A Tutorial on Radar System Engineering

Dr LEE Kar HengChief TBSS Group

Ph.D, M.Eng, M.Sc, B.Tech(Hons), MIEEECertified Teacher in Higher Education SEDA & TP

RADAR SYSTEM ENGINEERING

IEEE International Conference on Advanced Telecommunications Conference, Ho Chi Minh City,

Vietnam, 2015

2

• The speaker is thankful to TERMA for the use of TERMA Radar technical information in this tutorial

• Terma has delivered 7 radars in Vietnam– 2 Surface Movement Radars: Tan Son Nhat

International Airport and Noi Bai International Airport

– 3 Vessel Traffic Radars: Port of Ho Chi Minh City– 2 Vessel Traffic Radars: Port of Hai Phong

Acknowledgement

3

Tutorial Objective• To understand the entire functionalities of radar

systems: its components and operations

4

Tutorial Outline

Scan PatternGener ator

Antenna Duplexer

WaveformGener atorTransm itter

Receiver S ignalPr ocessor

DataExtractor

DataPr ocessor

RadarDisplay

• A Typical Radar System

5

Quick Review• RADAR- Radio Detection and Ranging• Theory of reflection, absorption and scattering• Higher frequency gives better result (???)• Information from radar: Range, height, direction,

direction of motion, relative velocity

6

• Target distance is calculated from the total time (tdelay) taken by the pulse to travel to the target and back to the radar

• c = 3 x 108 m/s, speed of light

Quick Review

tdelay

7

Quick Review

Video

Noise Noise & clutter Noise & clutter

tdelay

8

Quick ReviewNoise (not necessarily visible)

Targets (land)

Targets (vessels)

Sea Clutter

Rain

VRM 63 km(34 NM)

A weather radar will see rain clutter as information.

A navigation radar will see rain clutter as noise.

9

Quick Review

10

• Block diagram of a typical pulsed radar

A Pulsed Radar System

TXRXAntenna

Motor

Motor Controller

Magnetron Modulator

HV, PW

Receiver

LimiterLNFE

IF AmplifierVideo Ampl.

STC

Video Processing

Power Supplies

Encoder, ARP & ACP’s

Control

Wave Guide

Circulator

Interface

Interface to external equipment

11

A Pulsed Radar System• A typical radar system

Antenna System

Transceiver• Transmitter• Receiver• Plot Extractor• Tracker• Control

Electronics• Interface

Two transceivers provides redundancy and hot-standby

12

A Pulsed Radar System• The Transceiver:

WG SSPA

WG Assy

SSPA

13

A Pulsed Radar System• The Transceiver – Waveguide Assembly:

13

14

A Pulsed Radar System• The transmission:

WG

WG

CouplerLimiter / STC

Adapter

Adapter

Transmission from antenna

SSPA

Signal coming from SSPA enters the circulator through WG SSPA

Signal passing the circulator cw and takes first exit

Signal enters the antenna through the installed WG

15

A Pulsed Radar System

• The reception:

WG

WG

CouplerLimiter / STC

Adapter

Adapter

Reflections from targets are captured by the antenna

RxTx

Signal arrives coupler, a -50 dB signal out for measurement purposes

Signal passing the circulator cw and takes first exit

Signal enters the circulator through the installed WG

RxTxControl

RxTx

Signal added through a-20 dB coupler - for calibration purposes

Attenuation of the signal up to 40 dB

Test port

RxTx

Signal passing the circulator cw and applied the RxTx module

16

• The antenna receives the EM energy from the transmitter and radiates the energy into the free space

Antenna System

• An isotropic antenna radiates the energy in a spherical pattern

• The energy is distributed equally in all directions

16

17

Antenna System• In practice, the radiated energy is focused in a

direction

• The radiation pattern describes how the energy is radiated (or focused)

• The characteristics of radiation pattern are beamwidth, gain and sidelobes

The ability to focus the EM energy gives the gain of the antenna

18

Antenna System• Radiation from a directional source• The energy is focused in a given

directions• This allows the energy to travel

further, hence a gain, G, compared to the isotropic source

18

19

Antenna System• Coastal Surveillance and Vessel Traffic System radars are

usually fan or inverse-cosecant-squared beams

fan beam pattern

Inverse-cosecant-square beam pattern

20

Antenna System• A 21’ Slotted Waveguide Array Antenna

21

Antenna System• Horn directs the EM energy and hence improves the

antenna gain (1)• The polarization filter gives circular or horizontal

polarization (3)• The antenna is protected by a

radome (4)• The antenna is radiated by the

slotted waveguide (2)Radome (front)

Horn

Polarization filterRF radiator(slotted waveguide)

Radome (back)

1

2 3

44 1

22

Antenna System• The beam pattern: beamwidth

23

Antenna System• Antenna performance:

Main Parameters:Frequency band 9.14 - 9.47 [GHz]

VSWR9.345 - 9.405 GHz ≤ 1.15

9.140 - 9.470 GHz ≤ 1.20

Gain ≥ 38 [dBi]

Integrated Cancellation Ratio ≥ 15 [dB]

Azimuth Pattern:Horizontal BW @ -3 dB ≤ 0.35 [º]

Side lobe level± 1.5º to ± 5º ≤ -28 [º]

± 5º to ± 10º ≤ -30 [º]

Exceeding ± 10º ≤ -35 [º]

Elevation Pattern:

Elevation beam form Fan

Vertical BW @ -3 dB ≤ 11 [º]

Min. coverage @ -30 dB -18 [º]

Tilt (fixed) -1.5 [º]

24

Antenna System• For mechanically-steered antenna, the bearing

information is obtained using encoder

Antenna

A B C

ABC

Direction (encoder)

25

Antenna System

Radome

Encoder Assembly and Rotatory Joint Module

Lineararray

Antenna

Tx

RxFlared horn

Connectionbox

Power

Encoder(s)Thermalsensors

Encoder (s)Rotary joint

WaveguideRF Flange

Turning unit

Gearbox

Thermalsensors

Motor

Antenna unit

Slotted waveguide (SWG)

26

Antenna System• The encoder:

ENCODERStationary Waveguide Entry

(to transceiver)

Rotating Waveguide Entry(to Antenna)

ENCODER INTERNAL

27

Antenna System• The encoder:

Channel A

Channel A

Channel B

Channel B

Channel N

Channel N

photodetector

LED light source

Rotating encoder disk, gray or Manchester coded generate the counting pulse

Bipolar pulses (RS422)

28

Antenna System• Whenever a stationary waveguide is to be connected to

a rotating antenna, a rotating joint must be used• In radars, rotary joints connect transmitter and/or

receiver to its rotating antenna• A circular waveguide is normally used in a rotating joint

Stationary Waveguide Entry(to transceiver)

Rotating Waveguide Entry (to Antenna)

29

Control

• It consists of a timing control that generates the synchronization timing signals

Timing Control

Modulator and Transmitter

Receiver

Signal Processor

Duplexer Antenna

To other modules

Scan Pattern Generator

(RPM, Beam Control, …)

30

Modulator• The modulator produces a high power DC pulse to the

transmitter• A modulated signal is generated and sent to the antenna by the

modulator and transmitter block• A train of narrow rectangular shaped pulses modulating a sine

wave carrier is transmitted

Pulse width

Pulse Repetition Time (PRT)

Rest Time (Listening)

Radar Carrier Frequency

30

31

Magnetron Transmitter• The transmitter is a high power

oscillator, e.g. a magnetron• The magnetron generates high power

RF wave • Transmitted pulse is high power, short

duration• The EM wave is sent to the duplexer

via waveguide (transmission line)• Transmitter remains silent during the

listening period

Magnetron

32

Magnetron Transmitter

• The two main limitations of magnetron are – Limited average power– Poor ability to detect moving targets in heavy clutter

• The peak power of several MW can be produced by magnetron, but the average power is limited to 1 -2 kW

• The pulse width limitation prevents the magnetron from being used with pulse compression where frequency or phase modulation is difficult

• Magnetrons are noisy outside the operating frequency

33

Marine Radar Magnetrons• The magnetrons have been used extensively in civil

marine radars• They are compact and can generate peak powers

between 3 and 75 kW with average powers from a few W to a few 10s of W

• They offer reliability which sea-goers require • Marine radar operates at a fixed frequency within

the band 9.38 GHz to 9.44 GHz

34

Marine Radar Magnetrons• A RF assembly that houses

both the transmitter and receiver Magnetron

Circulator LNFE

TR Limiter

IF AMP Assembly

35

The Solid-State Transmitter• Solid-state RF power generation is becoming more

common presently in VTR and marine radars• It is designed with transistor amplifiers (BJT or FET)• One stage transistor amplifier is low power and

low gain but it operates with low voltages and has high reliability

• To increase the power, parallel configuration and multistage transistor amplifiers are used

36

The Solid-State Transmitter• The advantages of Solid State Transmitter

– Individual solid-state devices have long MTBF– Maintenance is easy with modular design and

construction of solid state– Broad bandwidth – No cathode heating required (no need for high

voltage and warm up time)– No pulse modulator required– Solid-state transistor amplifiers have low noise and

good stability

37

The Solid-State Transmitter• Advantages of Solid-State radars over magnetron radars

PARAMETERS SOLID-STATE MAGNETRON

Maintenance Low High – annual magnetron replacement

Remote Control and Monitoring

Fully digital and remote controller

Requires digital interface for full remote control

Built-in-Test (BIT) Extensive and fully remote accessible

Limited

MTBF 50,000 hours 3,000 hours

Antenna Speed Electronically selectable, high speed scanning gives improved track quality

Standard 24 rpm, requires motor and gear replacements

38

The Solid-State Transmitter• Advantages of Solid-State radars over magnetron radars

PARAMETERS SOLID-STATE MAGNETRON

Antenna Tilt Yes No

Emitted Power Density

Low power mode for short range use and operations near hot ordinance

High

Variable Frequency Yes – electronic control No

Start-Up Instantaneous operation Warm up time

Spares Low High (replace magnetrons)

Frequency diversity (improved performance)

Yes No

Doppler processing Yes No

39

The Solid-State Transmitter• Meeting IEC 60963-1/62388 Requirements

Height (m)

RCS (m2)

IEC 60936-1 (nm) IEC 623880 (nm) Solid-State (nm)*

60 50000 20 20 24.2

6 5000 7 8 12.7

3 2500 --- 6 10.4

10 50000 7 11 15.6

5 1800 --- 8 11.6

4 7.5 --- 5 7.8

3.5 10 2 4.9 7.6

3.5 5 --- 4.6 7.2

2 2.5 --- 3.4 5.8

1 1 --- 2 4.1

40

The Solid-State Transmitter• The parameters of a solid-state radar

PARAMETERSX-BAND

S-BANDModel 1 Model 2

Overall length 3.7 m 5.5 m 3.9 m

Turning circle diameter 3.8 m 5.6 m 4 m

Frequency band 9.22 – 9.44 GHz

Gain 32.7 dB 34.5 dB 27.5 dB

Horizontal beamwidth 0.7 0.45 2 max.

Vertical beamwidth 2 2

Horizontal sidelobes within 10

-26 dB At least -28 dB dowm

Horizontal sidelobes outside 10

-33 dB At least -35 dB down

41

The Solid-State Transmitter• The detection performance of the X-band radar:

Antenna Height

Target Type

Modeled as fluctuating point target

Detection and Tracking distance(nm)

RCS (m2) Height (m) Clear Weather 10 mm/h Rain

50 mAMSL

1 1 1 m AMSL 10/8.25 (SS4) ----

2 3 2 m AMSL 12/11.23 (SS5) 9/10.64 (SS 5)

3 10 3 m AMSL 14/13.3 (SS6) 12/12.71 (SS6)

4 100 5 m AMSL 17/16.29 (SS7) 15/15.83 (SS7)

5 1000 8 m AMSL 20/19.2 (SS8) 18/18.71 (SS8)

• The results are obtained from CARPET based on PFA = 10–6 and PD = 80%

42

The Solid-State Transmitter• A Typical Transmitter Design based on SSPA

SSPA

WGAssy

RxTx

BP

BP

BP

BP

Com

bine

r

1300 MHz7825 MHz (VTS)7600 MHz (SMR)

1300 MHz 7825 MHz (VTS)7600 MHz (SMR)

9.225 - 9.5 GHz (VTS)9.0 - 9.275 GHz (SMR)

Tx R

AM

/ FP

GA

DAC /DDS

DAC /DDS

BP

BP

Atte

nuat

ion

Atte

nuat

ion

RxTx Control

100 - 375 MHz (low)

100 - 375 MHz (high)

43

A Typical Solid-State Transmitter• The design of a solid-state transmitter:

Power Amplifier

Power Amplifier

Module 1

Module 2

Pow

er s

plitt

er

Pow

er a

dder

Inco

min

g si

gnal

Tran

smitt

ed s

igna

l

44

A Typical Solid-State Transmitter• The solid-state power amplifier (SSPA) amplifies

the signal before transmission• The transmitted power ranges from 50 W (short

range) to 200 W (long range) typically• The required power is constructed using multi-

stage and parallel configuration of single stage power amplifiers PA module

In Out

8 W power transistor

50 W power amplifier

45

A Typical Solid-State Transmitter• The typical solid-state power amplifier

50 W

200 W

46

A Typical Solid-State Transmitter• The degradation performance

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0% 25%

50%

75%

100%

Percentage of power transistors in failure

Available SSPA power

Output power

Free space range

47

Radar Waveform

• The carrier is modulated using a scheme that is similar to that used in communication systems– ASK or ON-OFF AM

Pulse width, t

(TX)

Pulse Repetition Time (PRT)

Rest Time (Listening)

(RX)

Radar Carrier Frequency

48

Radar Waveform• The Chirp Pulse:

Time

Amplitude

Time

Frequency

f [MHz]

1 2 3 4 5 6

Chirp BW Separation

Time

Amplitude

Time

Frequencyor

35 Mhz 6 Mhz

f [MHz]100 375

f [MHz]100 375

1300 MHz

7825 MHz (VTS)7600 MHz (SMR)

100 - 375 MHz

1400 - 1675 MHz

9225 - 9500 MHz (VTS)9000 - 9275 MHz (SMR)

9260 9301 9342 9383 9424 9465

9441.5

9447.5

49

Radar Waveform• High resolution radar – short transmission pulses

are required– Short transmission pulse requires high transmission

power for long distance (B = 1/t)– Short transmission pulse gives large bandwidth

(receiver noise must be considered)• Best

– Long pulse for long distance– Short pulse width for high resolution– Small bandwidth for large dynamic range

50

Radar Waveform• The signal flow:

Antenna

Transmitter

Receiver /Processing

Power

TimePower

TimeTime

Received Echo

Transmitted Chirps withfrequency sweep

A B

B

A

AB

Equivalentcompressed

power

51

Radar Waveform• Pulse Compression

Every little part of the waverepresents a specific frequency

Low frequenciesare slowed down

High frequenciesare speeded up

Processing

52


Transmitted chirp

Received chirp

53


f1f2f3f4

Σ

Σ

Σ

f1

f2

f3

f4

ΣBandpass Delay Summation

54


f1f2f3f4

f1

f2

f3

f4

55


Σ

Σ

Σ

56

Radar Waveform• Pulse Compression – cross correlation

57

Radar Waveform• Pulse Compression – cross correlation

58

Radar Waveform• Pulse Compression – time side lobes

Antenna

59

Radar Waveform• Pulse Compression – windowing

60

Radar Waveform• Pulse Compression – windowing

61

Radar Waveform• Pulse Compression • Requirement

– Range resolution – 30 m– Pulse compression ratio – 75

• The compressed pulse duration

• The compressed bandwidths

cRt resolution

compressed 2.01033022

8

MHz5102.03011

6

compressedtBW

62

Radar Waveform

• Pulse Compression • Requirement

– Range resolution – 30 m– Pulse compression ratio – 75

• The required transmission pulse width

• If the SSPA gives 200W, the virtual transmitted power kW1575200 PCRPP SSPAvirtual

sPCRtcompressed t 1575102.0 6

63

Radar Waveform• Radar video with pulse compression

64

Radar Waveform• Radar video without pulse compression

65

Radar Waveform• For a single pulse, the maximum unambiguous

range, Ru,max, is determined by the PRF (1/PRT),

• High PRF is unambiguous in Doppler but highly ambiguous in Range since it meets the Nyquist sampling criteria for Doppler shift of all targets design to detect but there is little time between pulses for ranging

PRFcR

RcPRF uu .22 max,

max,

66

Radar Waveform• Medium PRF radar may be ambiguous

in both Doppler and range since it samples too fast for echoes from long range but too slow to Nyquist sample the Doppler shift of all targets

• Medium PRF however has the best of both worlds, a compromise performance between unambiguity in Doppler and range

• Low PRF is unambiguous in range but high ambiguous in Doppler since it waits until the last transmitted pulse arrives before the next transmission

Low PRF

tPRT = 1/PRF

f0 PRFPFF

High PRF

t

f0 PRFPFF

67

Radar Waveform• Different PRF is used in different application such as Low

PRF or Medium PRF is suitable for detection of vessels while High PRF is used in detecting air targets

• Careful selection of PRF will also result in better tracking performance since one must be able to detect a target before tracking it

• Mixture of pulse durations enhances the performance

68

Radar Waveform• Time duration variations of chirps:

Rx

Tx

Rx Rx Rx

Tx Tx Tx TxRx

Tx

Rx

Sweep n Sweep n+1 Sweep n+2 Sweep n+3 Sweep n+4 Sweep n+5

tInstrumented range

Minimum range

Antenna rotation

SM

L Sweep n

Sweep n+1

Sweep n+2

Sweep n+3

Sweep n+4

Sweep n+5

Depending on range required and application, three combinations of chirps can be used:1. combination: SHORT, MEDIUM and LONG

69


Rx

Tx

Rx Rx Rx

Tx Tx Tx TxRx

Tx

Rx

Sweep n

Sweep n+1 Sweep n+2 Sweep n+3 Sweep n+4 Sweep n+5

t

Instrumented range

Minimum range

Antenna rotation

S

M Sweep n

Sweep n+1

Sweep n+2

Sweep n+3

Sweep n+4

Sweep n+5

2. combination: SHORT and MEDIUM

70


Rx

Tx

Rx Rx Rx

Tx Tx Tx TxRx

Tx

Rx

Sweep n

Sweep n+1 Sweep n+2 Sweep n+3 Sweep n+4 Sweep n+5

t

Instrumented range

Minimum range

Antenna rotation

S Sweep n

Sweep n+1

Sweep n+2

Sweep n+3

Sweep n+4

Sweep n+5

3. combination: SHORT only

71

• The receiver amplifies the radar returns and prepares them for signal processing

• Extractor of target information from the returns is performed by the signal processor, tdelay is determined for each detected target

• Target distance is calculated from the total time (tdelay) taken by the pulse to travel to the target and back

c = 3 x 108 m/s, speed of light1 s = 300,000,000 m= 300,000 km1ms = 300 km

Target Range

72

Target RangeThe PRF determines how often the radar transmits (PRF = pulses/s = 1/PRT or 1/PRI) and the maximum range of the system (PRT = Pulse Repetitive Time, PRI = Pulse Repetitive Interval)

Transmitted pulses

Pulse travels 150 km in 0.5 ms

Maximum range = 0.25 ms (75 km)Echo needs 0.25 ms to return

Transmitted pulses

Pulse travels 300 km in 1 ms

Maximum range = 0.5 ms (150 km)Echo needs 0.5 ms to return

PRF = 2 kHzPRT = 0.5 ms

PRF = 1 kHzPRT = 1 ms

1 ms

73

Average Power• In each cycle (PRT), the radar only radiates from

t sec (known as pulse width or PW)• The average transmitted power is

where Pt = peak transmitted power andPRF = Pulse Repetition Frequency,

. .av t tP P P PRFPRTt t

1PRFPRT

74

Range Ambiguity• The range that corresponds to the 2-way time delay is

the radar unambiguous range, Ru

• Consider detection of 1 target at R1 in two separate transmissions

TransmittedPulses

ReceivedPulses

Pulse 1 Pulse 2

echo 1 echo 2

PRT

t

tdelaytdelay

Ru

(R )1

R 2

75

Range Ambiguity

• Echo 1 is the return from target at range R1,

• Echo 2 is the return from the same target at range R1, from the 2nd transmission

• Echo 2 can also be taken as an echo from a different target from the 1st transmission

1 2delayct

R

2 1 2delayct

R R

2 2

delayc PRT tR

ERROR!!!

76

Range Ambiguity• The maximum unambiguous range: ,max 2 2u

cPRT cRPRF

Page 2 of 4

Range: 5 NM

Transmit

Pulse 1t = 0 Next pulse

Time needed for the pulse to hit the target: T = 31 µs

Time needed for the pulse to return to the antenna: T = 31 µs

Time needed between two pulses: T > 62 µs or PRF < 16.1 kHz (corresponding 2 x range, here 10 NM)

Target at 5 NMt = 31 µs

Receive

Pulse 1t = 62 µs

PRF [Hz]

10001500220040008000

[km]

15010068.237.518.7

Max. range [NM]

8154

36.820.210.1

300.000 km/s

300.000 km/s

Max. range = c2PRF

150.000PRF

81.000

PRFkm NM

76

77

Range AmbiguityThe illustration of range ambiguity using lattice diagram:

t

PRF 3.75 kHz, T = 266.7 µs

30 km

T = 266.7 µs

Tx pulse n

Tx pulse n+1

Tx pulse n+2

Tx pulse n+3

40 km

T = 266.7 µs

T = 266.7 µs

T = 266.7 µs

RANGE 40 km

Tx pulsesEchoes 30 kmEchoes 60 km

Range

60 km

200 µs

200 µs

200 µs

200 µs

133.3 µs

133.3 µs

133.3 µs

78

Range Ambiguity

Fals

e ta

rget

True

targ

et

T0

t-nt

t-nt

t-nt

t-nt

T0Sweep 1

T0Sweep 2

T0Sweep 3

T0Sweep 4Fa

lse

targ

et

True

targ

et

T0AND

T0ADD

79

Range Ambiguity• Staggered PRF is used to avoid “jamming” or interference from other

radars’ transmitting and “second time around”• The change of repetition frequency does that the radar on a pulse to

pulse basis can differentiate between returns from itself and returns from other radar systems with same frequency.

T = 1PRF

Without staggerPRF: 1 kHz

= 1 ms T = 1 ms T = 1 ms T = 1 ms

T = 1PRF

With stagger 8%PRF: 1 kHz

+ 8% = 1.08 ms T = 1PRF - 2% = 0.98 ms T = 1

PRF + 4% = 1.04 ms T = 1PRF - 8% = 0.92 ms

80

Range Ambiguity

T0

Fals

e ta

rget

t-n t-m t-p t-qTr

ue ta

rget

t t t t

Fals

e ta

rget

True

targ

et

T0

T0

T0

T0

Sweep 1

Sweep 2

Sweep 3

Sweep 4

AND T0

ADD T0

81

Range Resolution• Range resolution, R, is the radar ability to

detect targets in close proximity as 2 distinct targets

• 2 close proximity targets must be separated by at least R to be completely resolved in range

• Consider 2 targets located at ranges R1 and R2, corresponding to time delays t1 and t2 respectively, the difference between the 2 ranges is 2 1 2 12 2

c cR R R t t t

82

Range Resolution• To distinguish the 2 targets, they must be

separated by at least 1 pulse width t,

where B = radar bandwidth2 2c cR

Bt

ReceivedPulses

returntarget 1

c

returntarget 2

t ct

target 1 target 2

ct

R R1 2

83

2 2c cR

Bt

Range Resolution

Range resolution:

Pulse width = 1 µs, length of pulse flying through the air = 300 mDistance is half of 300 m = 150 m

For a 150 ns pulse, the range resolution,

m 5.222

101501032

98

tcR

Two targets must be separated by at least 22.5 m so that they can be detected as two distinct targets by the radar with a pulse width of 150 ns.

84

Range Resolution

PW: 1 μsLength: 300 mRange resolution: 150 m

150 m

120 m

PW: 250 nsLength: 75 mRange resolution: 37.5 m

50 m

30 m

PW: 50 nsLength: 15 mRange resolution: 7.5 m

10 m

7 m

85

• The number of wavelengths contained in the two way path between the radar and the target,

• Total phase shift,

• Radar is able to give radial velocity of a moving target from Doppler Effect

• Doppler effect causes a shift in frequency of the received echo signal from a moving target

• Doppler frequency shift• Let R be the Range of the target

Doppler Effect

RADAR TOWER

INBOUNDECHO

RADARANTENNA

TRANSMITPULSE

OUTBOUNDECHO

2Rn

42 Rn

86

• When the target is moving, R and φ change continuously

• The rate of change of φ is angular frequency

where vr = radial velocity of the target towards the radar

• The Doppler frequency shift,

where

Doppler Effect

442 rd d

vd dRfdt dt

22 r ord

v fvfc

cosrv v

87

Duplexer• The duplexer is a waveguide switch• It passes the transmitted high-power pulses to the

antenna and the received echoes from the antenna to the receiver

• Duplexer switches automatically based on the timing control signal

88

Receiver• The receiver picks up the target signal from the

received signal, which includes noise and interference• Performance requirement:

– To detect the targets with highest probability of detection (PD)

– To minimize the false alarm (classifying noise, clutter or interference as target) (PFA)

• PD and PFA are related to the SNR, stronger signal power over the noise power gives better performance

89

Receiver

Time

Rec

eive

d po

wer

ThresholdTargetreturns

Randomnoise

• 3 out of the 4 targets detected are real targets, the PD = 75%

• If the no. of noise spikes above the threshold is 1 out of 50, the PFA = 2%

90

Radar Detection• Radar detection is a binary decision problem and

the process is statistical

91

Radar Detection• It is necessary to decide the presence of a target

when– The target is usually embedded and corrupted by random

noise from the atmosphere, thermal, …– The target RCS is also statistical

• Consider the following radar system:

92

Radar Detection• The radar transmits and the

detector output is measured 100 times at a fixed time interval without backscattering

• The 100 samples are tabulated using a histogram

• The histogram is a discrete probability function with finite probability interval and limited number of measurements

V Relative Frequency,

n0 – 0.1 2

0.1 – 0.2 10

0.2 – 0.3 16

0.3 – 0.4 22

0.4 – 0.5 19

0.5 – 0.6 13

0.6 – 0.7 8

0.7 – 0.8 5

0.8 – 0.9 3

0.9 – 1.0 1

> 1.0 0

93

Radar Detection• The discrete probability histogram,

0.02

0.1

0.16

0.220.19

0.130.08

0.05

0.030.01

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

No. of Samples, N = 100

V

nN

1V

nN

0S

94

Radar Detection• As the interval approaches 0, the discrete time

histogram becomes a continuous time function, called Probability Density Function

PDF for S/N = 0 (no target)

95

Radar Detection• The voltmeter is replaced by a threshold

detector and indicator where when

0 when o T

T

V V V VV V

96

Radar Detection• As there is no target, any measured voltage that

is greater the threshold gives a false alarm

Single pulse Probability of False Alarm

97

Radar Detection

• The probability of false alarm (PFA) depends on the threshold

• The threshold can be set to give a desired PFA, for instance, PFA = 0.1 (this means that the shaded area is 10% of the total area under the PDF curve)

• The measurements can be carried with S/N = 1, S/N = 2, S/N = 4 for different targets

98

Radar Detection• The PDFs for different signal strengths

Single pulse PDFs for different S/N

, , ,T

D T D FA TV

S S SP V P P P V V p V dvN N N

99

Radar Detection• The area under the PDF curve for S/N > 0 give the

probability of detection (PD) for the required PFA = 0.1• For a different PFA, the PD will change when VT changes

For a smaller allowed PFA, the PD is smaller for a given S/N

100

Ordered statistics Page 1 of 2

42111402

16639

104

177

1211581

13

13151135169101211

58141151447119

1016401751217155

0 - 2

3 - 5

6 - 8

9 - 11

12 - 14

15 - 17

Interval Frequency

5

12

5

12

8

8

50

Acc. frequency in %

10%

34%

44%

68%

84%

100%

Frequency in %

10%

24%

10%

24%

16%

16%

100%

Receiver Processing• Ordered Statistics

101

Ordered statistics Page 2 of 2

0 - 23 - 56 - 89 - 11

12 - 1415 - 17

Interval Frequency

51251288

Acc. frequency (%)

10%34%44%68%84%

100%

0 - 23 - 56 - 89 - 11

12 - 1415 - 17

Interval

Receiver Processing• Ordered Statistics

102

u

t

PD = 20% (1 target of 5)

PD = 80% (4 targets of 5)

PFA = Too high (threshold too low)

PFA = Constant (CFAR)

Receiver Processing• Constant False Alarm Rate (CFAR)

Adaptive Threshold

103

u

t

PPA = Constant, ex. 10-6

u

t

Receiver Processing• Received signal

104

0 m End of range

0 m End of range

Range cell containing 8 bit video information

0 m End of rangeOrdered statistics

Ordered statistics

Ordered statistics

Ordered statistics

Ordered statistics

.....and so on

Receiver Processing• Range cells

105

Cell Level Cell Level Cell Level Cell Level1 38 33 99 65 132 97 452 52 34 66 66 154 98 823 60 35 112 67 136 99 374 30 36 98 68 100 100 415 45 37 85 69 165 101 566 63 38 125 70 143 102 477 61 39 100 71 154 103 628 38 40 105 72 134 104 619 43 41 127 73 112 105 35

10 60 42 75 74 143 106 8011 51 43 92 75 153 107 1712 37 44 95 76 154 108 1513 44 45 81 77 199 109 3114 39 46 115 78 167 110 3215 51 47 220 79 182 111 3416 42 48 251 80 176 112 1417 6 49 238 81 165 113 3218 8 50 243 82 181 114 5519 16 51 243 83 200 115 3120 20 52 232 84 210 116 4221 15 53 222 85 179 117 1322 29 54 249 86 172 118 2723 36 55 228 87 213 119 5424 12 56 241 88 203 120 2225 24 57 190 89 176 121 3026 14 58 237 90 186 122 5727 21 59 220 91 199 123 4328 30 60 201 92 105 124 5229 18 61 247 93 117 125 3430 25 62 160 94 38 126 831 32 63 135 95 55 127 1532 82 64 143 96 32 128 3

0

50

100

150

200

250

300

1 8 15 22 29 36 43 50 57 64 71 78 85 92 99 106 113 120 127

Cell

Bit L

evel

Bit Level

Floating window covering 128 cells

Receiver Processing• Range cells statistics

106

Interval Frequency0-15 11

16-31 1532-47 2348-63 1564-79 280-95 796-111 6

112-127 6128-143 7144-159 4160-175 5176-191 7192-207 5208-223 5224-241 4240-255 6

Diagram showing the intervals and the frequency in each.


107

LUT (Look-up Table)

The difference in signal level between 50% and 90%

fractile, is used to point out an additional attenuation in the

LUT

- Att.

Signal Level(90% - 50%)

Interval Frequency [%] Frequency [%], accumulated

0-15 8,6 8.616-31 11,7 20.332-47 18,0 38.348-63 11,7 50.064-79 1,6 51.680-95 5,5 57.196-111 4,7 61.8

112-127 4,7 66.5128-143 5,5 72.0144-159 3,1 75.1160-175 3,9 79.0176-191 5,5 84.5192-207 3,9 88.4208-223 3,9 92.3224-241 3,1 95.4240-255 4,7 100.0


108

Noise Sea

50%

90%

50%

90%

256 (total)

50% 90%

25

Delta values (value at 90% - value at 50%) are used to point out a value in the LUT - a

guesstimate of the level in the upper 10%

Delta values (value at 90% - value at 50%) are used to point out a value in the LUT - a

guesstimate of the level in the upper 10%


109

90%

50%

Δ = Hi

Δ = Lo

50%

90%

Sea

Δ = Hi

99.9999%

50%

90%

Noise

Δ = Lo

99.9999%


110

50%

90%

Sea

Δ = Hi

99.9999%

50%

90%

Noise

Δ = Lo

99.9999%

Δ valueThe delta value is used to guesstimate the value at 99.9999%, PFA = 10-6.The value of the attenuation to obtain this, is fetched from the LUT and therefore the attenuation depends on the distribution curves

Δ value

LUT

Δvalue

90% CFAR: 10-6


111

Noise Clutter Noise

Signal consisting of noise and sea clutter

Attenuation working OK and the clutter is attenuatedto the level of the noise.

Undercompensated - the clutter is not sufficiently attenuatedand will make a concentration of clutter on the radar image.

Overcompensated - the clutter is attenuated too hard and willmake a "hole" in the noise on the radar image.


112

Page 11 of

11

Receiver Processing• CFAR effect:

113

Average is stored in weep n+3

For the next integration cycle sweep n+1 will be sweep n

Sweep nVideo

Noise

Receiver Processing• Sweep Integration:

114

Sweep n

Sweep n+1

Sweep n+2

Sweep n+3

ADD

ADD / 4


115

0 100 200 300 400 5000

2.5

5

7.5

10

12.5

1514.686

2.19 10 3

single0i

single1i 4

single2i 8

single3i 12

4990 i

4 sweeps

4 sweeps integratedgives an improved S/N

0 100 200 300 400 5004

5

6

7

8

9

109.773

5.793

inco_inti

4990 i


116


117

Receiver Processing• Scan Correlation

– The radar performs sliding window scan-to-scan correlation for each range-azimuth cell to provide clutter discrimination

– This is done to discriminate between clutter and targets that are of interest to the operator

118

Scan Correlation• One channel performs sliding window scan-to-scan correlation

over 3 consecutive scans to discriminate between clutter and extremely small targets with speeds up to e.g. 8 knots - the actual speed limit is determined by the antenna rotation speed, the pulse width and the range

• A second channel performs sliding window scan-to-scan correlation over 2 consecutive scans in order to discriminate between clutter and small targets with speeds up to e.g. 16 knots - the actual speed limit is also here determined by the antenna rotation speed, the pulse width and the range

• A third channel without scan-to-scan correlation to detect targets of medium and large size (RCS) at any speed

Receiving Processing

119

Receiving Processing• Scan Correlation

120

Receiving Processing• Sweep Integration:

Gate signal

No gate signalAN

D

Gate signal

No gate signalAN

D

1 3 5 7 92 4 6 8 10Scan n-2

1 3 5 7 92 4 6 810Scan n-1

3 s1 3 5 7 92 4 6 8

10

Clutter

1 3 5 7 92 4 6 810Scan n

3 s

Target Returned echo (PW = 500 ns)Resolution cells (10 x 7.5 m)

ThresholdLevel

Target Returned echo (PW = 500 ns)Resolution cells (10 x 7.5 m)

ThresholdLevel

121

AN

DA

ND

1 3 5 72 4 6 8Scan n-2

1 3 5 72 4 6 8Scan n-1

3 s

1 3 5 72 4 6 8Scan n

3 s

Gate signal

Receiving Processing• Sweep Integration:

122

A Typical Tracking Radar System• The receiver decides that an object has been

detected (signal strength > threshold)• The signal processor will determine the presence of

real targets and rejects the unwanted returns

123

A Typical Tracking Radar System• The data extractor will determine the target

measurements such as range, azimuth and/or elevation (2D or 3D)

124

A Typical Tracking Radar System• The data processor will perform the association,

tracking and prediction task and maintains the target database

• Echo – a group of coherent detections within the same sweep

• Plot – a group of echoes within the same radar scan correlated in range and bearing in accordance to some pre-defined criteria such as center of gravity, peak amplitude, peak size, …

125

A Typical Tracking Radar System• Tag/ID – an identification generated and

assigned to a target• Overlay – a group of maps, symbols, navigation

channel/pathway and text that are presented on radar display

• Track – a sequence of kinematic state (position, speed, course, …) estimate of a target based on past plots

126

A Typical Tracking Radar System• A typical radar with

tracker is made up of tracking module/algorithm, display, scan conversion and user interface

• The exchange of data is achieved via a local area network (LAN)

LAN AREA NETWORK

127

A Typical Tracking Radar System• The processors are connected to the system bus

or network where radar input signals (ACP, ARP) and videos (targets, clutters) are processed and shared

• The radar videos are packaged in LAN message protocol for sharing and further processing

• The target echoes are identified as plots and tracking process is initiated

• The tracking information is then distributed via the LAN

128

A Typical Tracking Radar System• A typical radar data distribution over network

Video, plots, tracks, BITE,

control

LAN

RADAR TRANSCEIVER

AND PROCESSOR

3rd PARTY CONTROL AND COMMNAND

SYSTEM

RADAR DISPLAY AND

WORKSTATION

RADAR SERVICE TOOL

129

A Typical Tracking Radar System• Data interface and distribution

RADAR TRANSCEIVER

TRACKING MODULE

Radar Video

LAN

Own speedPosition HeadingMap

Control equipmentMonitoring equipment

Display

3rd Party Systems

TracksTrails

130

A Typical Tracking Radar System• Usually, extractor and tracker are housed under

one single module

Data Extractor

Data Processor (Tracker)

Plots

Settings

Radar Video

Platform and Navigation Data

Map Data

Status

Tracks

Plot Position

External Tracks

131

Radar Scan Conversion• The radar scan

converter receives raw radar video and converts it into a format for raster scan display

• The radar interface takes in – video and antenna signal

from the radar– the radar input gain and

132

Radar Scan Conversion• A surveillance radar supplies its measured data in polar

coordinates, i.e. in range-azimuth format (r,b)• Radar scan conversion involves the transformation of data in a

range-azimuth format to an x-y format for a raster-scan displayRad

ar be

am

133

Radar Scan Conversion• The conversion is carried out using x = rsinb + Xc and

y = rcosb + Yc where (Xc, Yc) is the radar center• The radar stores its raw data into a table called Polar

Store (e.g a 256K cells)• The table columns represent all possible measurable

range steps• Each table row represents a complete pulse period,

which took place in a certain azimuth angle• The first line is the azimuth angle of 0°, the north

direction

134

Radar Scan Conversion• The radar interface will also accept the existing

azimuths (2048 – 11-bit antenna word) per 360• The actual acquired data in the form of rho-theta is

stored in the display frame stores (e.g. 2048 by 2048)• The scan conversion algorithm handles the 2048 by

2048 resolution display in real time up to 60 r.p.m. antenna rotation rate

• The scan conversion process is controllable • General graphics drawing capability is provided for

flight/navigational plan and mask definition

135

Radar Scan Conversion• Scan conversion can also be performed in several

windows • The display controller graphics accelerator of the

display card is supported by at least 2 GB of RAM to provide exceptional flexibility for generating combined graphics and radar video and with an output resolution of up to say 1600 by 1280

• The display shall consists of symbols and targets layer, radar video layer which fades and become transparent and maps and charts layer

136

Plot Extractor• A radar plot is a group of connected range-azimuth cells

whose measured signal strength exceeds a defined threshold

• A plot is displayed which the video signal strength is larger than the threshold

• Technique such as STC (Sensitivity Time Control) improves the target detection by suppressing the echoes nearby the radar

137

Plot Extractor• In ideal situation, every target (wanted or

unwanted) illuminated by the radar beam during the present sweep should result in only one echo

• The plot extractor processes the radar data produced by the radar input to generate echoes every sweep by identifying echoes that have a size, shape and signal strength that is consistent with a valid return

138

Plot Extractor• Plot extractor is software application that analyses radar video

and detects potential targets• Processing is based on three main phases:

– clutter suppression– adaptive thresholding – plot extraction

• Clutter suppression can be based on Clutter Map Constant False Alarm Rate (CFAR) algorithm

• Adaptive thresholding implemented can be Cell Averaging CFAR (CA-CFAR) detection

• Plot extraction is detection of target-like shapes among video samples that exceed the threshold using standard m-of-n separation criterion

139

Plot Extractor• Plot Extractor and radar input is direct (not TCP/IP) • Since Plot Extractor is the only component that

requires high sample rate, it is optimal to place it on the same machine as the signal source

• Another consumer of raw video samples is Radar Display

• Since Radar Display requires lower sample rates, Plot Extractor also performs downsampling of the video before streaming it on the network

140

The Tracking Process• A tracking process is generally made up of 3

stages namely, Initiation, Updating and Termination

141

The Tracking Process• A good tracker will be able to provide better

detection of highly maneuvering small targets

142

The Tracking Process - Initiation• When a target is acquired, the initiate position

(Cartesian coordinates) is taken from the tracker ball (some call it order marker) or generated by processor

• The radar return defined by range and bearing (polar coordinates)

143

• The conversion from polar to Cartesian coordinates is carried out every antenna scan, when a track is confirmed, the target speed and course can be calculated by the data processor

The Tracking Process - Initiation

144

• The initiation process

Predicted position

Search window

Track

Radial accelerationTangential accelerationMinimum speedMaximum speed


145

• In order to associate correct plots to correct tracks, false alarms must be minimized, one conventional way is to correlate plots from more than 1 scan

• The antenna is divided into several azimuth sectors, plots received from, say, 2 previous scans (e.g. N-1 and N-2) are stored in accordance to the sectors they are detected

• The plots from current scan are then compared to these stored plots, the decision criteria for a successful correlation can be 2 out of 3 or generally, m-out-of-n


146

• Consider a plot PN in sector S of antenna scan N, an initiation window will be opened at this plot position, a search is then made in N-1 plot database to find a plot within this window


147

• Assuming only plots in sectors S-1, S and S+1 are search (for minimized comparison time) and that a plot PN-1 is found, then another search is carried out in N-2 plot database


148

• If the search is successful in N-2 plot database then the primary track initiation is successful, however, if there is no plot in N-2 plot database, then the primary track initiation can also be considered successful if the decision criteria is 3-out-of-2


149

• The available plots are then used to calculate the target position in x-y coordinates, the speed vector and the predicted next position (by filtering)

• They are also used to determine the quality of the track (centroid) and subsequently the gate size

Centroid is of a target is defined as the value of independent variable (in time) where the area under the target spectrum to the right equals to the left


150

• Given a plot at radar scan N, it is necessary to determine the shade (usually not implemented) and size of the correlation gate in order to capture the plot at scan N+1, the gate is usually defined in terms of range and azimuth


151

• The prediction processWith good

predictions the search window

gets smaller

With bad predictions the search window

gets bigger


152

• Correlation window eliminates the unlikely plot-to-track pairings and hence reduce unnecessary computation

• During the initial acquisition, usually the 1st 2 scans, the processor does not have sufficient knowledge of the target’s course and speed, the gate is larger to increase the match probability

• After several scans, the target’s motion is better known, the gate size reduces, the gate size is determined by the radar measurement error and possible maneuvering


153

• After the new track is initiated and vector generated, track updating is necessary to update the track with correct plots obtained so as to update current track information (position, speed, course) and predict future position

• The 1st task involves comparing the incoming plots with known tracks to obtain a list of all possible plot-track pairs

• One commonly used comparison method which results shortest possible comparison time is to organized the tracks such that they are compared with smallest possible no. of plots

The Tracking Process - Updating

154

• Depending on implementations, a central sector covering >1 track sectors, e.g. 2 track sectors and therefore 3 plot sectors, is defined for detecting plots that have evolved to non-adjacent sectors (fast moving targets or small sector size)

• If there exists a central sector, an unsuccessful correlation of plots (from the 2 adjacent plots) to a track, comparison is made with plots in the central sector


155

• 1 simple way to resolve such problem is to generate a plot-track distance matrix, a plot is then associated to a track it lies within the gate, the distance between the correlated plot and the track is computed and stored in the matrix

• After the comparison cycle, some common situations can occur– 1 plot is associated to 1 track – 1 plot is associated to several tracks– several plots are assigned to 1 track– no associated of plot to any track


156

• Consider a plot-track distance matrix, it is obvious that plot 3 is associated to track 3 (deleted from matrix), plot 1 is associated to track 1 (delete from matrix)


Track 1 Track 2 Track 3Plot 1 1 4 15Plot 2 3 3 13Plot 3 12 10 3Plot 4 5 3 10

157

• Plot 2 and 4 can be associated to track 2, one simple way to resolve the problem is to assign plot 2 randomly (wrong assignment will be rectified after a few more scans)

• The remaining plots are considered for new track initiation and remaining tracks can considered termination


Track 1 Track 2 Track 3Plot 1 1 4 15Plot 2 3 3 13Plot 3 12 10 3Plot 4 5 3 10

158

• After association, the track information are stored in the target database; the measured position (if there is an associated plot) is updated, otherwise the predicted position is used

• Predicted position is obtained by filtering (e.g. Kalman or ab filter) which filters and predicts target position


159


Predicted position

Search window

Track

Updated track

Out of search window

Out of shape

New search windowNew prediction

• The plot-to-track correlation

160

• The quality factor, status of the track, gate size and position is computed

• The quality factor of a track depends on the track history (how good it was), it is used to compute the gate size for the next scan (smaller or larger)

• The quality factor is computed as

where K = +ve when there is a plot in the gate and K = -ve there is no plot in the gate


KQQ nn 1

161


162

• This process is performed once every scan for every track, an associated plot is known as a hit where a non-associated plot is a miss

• A track is cancelled automatically when plot is absent for a predefined no. of scans consecutive or Q = 0 (no point tracking)

• A track is also terminated manually when drop track function is invoked

• This erases all data corresponding to the track in the target table, the lost target symbol is displayed for reference

The Tracking Process - Termination

163

• A miss

The Tracking Process - Termination

Predicted position

Search window

Track

This will count as a lack –no update found

164

The End

A Tutorial on Radar System Engineering

Engineering

Transcript of A Tutorial on Radar System Engineering