1

Radar Fundamentals

1.1 INTRODUCTION
This chapter introduces some of the major radar concepts that will be described
in greater detail in subsequent chapters. One key concept, however, is covered
primarily in this chapter: the radar range equation (RRE). Although specific
forms of the RRE are discussed in other chapters, its derivation and definitions
of all major terms are covered here. The remainder of the chapter provides an
overview of other significant radar concepts.

1.2 SEARCH AND TRACK FUNCTIONS

Typically, the functions performed by radars enable the sensing of a target’s
presence and its physical location and the ability to predict a target’s future posi-
tion. These basic capabilities are fundamental to military radars, such as for fire-
control applications, and commercial radars, such as those for air-traffic control
(ATC).
The major functions that radars can execute include:

• Search:
– To survey a volume of space and report locations of targets
– To measure target position in two or three dimensions (i.e., 2-D or 3-D)

1
2 Chapter 1

• Tracking:
– Obtain more accurate target location by “smoothing” measurements
– Estimate the “state vector” of the target (i.e., positions, rates, and possibly
accelerations) to predict the target state vector for a future time
• Track-While-Scan:
– Combine the search and tracking functions into one radar mode
– Use data processing to initiate and maintain tracks, while simultaneously
searching for new targets
– Add tracking capability without using additional radar resources.

In order to provide these high-level capabilities, many additional lower-level

functions must be executed by radars. Some of these functions are described in
the following sections.

1.3 TARGET DETECTION, RESOLUTION, AND CLUTTER

CONCEPTS
Three key concepts affecting radar operation are target detection, radar resolu-
tion, and clutter backscatter (commonly referred to as simply “clutter”). The fol-
lowing subsections provide descriptions of each and their importance to the
operation, capabilities, and performance of different radars.

1.3.1 Target Detection

Radar energy is typically transmitted as a pulse or group of pulses at a carrier or

operating frequency ranging between tens and thousands of Megahertz (MHz).
The instantaneous bandwidth of each pulse can be on the order of a Megahertz
or less (i.e., narrowband operation) or a Gigahertz (GHz) or more (generally con-
sidered wideband operation). The reflected energy received by the radar is used
to decide whether a target is present or not. Usually this function is performed
automatically for targets at all ranges (from the radar) of interest. This is defined
as target detection. Target detection is the crucial first function performed by ra-
dars. In other words, target detection is a prerequisite for all subsequent radar
functions, such as tracking and target classification.
 Radar Fundamentals 3

The reflected radar energy (or waveform) is subject to additive thermal noise,
primarily due to active electronics in the receive chain. It is in this “noisy” envi-
ronment within which targets must be sensed. This radar return or “echo” is de-
picted in Figure 1.1.
Two key target detection attributes are the “probability of detection” (Pd) and
“probability of false alarm” (Pfa). The radar’s design objective is to maximize Pd
while maintaining a small (and possibly constant) Pfa.
Detectability (or Pd) depends on the ratio of reflected target energy to average
thermal noise power, which is defined as the signal-to-noise ratio (SNR). SNR is
an important radar performance metric. The performance of nearly all radar
functions depends on the SNR. For example, Pd increases monotonically with in-
creasing SNR.
Two approaches to increasing SNR are to use higher-energy (i.e., larger ampli-
tude or longer duration) waveforms or to add (or “integrate”) multiple return
pulses, either coherently (in-phase) or non-coherently (summing magnitudes or
without phase coherence).
A target-present decision when no target (only noise) is present is called a
“false alarm.” The converse to this error is declaring “no target present” when a
target is actually present. This latter error is referred to as a “missed detection.”
Multiple pulses can be transmitted as a pulse “train” or “burst.” This type of
waveform is depicted in Figure 1.2. If each pulse in a burst has energy E, then
the waveform illustrated in the figure has total energy NE.
The minimum “non-self eclipsing” range is based on the pulse duration τ and
is given by:

, (1.1)

where c is the speed of light. This phenomenon arises from the fact that recep-
tion cannot begin until transmission is complete. The maximum “un-eclipsed”
range is therefore defined as:

 T −τ  ,
R max = c (1.2)
uneclipsed  2 
4 Chapter 1

Figure 1.1 Illustration of Received Target Signal or Echo

Figure 1.2 A Radar Pulse Train or Burst Waveform

where T is defined as the pulse repetition interval (PRI). Alternatively, the pulse
repetition frequency (PRF) is defined as:

1 1 . (1.3)
PRF = =
T PRI

1.3.2 Radar Resolution

When targets are closely spaced in range, angle, or Doppler (proportional to tar-
get range-rate) they may not be “resolvable” by the radar. Resolution capability
is defined as the minimum separation necessary to declare two targets when
there are two targets present. Resolution is a function of waveform duration or
bandwidth (range resolution), antenna beamwidth (angular resolution), and co-
herent integration time (range-rate or Doppler resolution).
A single pulse is a commonly used radar waveform. A simple pulse (enve-
lope) of duration τ in time has a bandwidth, B, approximately given by:
 Radar Fundamentals 5

1. (1.4)
B =
τ

Figure 1.3 shows an ideal transmitted rectangular radar pulse. Two targets (of
equal amplitude) separated by much greater than τ in time are easily “resolved.”
The inherent range resolution capability of the waveform is defined as:

c .
δR = (1.5)
2B

In real-world applications, to ensure target resolution with a high probability,

targets must be separated by 2 to 3 δR depending on the targets’ relative sizes
and the time-sidelobes of the matched filter output, as described in a subsequent
section of this chapter.

1.3.3 Clutter Backscatter

The reflected energy received by the radar may originate from unintended ob-
jects such as land or sea backscatter, weather (e.g., rain), or man-made objects
such as buildings or other structures. This type of radar return is referred to as
“clutter.” When the radar is used for weather detection, clutter is the desired tar-
get.
A key discriminating feature used to differentiate targets from clutter is the
observed or measured speed (or range-rate). This method is based on exploiting
the Doppler-shifting effect, that is, the phenomenon that returns from higher

Figure 1.3 Simple Rectangular-Pulse Waveform

6 Chapter 1

range-rate targets of interest are “shifted” in frequency more than lower-speed

targets, which in this case are clutter returns. The Doppler frequency-shift based
on the target speed is given by:

, (1.6)

where Ṙ is the target or clutter range-rate and λ is the radar operating wave-
length.
A second key discriminating feature used to separate targets from clutter is
the “polarization” of the reflected electromagnetic (EM) signal. This is especially
true for clutter arising from rain. The return from spherical rain drops when us-
ing a circularly polarized waveform is reduced compared to the target response
when a linearly polarized waveform is employed. Different polarizations will
reflect (or refract) distinctly from different shaped targets; for example, sharp
edges tend to spread energy into the primary and orthogonal polarizations,
whereas smooth reflectors result in a single reflected polarization. Polarization is
an important feature for separating certain classes of targets.

1.4 SURVEILLANCE RADARS

Surveillance radars perform an important role for many missions. A surveillance
radar’s objective is to detect objects in a volume of space and to “acquire” them
(i.e., initiate a target track). Usually, these target tracks are handed off to tracking
radars or, for multifunction radars, to other radar functions such as tracking or
target classification.
A search radar illuminates the volume (i.e., radiates energy into it) with one or
more antenna beam positions on transmit. The antenna may be mechanically or
electronically scanned across the coverage volume. The “scan” or “frame” time
is the time necessary to illuminate the entire volume once.
Most radars use the same antenna to transmit and receive energy. A switch or
radio frequency (RF) duplexer is used to direct energy from the transmitter to
the antenna or from the antenna to the receiver. The received waveform is very
weak relative to the transmitted waveform and must be amplified in the receiver
prior to making detection decisions. The receiver also translates the RF energy
 Radar Fundamentals 7

from its operating frequency to a lower intermediate frequency (IF) or baseband

for further processing.
There are a number of different types of searches that these radars can execute.
The following subsection describes one of the major search types: volume
search. Chapter 4 covers a number of other commonly used searches.

1.4.1 Volume Search

Most surveillance radars that operate in general air search modes, such as in air
defense applications (e.g., search and tracking of air targets: aircraft, cruise mis-
siles, unmanned air vehicles), perform volumetric-type searches. As indicated
by its name, volume searches survey a volume defined in space, specified by a
range extent, and azimuth and elevation extents. A typical volume search
beam-pattern (or “raster”) employed by a phased-array radar is depicted in
Figure 1.4.
A volume search is specified by the physical extents defined in Figure 1.4, the
time required to execute the search (i.e., the frame time), the allowed false alarm
rate, and the cumulative probability of detection, Pd cum.

1.5 RADAR BLOCK DIAGRAM

Figure 1.5 depicts the block diagram of a basic phased-array radar. As can be
seen, the major functions are the phased-array antenna, beam steering generator,

Figure 1.4 Volume Search Beam Raster for Phased-Array Radars

8 Chapter 1

Figure 1.5 Basic Phased-Array Radar Block Diagram

waveform generator, exciter and transmitter, receiver, signal processor, data pro-
cessor, and operator displays and controls.
Note that for reflector-antenna radars, the phased-array is replaced with a re-
flector antenna. Also for mechanically scanned radars, the beam steering genera-
tor is replaced with antenna pedestal control. In the case of solid-state transmit-
receive configuration, the transmitter is replaced with T/R modules (this type of
radar is referred to as active-aperture radar). Most often in modern radars, the
signal processing is performed in software, and the signal processor and data
processor are combined into a single signal/data processor (i.e., computer or
multiprocessor).
Figure 1.6 is a block diagram of a phased-array antenna (also known more
generally as an electronically-steered array).
A phased-array antenna is a “sampled aperture,” where the antenna consists
of N small antennas or antenna elements. Each is “phased-steered” to receive or
transmit from an angle θ steer using phase shifters. The phase shifters are usu-
ally digitally selected to obtain quantized phase (e.g., 0, 22.5°, 45°, 67.5°, 90°).
This type of antenna is used for narrowband operation. When wideband opera-
tion is required, some combination of time-delay and phase-shift steering is
needed.
 Radar Fundamentals 9

Figure 1.6 Basic Phased-Array Antenna Block Diagram

1.6 RADAR RANGE EQUATION

The radar range equation is the fundamental relationship that defines radar per-
formance for a given set of radar parameters (e.g., peak transmit power, transmit
antenna gain, receive antenna gain, wavelength), or the tool used to design a ra-
dar to satisfy certain performance requirements (e.g., single-to-noise ratio on a
target with a specified radar cross section [RCS] at a specified range from the ra-
dar). Figure 1.7 illustrates the basic derivation of the RRE based on the laws of
physics such as electromagnetic scattering, EM wave propagation in a vacuum,
and so on. As can be seen in the figure, the transmitted power density is defined
as:

Pt Gt
Transmit Power Density = , (1.7)
4π R 2 Lt

where Pt Gt is the product of the peak transmitter power and the transmit an-
tenna gain, and R2 Lt is the product of range of the target from the radar squared
and the total transmit losses. For a target with an RCS of σ, the echo or reflected
power density is given by:
10 Chapter 1

Figure 1.7 Derivation of the Radar Range Equation

Pt Gt σ
Echo Power Density = . (1.8)
( 4π R )
2
2
Lt

The received power at the radar antenna aperture of area Ar and receive loss of
Lr is then:

Pt Gt σ Ar
Power at Radar Aperture = . (1.9)
( 4π R )
2
2
Lt Lr

When the thermal front-end noise can be modeled as having a “white” spectrum
with power spectral density of kTs, that is, the product of Boltzmann’s constant
and the system noise temperature, and the radar has a processing (or noise)
bandwidth of B, the noise at the output of the matched filter is given by:

Noise Power = k Ts B . (1.10)

 Radar Fundamentals 11

Therefore, the average signal power-to-RMS noise power, commonly referred to

as the SNR, is defined as the ratio of equation (1.9) to equation (1.10), or:

Pt Gt Ar σ
SNR = . (1.11)
( 4π R2 )
2
k Ts B Lt Lr

If the antenna gain is defined as:

4π Ar , (1.12)
Gr =
λ2

then solving for the receive aperture and substituting into equation (1.11) yields
the commonly used “sensitivity” form of the RRE:

Pt Gt Gr λ 2 σ
SNR = . (1.13)
( 4π ) 3 4
k Ts B R Lt Lr

The RRE is often calculated in decibel units via:

SNR dB = 10 log10 SNR . (1.14)

The evaluation of equation (1.14) in tabular form is referred to as a “Blake chart”

(named after L. V. Blake). An example of a completed Blake chart for an X-band
radar appears in Figure 1.8.

1.6.1 Jamming Effects on Signal-to-Noise Ratio

Equations (1.11) and (1.13) are referred to as the RRE for radars in the “clear”
(i.e., operating in a thermal noise environment only). When severe intentional or
unintentional interference, commonly referred to as jamming, are present and
the interference is much larger than the thermal noise, the importance of the
SNR is supplanted by the signal-to-interference ratio (SIR) given by:

S S 1
SIR = ≈ = . (1.15)
I N+ I ( SNR ) −1
+ ( SIR )
−1
12 Chapter 1

Figure 1.8 Example of a Blake Chart for an X-Band Radar

 Radar Fundamentals 13

Therefore, when the interference level is much lower than thermal noise level,
the equations (1.11) and (1.13) are a useful performance figure-of-merit (FoM) for
radars. However, when the interference level approaches the noise level and ex-
ceeds it, the SIR defined in equation (1.15) becomes the performance FoM.
As can be seen from equation (1.15), when the thermal noise is exceeded by
the thermal noise-plus-interference, target detection sensitivity is reduced and is
now limited by the SIR. Mathematically, this can be expressed as:

2 2
P J G J G rj λ P J G J G rj λ
kT s B → kT s B + -------------------------
- ≈ -------------------------
-, (1.16)
2 2
( 4πR J ) ( 4πR J )

where PJ GJ is the product of the jammer power and antenna gain (referred to as
the effective radiated power of the jamming source), RJ is the range of the jam-
mer to the radar, Grj is the radar receive antenna gain in the jammer direction,
and jammer bandwidth ≥ B. Note that the jamming energy impinging on the ra-
dar antenna aperture decays as an inverse-range-squared law, compared with
the target return, which goes as inverse-range to the fourth power. This enables
relatively low-powered jamming sources to degrade radar performance, which
is the inherent advantage of the jammer over radars.
Substituting equation (1.16) into (1.13) yields the RRE for the case where jam-
ming levels far exceed thermal noise:

Pt Gt Gr λ 2 σ RJ2
SIR = . (1.17)
( 4π ) 2
PJ GJ R 4 Lt Lr

1.6.2 Other Forms of the Radar Range Equation

There are many forms of the RRE that can be derived from equation (1.13) for
specific applications. The RREs used for search and tracking are fundamental
variants. These are addressed in the subsequent sections.

1.6.2.1 Radar Range Equation for Volume Search. For volumetric searches as de-
picted in Figure 1.4, the RRE is adapted in the following manner. First, note that
the antenna beamwidth is related to the antenna gain via the relationship:
14 Chapter 1

λ λ π
θ3 ≈ = = 2
A λ2G G . (1.18)

4π

The area of a search beam is approximately given by:

π .
Ab ≈ θ3 AZ θ3 EL = 4 (1.19)
G
Also, the average transmitted power can be defined as:

Pt PRF . (1.20)
PAVE = Pt DF = Pt τ PRF =
B
where DF is the radar duty factor (percentage of time allowed to transmit) and
PRF is the pulse repetition frequency. Noting that:

Ψ ΨG
= = PRF (or beams / s) , (1.21)
θ3 AZ θ3 EL TSC 4π TSC

where Ψ is the area to be searched in radians2 and TSC is the scan or “frame”
time for the search, and substituting equation (1.20) into (1.21) yields:

ΨG B PAVE ,
= PRF = (1.22)
4π TSC Pt

or:

PAVE 4π TSC B .
Pt = (1.23)
ΨG

Now substituting equation (1.23) into the RRE of equation (1.13) results in:

PAVE Gr λ 2 σ TSC σ TSC

SNR = = PAVE Ar . (1.24)
( 4π ) 2
k Ts R Ψ Lt Lr
4
( 4π ) k T s R 4 Ψ Lt Lr
 Radar Fundamentals 15

Equation (1.24) is proportional to the PAVE Ar product and is not a function of op-
erating frequency. Also note in contrast that equation (1.11), the sensitivity form
of the RRE, is proportional to the product Pt Gt Ar . Theoretically then, radars at
any operating frequency can search a volume equally well if they possess the
same PAVE Ar .
In practice, however, lower operating frequencies are usually employed for
search radars since the number of beams required to search a given volume, for
a fixed antenna aperture size, is many fewer than at higher frequencies. At
higher frequencies this can lead to timeline occupancy issues since the larger
number of beams required will demand that longer portions of the timeline be
scheduled. When these scheduling periods exceed the specified frame time, the
radar is said to be “occupancy-limited.”

1.6.2.2 Radar Range Equation for Horizon Fence Search. Early warning (EW) mis-
sile surveillance radars and ballistic missile defense (BMD) radars often employ
horizon fence searches to detect and acquire targets. Figure 1.9 depicts a typical
search fence that covers ± 60 degrees of azimuth.
The fence concept is based on the fact that if radars have adequate detection
range on the targets of interest, then any ascending ballistic target must fly
through a fence and will be detected. Therefore, instead of performing a radar
resource-intensive volume search (i.e., due to the much larger number of
beams), a single row of beams at or above the horizon is sufficient for missile
search and acquisition applications.
Starting with the RRE of equation (1.24) for volume search, modifications can
be made to take advantage of the horizon fence characteristics. First note that for
missiles flying through the fence:

θ3 θ3 R
TSC = = , (1.25)
E
T N vT

where Ė T is the target elevation rate, R is the target range, νT is the vertical tar-
get velocity, and N is the number of looks required for detection. Noting the
beamwidth relationship in equation (1.18), (1.25) can be expressed as:
16 Chapter 1

Figure 1.9 Typical Horizon Fence Search Pattern

2R π .
TSC = (1.26)
N vT G

Substituting equation (1.26) into (1.24) yields:

σ PAVE Ar
SNR = . (1.27)
( 2 π ) k Ts R3 Ψ N vT Lt Lr Gr

As can be seen from equation (1.27), the RRE for horizon fence search is weakly
a function of operating frequency due to the G term and is also follows an in-
verse R3 law rather than R4 as do the RREs in equations (1.13) and (1.24).

1.6.2.3 Radar Range Equation for Tracking. The primary driving requirement on
tracking is angle accuracy given by:

θ3
σθ = , (1.28)
km 2 SNR η

where km and η are the monopulse slope and number of independent measure-
ments smoothed by the tracking filter. The SNR is defined by the track sensitiv-
ity form of the RRE given by equation (1.11). Substituting average power for
peak power as defined in equation (1.20) yields:

PAVE Gt Ar σ
SNR = , (1.29)
)
PRF ( 4π R 2 2 k Ts Lt Lr
 Radar Fundamentals 17

Substituting equation (1.29) into the square of equation (1.28) yields:

θ32 PRF ( 4π ) R k Ts Lt Lr .
24

σ 2
= (1.30)
θ
2 km2 η PAVE Ar Gt σ

Now substituting for antenna beamwidth from equation (1.18) and noting that η
= PRF Tt where Tt is the time in track (and PRF refers to the track update rate)
yields:

σ 2
=
( 4π )
3
k Ts R 4 Lt Lr . (1.31)
θ
2 km2 Tt PAVE Ar Gt Gr σ

As can be seen, the tracking accuracy is inversely proportional to PAVE AG2, or

equivalently PAVE A3/λ4, and is therefore highly dependent on operating fre-
quency. For a given size antenna aperture, superior tracking accuracy is
achieved by higher-frequency radars.

1.6.2.4 Radar Range Equation Summary. Tables 1.1 and 1.2 provide a summary of
the RREs for the radar applications discussed in this section.

1.7 DETECTION IN NOISE

This topic is covered in detail in Chapter 2 for detection in noise, clutter, and
jamming environments, respectively. However, some of the basic concepts are
introduced in this chapter.
In general, detection performance is a function of SNR (or in cases of clutter or
jamming, signal-to-clutter ratio (SCR) or SIR, respectively) and typically em-
ploys threshold tests to declare target detections. References [6] and [7] address
the theory for this critical radar function. Most modern radars use matched-filter
receivers to maximize the SNR at the processed output prior to detection deci-
sions.
18 Chapter 1

Table 1.1 Forms of the Radar Range Equation (Square Antenna)

Radar
Radar Appropriate Form of the Parametric
Application Radar Range Equation Factor

σ TSC
SNR = PAVE Ar
( 4π ) k Ts R4 Ψ Lt Lr
Volume search PAVE A

σ PAVE Ar PAVE A
SNR =
Horizon fence search
( 2 π ) k Ts R Ψ N vT Lt Lr
3
Gr G

Pt Gt Ar σ
SNR = Pt AG
( 4π R )
Track sensitivity 2
2
k Ts B Lt Lr

Track accuracy σ 2
=
( 4π ) 3
k Ts R 4 Lt Lr
PAVE AG 2
θ
2 k Tt PAVE Ar Gt Gr σ
2
m

Table 1.2 Forms of the Radar Range Equation (Circular Antenna)

Radar
Radar Appropriate Form of the Parametric
Application Radar Range Equation Factor

σ TSC
Volume search SNR = PAVE Ar PAVE A
16 k Ts R 4 Ψ Lt Lr

πσ PAVE Ar PAVE A
Horizon fence search SNR =
16 k Ts R3 Ψ N vT Lt Lr Gr G

Pt Gt Ar σ
SNR = Pt AG
( 4π R )
Track sensitivity 2
2
k Ts B Lt Lr

Track accuracy σ 2
=
( 2π ) 4
k Ts R 4 Lt Lr
PAVE AG 2
θ
2 k Tt PAVE Ar Gt Gr σ
2
m
 Radar Fundamentals 19

The basic form of the detectors is:

H1
s(t) > VT , (1.32)
< VT
H0

where s(t) is the output of the optimal matched filter and VT is the detection
threshold. The detection threshold is usually based upon an assumption of
Rayleigh-distributed random noise magnitudes.

1.7.1 Target Models

Chapter 2 describes the performance of equation (1.32) for several analytical tar-
get models. For the purposes of this chapter, it is assumed that targets follow an
exponentially distributed random variation in power (equivalent to a Rayleigh-
distributed random fluctuation voltage model). This is referred to as a Swerling I
target model when returns are independent look-to-look (i.e., scan-to-scan
rather than pulse-to-pulse).
Detection is modeled as a statistical phenomenon with two sources of statisti-
cal or non-deterministic variability:

• Additive interference (e.g., noise, interference)

• Fluctuating target radar cross section.

The target’s RCS is an approximation devised to account for the portion of scat-
tered energy reflected back toward the radar from the target. The target RCS
may vary in amplitude (or fluctuate) versus time. The RCS depends on the RF
operating frequency of the radar (e.g., ultra-high frequency [UHF], L-band, X-
band) and on the reflecting structure of the particular target. Complex scattering
surfaces such as an aircraft are composed of many individual RF scatterers. Sim-
pler target shapes, like a conical missile warhead, may present only one or two
scattering sources.
The interpretation of the reflected energy scattered by the target then becomes
driven by the waveform range resolution:

c cτ eff . (1.33)
δR = =
2B 2
20 Chapter 1

When δR is ≥ target “length,” a single return occurs at the radar that is a sum of
the complex-valued individual scattering components. RCS fluctuations arise
due to constructive and destructive combination of individual scattering center
components within the resolution-limited waveform bandwidth.
However, when δR << target “length,” multiple scattering returns are re-
solved in range (resulting in little or no RCS fluctuation). Models of target RCS
fluctuation (e.g., Swerling Models I-IV, log-normal) were developed to allow
detection analysis to be performed without having actual RCS data versus
viewing angle.

1.7.2 Detection and False Alarm Probabilities

The probability of false alarm affects the number of “false target responses” that
must be evaluated by the radar processor. If a search volume such as that in Fig-
ure 1.10 consists of Nb antenna beams, where Nb = Nazimuth × Nelevation, and Nr
range gates, then the average number of false alarms per frame or search volume
is:

N FA = PFA ⋅ N b ⋅ N r . (1.34)

Typically, the search PFA is selected to result in ≤ NFA/second. Therefore, for ex-
ample, if NFA/second = 5 false alarms/second, and Nb = 100 and Nr = 1000, then
for a scan or frame time (TSC) of 2 seconds, the PFA allowed is:

N FA ⋅ TSC ( 5) ( 2 )
PFA = = = 10 −4 . (1.35)
Nb ⋅ Nr 10 5

Estimation of radar detection performance can be obtained analytically using

RCS models such as the Swerling models. Each model type has an associated
probability density (or family of probability densities) than can be integrated to
calculate PD:

fX ( x ) dx ,
∞
PD = ∫ VT
(1.36)
 Radar Fundamentals 21

Figure 1.10 Example of Search Volume

where fx (x) is the RCS fluctuation density, and VT is the selected detection
threshold, and:

VT2
∞ −
PFA = ∫ fn ( n ) dn = e 2σ 2 , (1.37)
VT

where the threshold can be calculated to be:

(
VT2 = − n Pfa 2σ 2 . ) (1.38)

Substituting (1.38) into (1.36) allows the calculation of PD for any analytical RCS
model.

1.7.3 Detection in Thermal Noise

Calculating PD for a specified PFA and target RCS model requires performing the
integration defined in (1.36). For exponentially distributed RCS (e.g., Swerling I),
the integral has a closed-form solution:

PD = (P )
FA
1 + SNR . (1.39)
22 Chapter 1

Equivalently, for a Swerling I target fluctuation (slow fluctuation—scan-to-scan),

the required SNR is given by:

n PFA
SNR REQ = − 1. (1.40)
n PD

For probability densities in (1.36) that do not have a closed-form solution, nu-
merical integration can be used to evaluate PD. PD has been evaluated for many
target models and is documented in texts such as references [6] and [7].

1.7.4 Constant False Alarm Rate Processors

The detection threshold defined in equation (1.38) is function of two parameters,

PFA and σ2. The first is specified as described in Section 7.2. The noise power, σ2,
is established by the receiver noise figure or system noise temperature (Ts) that
appears in the denominator of the expression for the RRE. Accurately estimating
the noise level, which is itself a random process, can be challenging depending
on the radar’s operating environment.
A constant false alarm rate (CFAR) processor is a method to estimate σ2 in a
localized manner around the range cell to be tested for detection. A typical
CFAR block diagram is illustrated in Figure 1.11. There are many variants of
CFAR algorithms. These are described in more detail in Chapter 3. A very com-
mon type is the cell-averaging (CA) shown in the figure. The basic concept is to
use two relatively short (in terms of the number of range cells) sliding windows
before (i.e., lagging) and after (i.e., leading) the cell-under-test (CUT) to be evalu-
ated as a candidate for detection to estimate the noise power.
Each of these sliding windows can provide an independent estimate of σ2, de-
2
noted σ̂ , with the estimation error variance being a function of the number of
cells employed in the estimate. These estimates can be based on simple arith-
metic averages, with and without censoring (i.e., removing large contributors
like adjacent targets or range sidelobes from consideration). Independent of the
precise formula used, the resulting estimate is used to compute:

VT 2 = − 2 σˆ 2 ln PFA . (1.41)
 Radar Fundamentals 23

Figure 1.11 Block Diagram of a CFAR Processor

Due to the variance on the total noise estimate, use of equation (1.41) in place
of equation (1.38) incurs a loss in detectability referred to as a CFAR loss. For this
reason, CFAR tends to be used where the additional signal processing loss (in-
cluded in the denominator of the RRE as a component of Lr) can be tolerated.
Therefore, it is usually used for tracking but not for functions such as long-range
search where the additional loss is undesirable.

1.7.5 Detection in Clutter

A target can be detected in the presence of clutter (unwanted RF backscatter) if

there is some distinguishing feature (or features) that allows separation of the
two “echoes.” Examples of features include:

• Doppler shift (range-rate difference)

• Polarization
24 Chapter 1

where:

2 ∆R
, (1.42)
∆ fd =
λ

and ∆Ṙ and λ are the difference between the target or clutter range-rates and the ra-
dar operating wavelength, respectively. A typical case is illustrated in Figure 1.12.
For example, land clutter is usually distinguishable from moving targets
since the mean velocity of the land reflections is zero. However, Doppler shifts
for rain in wind conditions or sea clutter at high sea state can overlap with
those of slow-moving targets (especially surface targets like ships or tanks). As
a benefit to target detection, certain RF waveform polarizations, for example,
circular, can result in reduced reflections from rain clutter as compared with
those of the target.
There are two basic approaches to mitigating clutter effects to better detect
moving targets:

• Moving Target Indicator (MTI) Cancellers

• Pulse-Doppler Processing.

MTI cancellers operate by subtracting sequential radar returns with the objec-
tive that near-stationary clutter will be cancelled, but moving targets will not.

Figure 1.12 Doppler Frequency Separation of Target from Land and Sea
Clutter
 Radar Fundamentals 25

The simplest MTI canceller is the single-delay (or two-pulse) canceller depicted
in Figure 1.13. The clutter appearing at the canceller output is reduced substan-
tially by the attenuation of low-frequency clutter. Optimum performance is ob-
tained when target Doppler shift is near PRF/2 (i.e., least attenuation of target
response).
Pulse-Doppler waveforms are coherent bursts that consist of N pulses spaced
by a uniform delay of PRI (pulse repetition interval). A pulse-Doppler pulse
train waveform is shown in Figure 1.14.
The matched filter to a Pulse-Doppler waveform consists of sub-pulse
matched filtering (i.e., range processing), with the results stored for M range
cells, followed by N Doppler filtering. Clutter (and its “alias”) will appear in
lower and higher Doppler filters. This is depicted in Figure 1.15.

1.8 RESOLUTION AND MEASUREMENT ACCURACY

Resolution is defined (for any measurement) as the target separation distance
necessary to identify when two targets are present. Waveform resolution is de-
fined as the inherent separation capability of the waveform in range and/or

Figure 1.13 Single-Delay MTI Canceller and Associated Frequency Response

Figure 1.14 Pulse Train Waveform Used for Pulse-Doppler Processing

26 Chapter 1

Figure 1.15 Pulse-Doppler Processing and Associated Frequency Response

Doppler. Angular resolution is defined as the inherent angle separation capabil-

ity of the sum antenna pattern.
Definitions of resolution include:

c
Range : δ R =
2B
Angle : δθ = θ3 , (1.43)
1
Doppler : δ f =
T

where:
τ = pulse length
B = pulse bandwidth
c = speed of light
θ3 = 3 dB antenna beamwidth
T = integration time.

Some convenient rules of thumb include the following:

• Two-target separation capability for equal amplitude targets at “reasonable”

signal-to-noise ratio, is approximately 2δ, where δ is the relevant inherent reso-
lution
• For unequal target amplitudes, with “reasonably high” matched-filter side-
lobes, the approximate resolution can be in the range: 2.5δ–3δ.

Measurement accuracy can be modeled as a function of the inherent resolu-

tions: δR, δθ, and δf as:
 Radar Fundamentals 27

δR
Range accuracy ≅ = σR
2 SNR
δθ
Angle accuracy ≅ = σ θ ; km ≈ 1.6 . (1.44)
km 2 SNR
δf
Doppler accuracy ≅ =σf
2 SNR

As can be seen, for this model accuracy improves in an inversely proportional

manner to SNR where SNR is the signal-to-noise ratio at the output of the ra-
dar matched filter. This model is based on the approximate statistical “Cramer-
Rao bound” to the estimation error standard deviation for each parameter.
The basic range resolution of an “un-coded” pulse of duration τ is given by δR
= cτ/2, where c = speed of light. In general, since large values of τ correspond to
higher SNRs, resolution for this waveform is inversely proportional to detect-
ability.
Pulse compression was developed to counteract this phenomenon. By “cod-
ing” the simple pulse (modulating the carrier over the pulse duration), good res-
olution can be achieved without sacrificing SNR by increasing the effective
bandwidth (B). Two popular approaches to achieving pulse compression are:

• Linear frequency modulation waveforms (LFM, also known as “chirp” wave-

forms)
• Discrete phase-coded waveforms (digital modulation).

These two techniques are illustrated in Figure 1.16.

Figure 1.16 Linear Frequency Modulation and Phase-Coded Waveforms

28 Chapter 1

1.9 TRACKING RADARS AND THE MONOPULSE TECHNIQUE

Tracking radars are typically used to obtain and maintain improved accuracy
compared to a search or surveillance radar. This capability is usually achieved
using a combination of hardware and software techniques, including:

• Monopulse antenna processing for angle measurements

• Wider RF bandwidth to improve range resolution and accuracy
• Pulse-Doppler waveforms to improve range-rate resolution and accuracy (and
for clutter mitigation)
• Signal processing techniques, such as range interpolation, to improve range
accuracy
• Tracking filters to improve accuracy of position and rate estimates and predic-
tions.

Tracking radars predict the future position of targets by estimating range and
angular rates. These predictions are used to position the antenna beam for sub-
sequent radar transmit and receive actions.
Monopulse is a technique to measure target angles by separating two-dimen-
sional receive antennas into azimuth and elevation “quadrants”; and combining
them in a specific way to enable the estimation of angular positions. An “error”
pattern is formed as e ( θ ) = ( θ ) ⁄ Σ ( θ ) as shown in Figure 1.17. The amplitude
and sign of e ( θ ) indicates the distance of the target from the antenna pointing
angle.

1.10 TRACK-WHILE-SCAN RADARS

Track-While-Scan (TWS) is an approach that combines the search and track func-
tions:

• While searching a volume, target acquisition (verification and track initiation)

and track maintenance are performed using detections
• No dedicated tracking beams are scheduled, but the “normal” search scan is
used for both search and track functions.

For mechanically scanned radars, like air-traffic control (ATC) radars:

 Radar Fundamentals 29

Figure 1.17 Illustration of Monopulse Antenna Processing

• A rotating antenna performs a periodic scan over 360 degrees of azimuth

• A broad elevation beam is typically used to achieve maximum elevation target
illumination.

For phased-array radars, similar search volumes (e.g., raster-scanned discrete

antenna beams) are serviced, again both for search and track purposes:

• Usually, the previous two or three scans of data are used to perform track initi-
ation (TI)
• Subsequently, tracks are maintained (or updated) on each new scan (re-visit)
of a particular angular region.

1.11 REFERENCES
[1] R. Nitzberg, Radar Signal Processing and Adaptive Systems, Artech House, 1999
[2] D. K. Barton, Modern Radar System Analysis, Artech House, 1988
[3] D. R. Wehner, High Resolution Radar, Artech House, 1987
30 Chapter 1

[4] Y. Bar-Shalom & X. Li, Multitarget-Multisensor Tracking, YBS, 1995

[5] S. Haykin & A. Steinhardt, Adaptive Radar Detection and Estimation, Wiley-Inter-
science, 1992
[6] H. Van Trees, Detection, Estimation and Modulation Theory, Part 1, Wiley-Interscience,
2001
[7] J. DiFranco & W. Rubin, Radar Detection, SciTech, 2004

1.12 PROBLEMS
1. Consider a radar with a peak power of 100 kW, 45 dB transmit antenna gain,
a 0 dBsm target, a 10 square meters antenna aperture, a target range of
500 km, with transmit and receive losses of 3 dB each. Estimate the received
power at the radar aperture using the relationship below.

t tσ Ar
PG
Prec = .
( 4 π R2 )
2
Lt Lr

2. Consider a radar being degraded by barrage noise jamming. For a radar re-
ceiver with system noise of –143 dBm, a jammer effective radiated power
(PJGJ) of 10 W, a 0.01 square meter effective radar antenna aperture (equiva-
lent to a –30 dB sidelobe), with the jammer at 1,000 km:

PJ GJ Ar'
N effective = kTs Br + .
4π RJ

How much degradation in equivalent noise power does the jammer pro-
duce? Using the radar from problem 1, estimate the effective signal-to-inter-
ference ratio (SIR). Is this good or bad for the radar’s performance?

3. Consider a radar with 2 false alarms per second, 1,000 range cells, 16 beams,
and a 10-second search frame time. What probability of false alarm does this
require? What if the search requires 32 beams and a 5-second frame time?

N FA ⋅ Τ
PFA = .
Nb ⋅ Nr
 Radar Fundamentals 31

4. Consider the probabilities of false alarm from problem 3. If the SNR

achieved by the radar is 15 dB, then calculate the corresponding probabili-
ties of detection assuming a Swerling I RCS fluctuation.

n Pfa
SNR REQ = − 1
1

PD = ( PFA ) 1 + SNR
n Pd

5. If a probability of detection of 0.95 is required for the system, what SNRs are
required to achieve these for the probabilities of false alarm calculated in
problem 3? Are these SNRs reasonable for a radar? Why or why not?
6. Consider a radar with range, angle (both azimuth and elevation), and Dop-
pler resolutions of 15 meters, 20 milliradians, and 100 Hz. If an SNR of 10 dB
is available, calculate the measurement accuracies that can be expected. If
the original target was a fighter aircraft with a 10 dBsm RCS, what accuracies
would be achieved for a missile with a 0 dBsm RCS and for a 747 aircraft
with a 25 dBsm RCS? Are these reasonable?

δR
Range accuracy ≅ = σR
2 SNR
θ3
Angle accuracy ≅ = σθ
km 2 SNR
δf
Doppler accuracy ≅ =σf
2 SNR

7. If an LFM or “chirp” waveform with a 10 µsec pulse length and a 1,000 MHz
bandwidth is used, what is the effective range resolution? What range accu-
racies would be achieved for the three target types in problem 5?