abstracts in both official languages unless the text is bilingual).
(U)DRDC Ottawa has developed a multifunction radar (MFR) software simulator through several contracts with
Atlantis Scientific Inc and Sicom System Ltd. The latest version, called MFRSIM, is coded in MATLAB version 6.1
(by MathWorks Inc), running on an IBM compatible PC. MFRSIM has been developed specifically to evaluate the
detection capability of an MFR against anti-ship missiles (ASM) operating in a littoral environment. Both rotating and
non-rotating phased array MFRs can be simulated, as well as conventional rotating antennas such as volume search
radars. The simulation can include land clutter, sea clutter, chaff clutter, rain and angel clutter with jamming.
MFRSIM produces detection outputs in a causal manner one dwell at a time. The detector outputs are sent to the beam
scheduler and the associated tracker. The tracker automatically initiates new tracks and interfaces with the beam
scheduler to request tracking dwells. The beam scheduler controls the scheduling of surveillance; confirmation, cued
search and tracking beams based on the tracker and detector outputs. The parameters of the transmitted dwell can be
changed from dwell to dwell. A procedure has been developed to automatically select the waveform parameters for
the scheduled dwell from a library of predefined waveforms. Although this capability is still rudimentary, the intention
is eventually to select a given waveform as a function of the radar returns and environmental data.
(U) This report describes the major capabilities, features, models and applications of MFRSIM. The first section
provides background information on the development of the DRDC Ottawa simulator over the years. The second
section provides an overview of MFRSIM, including its modules and capabilities. The radar models of MFRSIM are
examined in detail in the third section. This will provide a clear understanding of the capabilities and limitations of
this simulator. The fourth section explains how to use MFRSIM and shows some examples. The last section
discusses the future R&D plans for development of MFRSIM and its main applications.
14. KEYWORDS, DESCRIPTORS or IDENTIFIERS (technically meaningful terms or short phrases that characterize a document and could be helpful
in cataloguing the document. They should be selected so that no security classification is required. Identifiers such as equipment model
designation, trade name, military project code name, geographic location may also be included. If possible keywords should be selected from a
published thesaurus. e.g. Thesaurus of Engineering and Scientific Terms (TEST) and that thesaurus-identified. If it is not possible to select
indexing terms which are Unclassified, the classification of each should be indicated as with the title.)
UNCLASSIFIED
SECURITY CLASSIFICATION OF FORM
Defence R&D Canada R & D pour la défense Canada
www.drdc-rddc.gc.ca