PRI Analysis and Deinterleaving

Richard G. Wiley, Ph.D.

Research Associates of Syracuse, Inc
111 Dart Circle
Rome, NY 13441
315-685-3135; [email protected]
1

Pulse Repetition Intervals (PRIs) are often the key to

identifying the signals of many radar systems. The first step is
to deinterleave signals from multiple radar systems. This
briefing is a a brief introduction to PRI analysis and
deinterleaving from the ELINT/EW point of view

PULSE REPETITION INTEVAL (PRI)

ELINT Implications of Range Equations and Radar Constraints

The effects of the one-way range equation of ELINT and the twoway range equation of radar on signal strength must be understood
and explored in order to appreciate the typical situations
encountered in ELINT and EW. Similarly, the constraints placed on
radar waveforms must be understood in order to correctly interpret
the functions and applications of the signals transmitted by radar
and also to be aware of the signal characteristics expected to be
encountered by ELINT. In many ways, understanding these aspects
of ELINT is what separates one who only observes signals from one
who both observes and analyzes signals.
Reference: ELINT, Chapter 2
4

Radar and ELINT Range Equations

PT GT G R O V
3
4
(4S ) R R LT LR
2

SR

PT GTE G E O2
2
2
(4S ) R E LT LE

SE
5

Ratio of ELINT Range to Radar Range

A significant aspect of these range equations is that the power
level transmitted by pulsed radar transmitters in order to detect
targets at long range is very high. This allows ELINT receivers to
detect radar signal at very long ranges even when observing the
sidelobes of the radars transmit antenna.
To simplify the discussion, suppose that the ELINT receiver
requires a signal level that is a factor G times the signal level
needed by the radar receiver, that is:

SE

RE
RR

G (S R )

4S 1 GTE G E LE
RR

G
V
G
G
L
T
R
R

1/ 2

1 10

4S
RR

G RG

100

Ma

RangeRatioSLi

m:
ea
b
in

1/ 2

GT

=3
E

RE/RR

ELINT Range/Radar Range

RE
RR

RE
RR

B
0d

RangeRatioMBi

e
lob
e
Sid

10

: GT

=0
E

RR 4S
G R G

1/ 2

dB

A
V  1 sq. m
G  30 dB
R

100

G  1
E

10

100
Ri
Range (km)

Figure 2-1 ELINT to Radar Range Ratio

7

2.2 Radar Constraints

ELINT signals of interest include radar signals of all types.
Sometimes, people concerned about ELINT attribute properties
to radar signals that are contrary to the constraints under which
radar systems must function. Avoiding this pitfall is an
important aspect of ELINT work. Understanding the
fundamental limitations faced by radar designers and the
associated ELINT implications is important. Consider this
statement: Radars of the future could transmit noise
waveforms over GHz bandwidths and be undetectable by
ELINT receivers. Should ELINT equipment be developed to
intercept and process this kind of signal? Probably not-because signals like this would not be useful for tracking or
search radars in military applications.
8

1 10

Range Resolution related to Bandwidth

Range resolution in radar is inversely proportional to the
bandwidth of the signal (assuming that it is processed
coherently). The fundamental relationship is:
c

'R

2B

Here c is the speed of light and B is the bandwidth of the signal

during the coherent processing interval; also called its
instantaneous bandwidth.
For example, to distinguish between two fighters in tight
formation 30m apart in range, BW must be about 5MHz. If one
postulates a value of B=1 GHz, the radar has a range resolution
of 15 cm. This means that the target echoes are resolvable in
15 cm range increments called range cells. The echoes from a
75m target are spread across 500 range cells.
9

RangeResolution
Resolution (meters)
(meters)
Range

1 10

100
RngRes bi

10

1
6
1 10

1 10
bi
Bandwidth
Bandwidth B(MHz)
(MHz)

1 10

Figure 2.2. Range resolution Related to Radar Coherent Bandwidth

10

This spreading of the echoes across a multiplicity of range cells reduces the apparent radar
cross-section (and thus reduces the SNR available) in a single range cell. For this reason,
radar designs generally have range resolution appropriate for their function. This leads to
choosing coherent bandwidths of 10 MHz or less. (10 MHz corresponds to range
resolution of 15 m.) In this sense, there is no such thing as a spread spectrum radar
what is transmitted is also received and the resulting range resolution is determined by the
bandwidth. What this means for ELINT is that the coherent bandwidth of radar signals is
likely to remain the same as it is now provided the radar performs the same task.

Range Resolution Required

Resolution (m)

Bandwidth (MHz)

30
60

5
2.5

2. Detect missile
separation at launch

15

10

3. Imaging of Ships,
Vehicles and Aircraft

.5-1

150-300

4. High Resolution
Mapping

0.15

1000

1.Count A/C in attack

formation

11

Moving Targets and Integration Time Constraints

If a radar is to detect targets moving in a radial direction (toward or
away from the radar), the amount of time the target will be present in a
given range cell is determined by the target velocity and the range
resolution. This limits the coherent integration time of present day
radars to
GR 'R

TCV

Here TCV is the maximum coherent integration time for a constant

velocity target with radial velocity v and R is the change in range
during that time. If the target is accelerating in the radial direction,
the maximum integration time is now a quadratic function of both
velocity and acceleration

T ACC

v  v  2a(GR)
a
2

12

0.5

v  v  2a('R)
a
2

0.5

Constraints on Time-Bandwidth Product or Pulse Compression

Ratio
Because range resolution is determined by bandwidth and integration
time is determined by velocity, there is a natural limit on the product
of the instantaneous bandwidth and the duration of the coherent
processing interval or pulse width. This is called the "timebandwidth product." The radar's pulse compression ratio is limited to
no more than its time bandwidth product. By combining Equations
for range resolution and integration time it is easy to see that the time
bandwidth product is limited to:

Bv
ac
c
o
BT 
1  2  1 
a o0
a
Bv
2v

13

BT Limit product BT
Maximum time-bandwidth

1 10

a=0 g

BT i  1
BT i  2

a=1 g

BT i  5

g
a=2

BT i  10

a=5

BT1 i

10
a=

Acceleration
0, 1,2, 5, 10 g's
Velocity=300m/s

1 10

5
4

1 10

1 10
bi
Signal Bandwidth
B (Hz)
Bandwidth

Figure 2-414
Limit on Time x Bandwidth

1 10

Constraints on Doppler Resolution

If the radar coherently integrates the echoes in one range cell for the
entire integration time, the minimum doppler filter bandwidth, Bf, is
approximately the reciprocal of the integration time,.T, which is
either TCV for constant velocity targets or TACC for accelerating
targets:.

1
Bf |
T

However if the target is accelerating, the doppler shift changes.

Clearly there is a relationship between acceleration and the time the
doppler shift of the moving target remains within the doppler filter
bandwidth.

'f acc

2aTf o
c

2aT

 Bf

15

Because the coherent integration time is approximately equal

to 1/Bf, substituting Bf=1/T into 2-12 gives the maximum
allowable coherent integration time and the minimum doppler
filter bandwidth as

O
2a
T
, Bf !
2a
O

16

1 10
6.502 u10

a=10g

Doppler Spread( kHz)

3
1 10

'f i  1
100

'f i  2

a=1g

'f i  5
10

'f i  10

1
0.65
0.1 3
1 10
1u10

0.01

0.1

3

Ti

Coherent Integration time T (s)

Figure 2.5 Doppler Spread and Maximum Signal Bandwidth

17

1 10

1000

1 10

Doppler Spread( kHz)

1 10

ple
Dop

'f i  1
'f i  2

100

ad
Sp r e

'f i  5
'f i  10

- ri gh

le
t sca

Ma
xi m
um

10

a=10g
a=5g
a=2g g
a=1

Bi

Sig
n

al B

10

and
w

0.65
0.1 3
1 10

0.01

100

idt
h

-le
f

t sc

ale

0.1

.001

Ti

Coherent Integration time T (s)

Figure 2.5 Doppler Spread and Maximum Signal Bandwidth

18

1
1

Bandwidth (MHz)

6.502u10

The doppler filter bandwidth must be no wider than the spread of

doppler frequencies expected. Figure 2-5 also shows the
maximum radar signal bandwidth. For the case where acceleration
has a minimal effect on the integration time, the maximum
acceleration of the target can be expressed in terms of the radar
signal's bandwidth as

a max

2
v
2B 2
c( RF )

19

Long integration times require small target acceleration. The

radar designer must choose a bandwidth that suits the range
resolution required and integration to suit the target motion
expected. Long integration time implies either slow targets with
little acceleration or else poor range resolution. High
acceleration targets require wider signal bandwidths. An
aircraft target approaching at 300m/s and maneuvering at 3 gs
needs a radar signal bandwidth of at least 2.5 MHz at 10 GHz.
Radar signals exhibit relatively constant characteristics during
coherent integration--important to know for ELINT analysis.
Tracking radars extend the coherent integration time when target
velocity and acceleration are known. Examining all possible
target velocities and accelerations requires huge processor
throughput and is generally not practical today.

20

Frequency Agility
From one coherent processing interval to the next, the radar can
change its carrier frequency without changing its range resolution
properties. The agility band is limited by the radar designers
ability to obtain sufficient power and to maintain beam width and
pointing angle--typically about 10% of the center frequency. (For
example, a 1 GHz agility band centered at 10 GHz.) What this
means for ELINT is that narrowband receivers have a low
probability of intercepting the complete radar transmission. If it is
sufficient to intercept only portions of the radar transmission,
narrowband receivers can be slowly tuned across the radar band
and the entire agility band can still be determined if the signals is
present for enough time. The coherent processing interval
determines the Doppler resolution. When FA is used with doppler
processing, the frequency is changed on a pulse-burst to pulseburst basis, not a pulse-to-pulse basis.
21

PRI Agility
Modern multifunction radar systems make use of multiple pulse
repetition intervals (PRI) values during one look at the target. It is
a requirement of todays pulse doppler radars that the PRI remain
constant during each coherent processing interval. For moving
target indicating (MTI) radar designs, there is usually a sequence
of PRI values that must be completed during one processing
interval. This repeated sequence is known as "stagger" and ELINT
analysts call the period of the stagger the stable sum. This is
because when consecutive PRIs are added, the sum is constant
when one adds together the PRIs which make up the stagger
period--regardless of which PRI is selected as the starting point for
the sum.
22

MTI radars operate by subtracting (in amplitude and phase) the

echoes from one PRI from those in the next PRI. Stationary targets
have the same phase and amplitude and thus cancel. Echoes
from moving targets generally do not have then same amplitude
and phase and so do not cancel. However if the target moves an
integer multiple of half wavelengths in one PRI, the phase of the
second echo is shifted by a multiple of 360 degrees from the first
and the echoes cancel. Such speeds are blind speeds. Changing
the PRI changes the blind speed. A PRI sequence is selected to
detect targets regardless speed Moving target detection (MTD)
radar systems use a doppler filter bank to divide the frequency
region between the PRF lines into several filter bands (for example:
8 bands). This requires repeated constant PRIs (say 10 pulses at
one PRI and then 10 pulses at another, etc.) Multiple PRIs are
required due to range and velocity ambiguities and make visible
target ranges and velocities eclipsed by transmitted pulses (in
23
time) or spectral lines (in frequency).

For constant PRI and RF, the maximum unambiguous range (Ru)
and the maximum unambiguous velocity (Vu) are given by:

Ru

c(PRI )
2

c
2( RF )( PRI )

Vu

Examples at 10 GHz:
PRI 1000 us, Vu=15 m/s and Ru=150 km
PRI 100 us, Vu=150 m/s and Ru=15 km
PRI 10 us, Vu=1500 m/s and Ru=1.5 km
As can be seen, the product of unambiguous range and velocity is a
constant. This means that the total ambiguity is fixed but changes
in PRI can increase the unambiguous range but decrease the
unambiguous velocity and vice versa.
c2

RuVu

24

4( RF )

1 10
6

10

Inverse relationship of unambiguous

range and unambiguous velocity at
common radar frequencies

Rui  1

Rui  31 105

22
5M

Rui  4
1 .3

Rui  5
Rui  6

10

1 10

Rui  7

35

Rui  8

1000

15
GH
z

GH
z
GH
z

5 .5

3G
Hz
GH
z

42
5M
GH
z

Hz

Hz

1 10

10
10

1 10

100

1 10

Vui  1  Vui  2  Vui  3  Vui  4  Vui  5  Vui  6  Vui  7  Vui  8

10

25
Unambiguous
Velocity (m/s)

Fi

2 7R

/V l it R l t d

Frequency
Agility Band

Frequency

Unambiguous Range (m)

Rui  2

(Depends
on
Component
Design,
ECM
Factors,
Designer
Ingenuity)

Coherent Processing Interval

(depends on radar mission)
Time
Determines Range Resolution Which
*BandwidthDepends
on Radar Mission

Figure 2-8. Modern frequency Agile Radar with 100% Duty Factor
26

USES OF PRI

27

UNAMBIGUOUS RANGE AND

VELOCITY DEPENDENCE
c
c

Analysis p. 147
28

RANGE-VELOCITY AMBIGUITY

Analysis p. 148
29

OPTIMUM PRI FOR MEDIUM PRF RADAR

30

Text p. 149

OPTIMUM PRI FOR MEDIUM PRF RADAR

Band Be Obscured at each PRF line

31

NOMINALLY CONSTANT PRI

32

PRI DRIFT

33

Analysis p. 153

CRYSTAL OSCILLATORS AND

COUNTDOWN CIRCUITS

34

Analysis pp. 191, 192

SCR-584

35

SEARCH RADAR PRI SELECTION

36

PRI STAGGER
Definition: Two or more discrete PRI intervals (elements) are alternating
in a periodic fashion.

Desired Parameters
- Number of intervals
- Number of positions
- Interval values
- Sequence
- Stable sum

T
T
T
Unmodulated Pulse Train
H

T- H

T+H

T+ H

Stagger Ratio

T- H

Typical Staggered Pulse Train

Two Interpulse Intervals Shown

Stagger Versus Jitter

37

RADARS WITH STAGGER

Radar

Pulse Width
(s)

Average PRI
(s)

Actual PRIs Stagger Mode

(s)

Stagger Ratio

Stagger Purpose

Radar Function

1.

6,18

100

2500
3500

5:7

To eliminate blind speeds

Surveillance

2.

3049

3032
3066

89:90

To eliminate blind speeds

Height Finder

3.

3000

2954.55
3045.45

0:97
(almost 100:103)

To eliminate blind speeds

Surveillance

4.

3000

2897
3103
613
1167

14:15

To eliminate blind speeds

Experimental surveillance

1000

5:7

5.

24

3000

2750
3250

11:13

To eliminate blind speeds

Surveillance

6.

1375

1250
1500

5:6

To eliminate blind speeds

Acquisition

7.

20

5247

5000
5494

0:91
(almost 10:11)

To eliminate blind speeds

Surveillance

8.

2777.9

2572.0
2777.8
2983.5

25:27:29

To eliminate blind speeds

Air route surveillance

9.

1.4, 4.2

1250

1240
1260

0.984
(almost 125:127)

To identify second-time-around pulses

Gap filter, surveillance and

interrogator

10.

2632-3226

Unknown
8-pulse stagger with three
programs

Unknown

To eliminate blind speeds

Air route surveillance

11.

42

1551.6

1408 (3)
1667 (3)
1460 (3)

Almost 1033:1225:1073

3 pulses at each interval for double

cancellation MTI to eliminate blind speeds

Detection; threat evaluation and

target designation (long range
mode given here)

12.

6.7

4000

3571.4 (3)
4405.1 (3)
3745.3 (3)
4255.3 (3)
4081.6 (3)

Exact order of 1 pulse

intervals is not known

3 pulse canceller for MTI. Stagger to

eliminate blind speeds

Surveillance

13.

1-100

400
62.1
2500

For first sequence only:

623.3
818.0
740.1
662.0
701.1

Various Sequences
16:21:19:17:20:18
16:17:16:17
16:19:16:19 38
16:21:16:21
16:17

To eliminate blind speeds. Has various

digital MTI processing including double
double-cancellation

Surveillance, tracking, kill

assessment, missile guidance

DESCRIPTION OF PRI VARIATIONS

Nature of Pulse-to-Pulse PRI Variations
Periodic

Discrete

Large

Type 1

Small

Random (non-periodic)

Continuous

Large

Small

Discrete

Large

Continuous

Small

Large

Small

(Large implies intentional, small implies incidental)

39

JITTERED PRI

Definition:

Pulse repetition intervals are intentionally varied on

interval-to-interval basis in a random or pseudorandom
fashion. The variations are usually more than one percent.

Intentional Jitter
- Discrete or continuous

Desired Measurements
- Mean PRI
- Peak PRI deviation limits
- PRI distribution (histogram)
- Number of discrete PRIs

40

RADARS WITH JITTER

PRI
(s)

Pulse
Width
(s)

Peak-to-Peak Jitter
(s)

6, 18

3000
1000

505

26

4629

92.6

200

4000

50

20
5
400

10204
10204
6666

999.9
918.4
653.3

0.9

416-1515
(Variable)

4-50
4-2.67

500-2777.7
3.3-4.0

Peak-to-Peak Jitter
(%)

1.7
5
20

3.75

83-303

9.8
9.0
9.8

20

60

2.2-12
None

Jitter Type

Jitter Purpose

Radar Function

Random

Anti-ECM and
interference

Sruveillance

Random

Anti-ECM and
interference

Target tracking

Random

Anti-ECM and
interference

Long-range
surveillance

Random
Or
Programmed

Anti-ECM. Results from

PRF being submultiple
of RF which is jumping

Decoy discriminator
target tracking
acquisition

Unknown

Unknown

High resolution
synthetic aperture
mapping

Random

To reduced inward
range gate stealers, antiinterference, reduce
second-time around
echoes

Multifunction

41

PRI DWELL/SWITCH PULSE DOPPLER

Definition:

Rapid (automatic) switching between discrete PRIs with a dwell at each PRI

PRI = T1

PRI = T2

Dwell Time 1

Dwell Time 2

Desired measurements
- Number of PRIs
- Value of PRIs
- Dwell times
- Total dwell time for sequence
- Dwell sequence
- Time to switch
42

SLIDING PRI
Definition: The pulse train has a PRI (PGRI) that is continuously changing in either
a monotonically increasing or decreasing manner between maximum
and minimum PRI limits.

Desired Parameters
- PRI limits (min and max)
- Sweep waveform
- Sweep time (limits)

43

OTHER PRI TYPES 1

Periodic Modulation
Definition:

Pulse train consists of discrete or continuous intervals that

periodically increase and decrease, e.g., with sinusoidal,
sawtooth or triangular waveform
- Modulating waveform and rate
- Mean PRI and peak deviation limits

Pulse Interval Displacement

Definition:

Insertion of a different pulse interval into an otherwise

periodic pulse train
- Displacement value

44

OTHER PRI TYPES 2

Interrupted Pulse Train

Definition:

Intentional interruption of the pulse train with no apparent periodicity

- Range of on-period
- range of off-period

Burst Pulse Train

Definition:

Pulse train that is transmitted for some purpose for a relatively short
time and then is off for a relatively long time
- Burst definition
- Number of bursts per second
- Relationships of burst to scan

45

SCHEDULED PRIs

Scheduled PRIs
Definition:

PRIs are computer controlled, vary with the target environment and
function being performed by radar, and cannot be described by other
definitions
- Number of intervals
- Interval values
- Typical sequences
- Reason for sequence

46

MUTLIPLE PULSE GROUPS

Constant and Cyclic Patterns

Definition:

Pulse group characteristics remain constant or vary cylically in

predictable manner
- Number of pulses in group
- Pulse intervals
- Group position data

Frames/formatted pulse trains (data encoded format)

Definition:

Pulse train includes marker and data pulses

47

SUMMARY OF PRI TYPES

Analysis p. 151
48

DOPPLER EFFECT
v = radial velocity
c = 3(108) m/sec
fo = transmitted RF

v
km/hr
FIGURE 3-1. DOPPLER EFFECT

f
1

fo x

100

cv
2v

# f o x 1 
c-v
c

Doppler Shift

f  fo
1

2v
x fo
c

49

50

Doppler Shift (Hz)

@ 3 GHz
@ 10 GHz
555.5

1851.8

1000

5,555

18518.5

2000

11,111

37,037.0

FOURIER TRANSFORMS

51

IDEAL VS. ACTUAL SPECTRA

FOR CW SIGNAL

52

FM THEORY
V(t)

A sin(2 f c t  (t))
Phase Disturbance
2 f c t  (t)

Total Phase

1 d
(total phase)
2 dt

Instantaneous Freq.
ASSUME

(t)
1 d
2 dt

fc 

1 d
2 dt

sin2f m t
f m cos2f m t

Let

F/f m , then

1 d
2 dt

Fcos2f m t

THEN :
V(t)

Asin(2 f c t 

f
sin2 f m t)
fm

INDEX OF
53
MODULATION
" M"

BESSEL EXPANSION
V(t)
A

Jo (m) sin c t  J (m)>sin(c  m )t  sin(c  m )t @

1
 J (m)>sin(c  2m )t  sin(c  2m t)@
2
 J (m) . . . . .
3

J0(m)

J1(m)

J1(m)

J2(m)

J2(m)

fc-2fm

fc

fc+2fm

fc-fm

fc+fm
54

BESSEL FUNCTIONS

55

MOD. INDEX LESS THAN 1

FOR COHERENT SIGNALS:

f
 1
fm

J o (m) | 1
J (m) | 0
2

i.e. f is small
m
J (m) |
1
2
J (m) | 0 ....... etc.
3

THEREFORE
V(t) # A[sinc t 

m
m
sin(c  m )t  sin(c  m )t]
2
2

V
SB
Vc

m
2

f
2f m

mA/2

mA/2

in dB

fc-fm

fc

V
20 log SB
Vc

fc+fm

56

20 log

f
2f m

EXAMPLES
20 log f
2f m
f
2f m

 40 dB

1 kHz

e.g. f m 1 kHz
f

20 Hz

IF f c 10 GHz, STABILITY IS
20 Hz
10 x 109 Hz

2 parts in 109

(AT 1 kHz RATE)

57

RANGE AMBIGUITY RESOLUTION VIA MULTIPLE PRIs

12 s = X

T1 = 40 s

2 s = Y
T2 = 30 s

Actual Round Trip Echo Time is T = 92 s

N1 T1 + X = T
Trial and Error
Solution

and

N2 T2 + Y = T

N1

N2

N1 T1 + X

N2 T2 + Y

1
1
2
2

1
2
2
3

52
52
92
92

32
62
62
92

Unambiguous Range

c x Least Common Multiple of T , T

1 2
2
c x >120 s@
2
58

Analysis p. 196

ERICSSON PS-05/A MULTIMISSION RADAR

59

ERICSSON PS-05/A MULTI-MODE OPERATION (1)

60

ERICSSON PS-05/A MULTI-MODE OPERATION (2)

61

MTI VIDEO

62

MTI PHASE SHIFTS

63

MTI BLOCK DIAGRAM

64

BIPOLAR VIDEO

65

DOPPLER RETURNS
TRAIN

CAR
Typical images displayed on TPS-25 ground
Surveillance radar. Shown are target images
of: 1) a train, 2) an automobile, 3) a walking man,
and 4) a walking girl. (US Army photograph.)

MAN WALKING
WOMAN
WALKING
66

PULSED-OSCILLATOR MTI

= 2E sin(SfdT) cos [2Sfd(t + T/2) + Io]

Zeros at 0, , n when
f

n
T

so blind speeds are V

b

cxn
2RF x T
n
x PRF
2

Barton, p. 192
67

Page M50.ppt

68

BLIND SPEED ELIMINATION

1
T

6
T

No Stagger

vb = n c/2(PRI)(RF)

T
T

Vbn = Vb (7 + 5)/2

5
7

Deep lobe
at 32/T

T
T

63
65

Null at
64/T

Ref:69Barton, page 222

IMPROVEMENT FACTOR OF CANCELLER

(S/C)out
(S/C)
in

signal  to  clutter ratio at output of canceller

signal  to  clutter ratio at input of canceller

Overall improvement factor I is found from:

1
I

1  1  1  ....
I
I
I
3
1 2

I1, I2, I3 are the individual improvement factors calculated on basis of PRI, pulse
amplitude, pulsewidth, transmitter frequency, .. stabilities

70

INSTABILITY LIMITATIONS

71

CLUTTER STRENGTH

72

MTI + PULSE DOPPLER = MTD

Weighting
And
Magnitude

8-Pulse
Doppler
Filter Bank

3-Pulse
Canceller

I,Q Data
From A/D
Converters

Threshold

Zero
Doppler
Filter

Clutter Map
(Recursive
Filter)

Magnitude
(I2 + Q2)1/2

Typical Applications
New FAA ASR radars (10 pulse dwell)
AN/SPS-49 USN-adjunct to AEGIS (6-pulse dwell)
RAMP (Canada)

Clutter
Memory

15 20 radar scans are

needed to establish
the clutter map

73

MTD PERFORMANCE
Theoretical
RMS Clutter Width
Processor

0.01 PRF

0.1 PRF

MTI Improvement
Factor

1 canceller
2 cancellers
3 cancellers

25 dB
50 dB
72 dB

8 dB
12 dB
16 dB

FFT Improvement
Factor

8 pulses

35 dB

22 dB

MTI + FFT
Improvement Factor

1 canceller +
8 pulse FFT
2 cancellers +
8 pulse FFT
3 cancellers +

60 dB

28 dB

80 dB

34 dB

100 dB

36 dB

(Reference: NRL Report 7533, G.A. Andrews, Jr.)

Practical
Performance of FAA ASR radar: 3 pulse MTI alone | 25 dB
3 pulse MTI + 8 pulse FFT | 45 dB
(Reference: Skolnik, Introduction to Radar Systems, 1980, p. 127-128)
74

Target
Detection

ELINT IMPLICATIONS OF MTD

Coherent carrier

RF stability is necessary

Constant PRIs
Constant RF
(for a certain
number of pulses)

Several PRIs of the same interval must be

transmitted at the same RF (typically 4,
8, or 16 pulses for the FFT plus pulses
to fill the canceller. For example, a
three-pulse canceller plus an eight-pulse
FFT requires 10 pulses).

Stagger to eliminate
blind speeds

For these radars, the pulse interval

stagger occurs not from pulse-to-pulse but
from pulse group-to-pulse group

Long PRI

MDT is generally used for long-range radars

where the low PRF creates very ambiguous
Doppler shifts.

75

PRI EXERCISES

1.

The analyst found a signal at 6 GHz which had two-interval, two-position stagger. The
intervals were 500 and 550 microseconds. What is the average PRI? What is the
stagger ratio? What is H? What are the new blind speeds?

2.

What is the improvement factor for MTI of a radar which has RMS jitter of 10 nanosec
and a pulse duration of 1.41 microsec?

3.

A discrete random jitter PRI train was analyzed and the PRIs were found to be one of
the following 5 nominal values:
Nom
PRI (sec)
2440.8
2428.7
2465.3
2453.1
2562.9
Is there a clock? If so, what countdowns are used and what is the clock frequency or
period? What common range mark is that closest to?
(This problem is discussed on p. 194-195 of analysis book.)

76

PRI EXERCISES #2 - ANSWERS

1.

(500 + 550)/2 = 525 microsec = average PRI

R = 550/500 = 1.1 (11:10)
H = 550 525 = 25 microsec
Blind speed before stagger = nc/(2 PRIave RF)
(3 x 108 ) m / sec
VB
 171.4 km / hr (106.5 mph )
2(525)(10 6 ) sec x 6(109 ) x 1 / sec
V/VB = (11 + 10/2 = 10.5)
V = (10.5) (171.4) = 1800 km/hr (1118.4 mph)

2.

Improvement factor due to PRI instability is:

IdB = 20 log [ W / 2 't BW )], 'W = jitter, W = pulse duration,
2
B = bandwidth
IdB = 20 log [1.41 (10-6) sec/ 2 10(10-9)sec)]
= 20 log [102] = 40 dB

3.
Period
2440.8
2428.7
2465.3
2453.1
2562.9

Periods
In Order
2482.7
2440.8
2453.1
2465.3
2562.9

Difference
12.1
12.3
12.2
97.6

Nearest
Countdown
199
200
201
202
210

Calculated
Clock Period
12.20452
12.20400
12.20447
12.20445
12.20428
12.204392 average

The differences 12.1, 12.3, 12.2 average 12.2

97.6 divided by 12.1 = 8
So use 12.2 to start for countdowns.
The average clock period is 12.204392 sec so reciprocal is 81.93777 kHz (2000 yards, see p. 192.)

77

NOISE EFFECT ON PRI

'T

Triggering Error

T
RISE x 'A
0.8
A

'A
'T

Noise

'A
'T

TRISE/0.8

78

Slope

A
(TRISE / 0.8)

PRI VARIATION DUE TO NOISE

V2
amplitude Noise Power
1
(Amplitude)2 Signal Power SNR
T
Rise 1
V
Time
0.8 SNR
V2
V2
 V2
2 V2
PRI
Time1
Time 2
Time
T
2 Rise
V
PRI 0.8 SNR

79

BANDWIDTH EFFECT ON SNR

t
SNR 3.125 V r

PRI

tr |

.35
Bandwidth

SNR Required for

Bandwidth (MHz)

Rise Time
Limit (ns)

1 ns Jitter

10 ns Jitter

100 ns
Jitter

0.1

3.5 Ps

81 dB

61 dB

41 dB

1.0

0.35 Ps

61 dB

41 dB

21 dB

10.0

35 ns

41 dB

21 dB

100.0

3.5 ns

21 dB

80

AMPLITUDE INDUCED ERROR

81

AMPLITUDE COMPENSATED TRIGGER

82

PERFORMANCE OF TRIGGER CIRCUIT

83

DOPPLER SHIFT OF PRI

In 1 PRI, the platform moves

VR PRI

Transmit time from transmitter to receiver changes by VR PRI/c

Example: VR = 600 M/S PRI = 3000 s

3
r 6 ns
Observed 'PRI = 600 x 3 x 10
3 x 108

84

DELAY AND PULSE JITTER

Delay D2
Delay D1

Peak-to-Peak Jitter
At Delay D1

Peak-to-Peak Jitter
At Delay D2
85

DELAYED SWEEP JITTER PHOTOS

~ 1 s Jitter
Delay = 1 PRI

~ 2 s Jitter
Delay = 5 PRI
86

SYNTHESIS OF AVERAGE PRI

87

PRI DRIFT MEASUREMENT

88

REAL TIME RASTER DISPLAYS

89

DUAL AMPLITUDE AND TIME

DELAYS

90

Analysis p.74

DTE MODE-CIRCULAR SCAN

RADAR

91

RTR SIMULATION ON A
PERSONAL COMPUTER

92

MEAN PRI ESTIMATES

93

MINIMIZING THE SQUARED ERROR

94

RMS ERRORS COMPARED

95

96

97

NONCUMULATIVE AND
CUMULATIVE JITTER

98

CRAMER-RAO BOUNDS COMPARED

99

PRI ESTIMATION PERFORMANCE

100

USING THE WRONG JITTER

MODEL

101

PRI HISTOGRAMS

102

ACTIVITY IN 0.1S INTERVALS

103

INTERVALS FORMED BY
PULSE PAIRS

104

DELTA-T HISTORGRAM
(10% JITTER)

105

DELTA-T HISTOGRAM-STAGGER

t0 = 0

1 2

5
t1 = 4

t2 = 9

t3 = 16

A.

(tn tn-1) = 4, 5 or 7

B.

(tn tn-2) = 9, 11 or 12

C.

(tn tn-3) = 16

D.

(tn tn-4) = 20, 21 or 23

E.

(tn tn-5) = 25, 27, or 2

F.

(tn tn-6) = 32

5
t4 = 20

7
t5 = 25

4
t6 = 27

(4 + 5, 4 + 7, 5 + 7)

(4 + 5 + 7)

6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

D
106

5
t7 = 31

t7 = 37

THREE POSITION STAGGER

107

DELTA-T HISTOGRAM:
TOA AUTOCORRELATION
n
G (t  t n )
n 1

f (t )

t1

.....
t2 . . . . .

h (W)

h (W)

f
G (t  t n ) G (t  t k  W) dt
f n
k

value only if t  t n
tn  tk  W 0

EXAMPLE
t1

t=0

t3

f
f (t ) f t  W dt
f

0 and
OR t n

t = t2

tk  W

t3

W
t = t1 + W

t4 . . . . .

108

t4

DELTA-T HISTOGRAM:
TOA AUTOCORRELATION
h(W) G(t n  t  W)
k
n k

2
2
h(W) G(t n  t k  W)
n k

1
1
A count of the number of
pulse pairs such that
W d tn  t W
1
k 2

THEREFORE:
A count of the number of pairs of pulses whose arrival
times differ by a value between W1 and W2 is equal to
the integral of the autocorrelation of the TOAs
109

JITTER ANALYSIS MODEL

Center Frequency
(average PRF)
Jitter
Waveform
Peak
Amplitude

FM
Oscillator

Trigger
Generator

Periodicities
Periods
Amplitudes

Drifts/Trends
Slopes
Random Components
Bandwidths
Variances
Probability Densities
110

Time of
Arrival
Sequence

INSTANTANEOUS
FREQUENCY ESTIMATION
500

700

600

500

400

2500
Freq

2000

1428.5

1666.7

2000

PRIs (s)

500

2000

Linear
Interpolation

Midpoints
of Intervals
111

DEINTERLEAVING DEVICE

112

DEINTERLEAVING VIA DELTA-W HISTOGRAM

113

PURE VS. IMPURE INTERVALS

114

NUMBER OF EMITTERS DEINTERLEAVED

115

COMPLEX DELTA-W HISTORGAM - I

116

COMPLEX DELTA-W HISTOGRAM - II

117

COMPARISON OF DELTA-W HISTOGRAMS

118

EFFECT OF A NEAR MULTIPLE PRI

119

EFFECTS OF JITTER ON DELTA-W

HISTOGRAMS

120

Delta-T Histogram for Ten Interleaved Pulse Trains

Delta-T Histogram

Histogram Count

100

dhist b
.75 max( dhist )

50

0
5
8 10

1 10

1.2 10

1.4 10

int vb  PRI k 10

1.6 10

6

PRI, Seconds
N

820

10 Interleaved Pulse Trains

121

Comparison of the Delta-T and Complex Delta-T Histograms

Comparing Delta-T Histograms
100
100
abchist b

dhist b
0

1.05 max( abchist )

1.05 max( dhist )

50
100
0

1 10

2 10

int vb  PRI k 10

820

3 10
6

4 10

  int v  PRI 10
b

PRI, Seconds

6

10 Interleaved Pulse Trains

Top Trace is the regular Delta-T Histogram;

Bottom Trace is the Complex Delta-T Histogram--Note how multiples of the PRIs are suppressed
The dots above the peaks indicate the true PRI values
122

Delta-T Hisotgram bin Count

Complex Histogram Absolute Value

150

Effect of Jitter on Delta-T Histograms

(Jitter=1 microsecond)
Comparing Delta-T Histograms

abchist b
1.05 max( abchist)

dhist b
0

50

1.05 max( dhist )

50

0
5
5 10

1 10

1.5 10

2 10

2.5 10

3 10

3.5 10

100

  intv  PRI 10

6

intv b  PRI k 10

6

PRI, Seconds
Jitnc

0.5

Jitcum

0.5

7

5 u 10

820 width

10 Interleaved Pulse Trains

123

Effect of Jitter on Delta-T Histograms

(Jitter=2 microseconds)
Comparing Delta-T Histograms
100

100

50
dhist b

abchist b
1.05 max( abchist)

50

1.05 max( dhist )

50

0
5
5 10

1 10

1.5 10

2 10

intv b  PRI k 10

2.5 10

  intv  PRI 10

6

3 10
k

3.5 10

100

6

PRI, Seconds
Jitnc

Jitcum

820 width
124

7

5 u 10

10 Interleaved Pulse Trains

Delta-T Hisotgram bin Count

Complex Histogram Absolute Value

Complex Histogram Absolute Value

50

Delta-T Hisotgram bin Count

100

100

Effect of Jitter on Delta-T Histograms (Jitter=5 microseconds)

Comparing Delta-T Histograms
100

50
dhist b

abchist b
1.05 max( abchist)

50

1.05 max( dhist )

50

0
5
5 10

1 10

1.5 10

2 10

intv b  PRI k 10

2.5 10
6

  intv  PRI 10
b

3 10
k

3.5 10

Delta-T Hisotgram bin Count

Complex Histogram Absolute Value

100

100

6

PRI, Seconds
Jitnc

2.5

Jitcum

2.5

820 width

7

5 u 10

10 Interleaved Pulse Trains

125

Complex Delta-T histogram: Original and Improved

Original Complex Delta-T Histogram

Improved Complex Delta-T Histogram

Uniform Jitter=0.002

Uniform Jitter=0.02

Shift time origin

To avoid excessive
Phase variation
Uniform Jitter=0.2

126

K Nishiguchi and M. Korbyashi,

"Improved Algorithm for
estimating Pulse Repetition
Intervals, IEEE Transactions on
Aerospace and Electronic Systems,
Vol. 36, No. 2, April 2000.

Example of Automated Peak Processing Results

Delta-T Hist.
Complex Delta-T
Input PRI Values
0

pk

0
1
2

1 10-4
1. 048 10-4
1. 1110-4

0
1
2

1 10-4
1. 0510-4
1. 1110-4

3
4
5
6

1. 1510-4
1. 162 10-4
1. 176 10-4
1. 1910-4

3
4
5
6

1. 1510-4
1. 164 10-4
1. 178 10-4
1. 192 10-4

7
8
9
10
11

1. 2110-4
1. 2310-4
1. 2610-4
0
0

7
8
9
10
11

1. 2110-4
1. 2310-4
1. 2610-4
0
0

12
13
14
15

0
0
0
0

12
13
14
15

0
0
0
0

pkc

6

PRI 10

0
1

1 10-4
1. 0510-4

2
3
4
5
6

1. 1110-4
1. 1510-4
1. 163 10-4
1. 177 10-4
1. 191 10-4

7
8
9

1. 2110-4
1. 2310-4
1. 2610-4

This example based on the method of B.

Frankpitt, J. Baras, A. Tse, "A New Approach
to Deinterleaving for Radar Intercept
Receivers," Proceedings of the SPIE, Vol
5077, 2003, pages 175-186

Jitter =10 ns cumulative and 10 ns non-cumulative

Histogram Bin size 200 ns.127
Pulse Train Spectrum of Ten Interleaved Pulse Trains
k

Amplitude

PRF Spectrum

0.01

Xj

0.00011 max( X)
0.005

0
6000

8000

1 10

1.2 10

1.4 10

f j  PRF k

8.705 u 10

PRF (Hz)
10 Interleaved pulse Trains

PRF Resolution 10 Hz
128

1.6 10

1.8 10

2 10

This plot is the FFT of

TOA
phase 2S (
)
T
R. Orsi, J. Moore and R. Mahony, "Interleaved
Pulse Train Spectrum Estimation," International
Symposium on Signal Processing and its
applications, ISSPA, Gold Coast, Australia,
August 25-30, 1996

PRF Spectrum

Amplitude

0.03

Xj

.025 0.02
.015
0.01

0
4000

6000

8000

1 10

1.2 10

1.4 10

1.6 10

1.8 10

2 10

f j 1  PRF k  2 PRF k

PRF (Hz)
10 Interleaved pulse Trains

1.741 u 10

Fewer Pulses--Degraded PRF Resolution (50 Hz)

129

Figure 13.10 Pulse Train Spectrum for a Shorter Record

k

PRF Spectrum

Amplitude

0.03

Xj

.03


0.02

.02
0.01

0
4000

6000

8000

1 10

1.2 10

1.4 10

1.6 10

f j 1  PRF k  2 PRF k

PRF (Hz)
10 Interleaved pulse Trains

871

Fewer Pulses--Degraded PRF Resolution (100 Hz)

130

1.8 10

2 10

PULSE SORTING ALGORITHM

C

C
B

A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A

A A

3 Adjacent Matching Intervals

Step 1. Find 3 adjacent matching intervals

Step 2. Extend in both directions to discover other numbers of the pulse train
Step 3. Remove this pulse train and go back to Step 1.
If no more pulses can be removed, go to Step 4.
Step 4. Consider all pairs of pulses to search for intervals which match; go to Step 2.

131

SORTER SOFTWARE PERFORMANCE

Score

Amp On

(Pulses Pr ocessed)  10 (Pulses Wrong)

Total Pulses  Noise Pulses
Amp On: 0.2 amp Tolerance
from pulse-to-pulse

0% Jitter

100
90

Amp Off

80

1% Jitter

ff
pO
m
A

70

Score

pO
Am

60
50

pO
Am

Simulated Data
Average Density 200 pps

40

Amp Off

30
20
8% Jitter

10
0

10s

100s
132
Time Tolerance

1000s

SIMULATION SCENARIOS
File Name

I.P. #

PRI

Variation

C2-3009-V05

2
1

30.0
70.0

0.5
0.5

C2-3009-V20

2
1

30.0
70.0

2.0
2.0

C3-3009-V05

1
3

3.0
120.0

0.5
0.5

C3-3009-V20

1
3

3.0
120.0

2.0
2.0

C4-3009-V05

50.0

0.5

C5-3009-V05

1
2
3

100.0
100.0
100.0

0.5
0.5
0.5

C5-3009-V20

1
2
3

100.0
100.0
100.0

2.0
2.0
2.0

Denotes initial pulse number.

Table 1. Simulation scenarios.
Ref: Kofler and Leondes
133

FIXED GATE DEINTERLEAVING

RESULTS
File Name

I.P.D.

PRI

% Misses

C2-3009-V05

20
28
30
35
63
69

90.0
105.0
60.0
69.9
180.1
90.0

90.9
37.5
50.0
81.8
0.0
0.0

C2-3009-V20

20
28
30
35

90.3
104.9
59.8
69.7

81.8
37.5
42.9
63.6

16
.
.
.
473

9.1
.
.
.
36.0

62.9
.
.
.
100.0

7
117
395

3.7
50.0
32.8

3.2
77.3
60.0

C3-3009-V05

C3-3009-V20
C4-3009-V05

No emitters detected

C5-3009-V05

19
30
47
48
56

100.0
100.2
199.9
99.9
200.0

16.7
92.9
25.0
75.0
0.0

C5-3009-V20

19
30
47
48
56

99.9
100.7
199.7
99.7
200.0

16.7
92.9
25.0
75.0
0.0

134

Ref: Kofler and Leondes

ADAPTIVE GATE DEINTERLEAVING

RESULTS
File Name

I.P.D.

PRI

% Misses

C2-3009-V05

7
18

30.0
70.0

0.0
0.0

C2-3009-V20

7
18

30.0
70.0

0.0
0.0

C3-3009-V05

5
168

3.0
120.0

0.0
0.0

C3-3009-V20

5
168

3.0
119.9

0.0
12.5

C4-3009-V05

50.0

0.0

C5-3009-V05

15
20
25

100.0
100.0
100.0

0.0
0.0
0.0

C5-3009-V20

15
20
25

100.0
100.0
100.0

0.0
0.0
0.0
Kofler and Leondes

135

d,W

/WZ/

136

W

137

WdW
WZ/W

138

PRI ANALYSIS EXERCISE

Two signals are observed with the same angle of arrival but on different frequencies. The PRI of one is nearly
stable at 3000 s. The PRI of the second jitters randomly with a mean value of 1500 s and a peak-to-peak jitter
of about 20 s. The analyst notices that the PRIs of the second signal can be paired such that their sum is nearly
stable at 3000 s; i.e., PRI #1 + PRI #2 = PRI #3 + PRI #4 = PRI #5 + PRI #6, etc. However, PRI #2 + PRI #3 z
PRI #4 + PRI #5. He also notices that the mean value of the second signals PRI is exactly one-half that of the
first signals PRI every time the two signals are reported. The first signal has a slow circular scan, the second a
faster sector scan. What conclusions might be drawn about these two radars?
What additional data would you request from the ELINT station?

139

PRI EXERCISE ANSWER

There is a good possibility that the second radar operates in PRI synchronism with
the first;
but at one-half the PRI. Alternate pulses are triggered by the master clock, the
intermediate
pulses are generated by one shot type delay circuit which is not stable.
The second radar may be a height finder using elevation sector scan and associated
with a long
range search radar.
Confirmation of this would be aided by using two receivers and making a recording
of both
Signals simultaneously to investigate whether the second signal is synchronized to the
first.

140

PRECISION PDWs
Pulse Descriptor Words are computed from pre-detection
burst recordings
Digitizer has detected presence of high SNR pulses,
and captured them
Different capture and processing techniques apply to low
SNR pulses
Standard PDWs computed are:
- Amplitude
- Frequency
- Time of Arrival - Bandwidth
- Pulse Width
Algorithms and accuracies are described

141

Condor Systems, Inc.

USEFULNESS OF PRECISION PDWs

Reveals fine details of pulse train jitter patterns
Permits very high accuracy computation of crystal
controlled PRIs with few pulses
Can use very accurate pulse width to sort pulses
Fine variations of frequency pulse to pulse reveal unique
emitter characteristics (e.g., frequency pulling effects due
to VSWR changes in antenna rotary joint, etc.)
Amplitude droop in transponder pulse groups
Precise antenna pattern scan envelope measurement
142

Condor Systems, Inc.

EXAMPLE OF PRE-DETECTION RADAR

PULSE RECORDING

143

Condor Systems, Inc.

CALCULATION OF AMPLITUDE, TOA,

PW

144

Condor Systems, Inc.

TOA MEASUREMENT ACCURACIES

Digitizer time base determines ultimate accuracy

Individual pulse time of arrival error determined by:

't

tr
2SNR

where 't RMS Error in TDOA

t r Pulse Rise Time
SNR Signal to Noise Ratio in Captured
Pulse Bandwidth
Example: 30 ns rise time, 37 dB SNR yields RMS error of
300 picoseconds per pulse

145

Condor Systems, Inc.

PULSE WIDTH MEASUREMENT

ACCURACY

'pw

2
't 2
r  'tf

where 'pw RMS error in pulse width

't r RMS error of pulse risin g edge time
RMS error of pulse falling edge time
't
f

Example: RMS errors of captured pulse edge times of

300 picoseconds yield 1.414 x 300 = 423
picoseconds RMS pulse width error per pulse.

Condor Systems, Inc.

146

EXAMPLE OF PULSE WIDTH

ACCURACY

147

Condor Systems, Inc.

PULSE FREQUENCY COMPUTATION

148

Condor Systems, Inc.

PULSE FREQUENCY ACCURACY

1
T SNR TW
in

'f

Technique applies to high SNR cases (>+15 dB), sine

wave pulse
where

'f

RMS frequency accuracy

Integration time (~ pulse width)

SNR
Input Signal  to  Noise Ratio in BW , W
in

Input Pr e  det ection bandwidth

Example: 1 microsec pulse, 30 dB SNR, 20 MHz Bandwidth

yields RMS accuracy of 7 kHz.
Condor Systems, Inc.
149

EXAMPLE OF PULSE FREQUENCY

COMPUTATION

150

Condor Systems, Inc.

Pulse Bandwidth

151

Condor Systems, Inc.

EXAMPLE OF PULSE FREQUENCY

COMPUTATION

152

Condor Systems, Inc.