TTank-automotive & AArmaments COMCOMmand

Advanced Acoustic Sensor TechnologiesNDIA Symposium

Session III Advanced Technologies20 June 2001

Briefer: Jeffrey HeberleyTechnical Executive, FSAC, TACOM-ARDEC 1/n

Report Documentation Page

Report Date 20JUN2001

Report Type N/A

Dates Covered (from... to) -

Title and Subtitle Advanced Acoustic Sensor Technologies

Contract Number

Grant Number

Program Element Number

Author(s) Heberley, Jeffrey

Project Number

Task Number

Work Unit Number

Performing Organization Name(s) and Address(es) TACOM-ARDEC

Performing Organization Report Number

Sponsoring/Monitoring Agency Name(s) and Address(es) NDIA (National Defense Industrial Association 2111Wilson Blvd., Ste. 400 Arlington, VA 22201-3061

Sponsor/Monitor’s Acronym(s)

Sponsor/Monitor’s Report Number(s)

Distribution/Availability Statement Approved for public release, distribution unlimited

Supplementary Notes Proceedings from Armaments for the Army Transformation Conference, 18-20 June 2001 sponsored by NDIA

Abstract

Subject Terms

Report Classification unclassified

Classification of this page unclassified

Classification of Abstract unclassified

Limitation of Abstract UU

Number of Pages 36

OUTLINE• BACKGROUND

– WHY ACOUSTICS– TECHNOLOGY EXPLOITED– PRIOR ARDEC PROGRAMS

• PRIOR TECHNOLOGY/PROGRAMS– FAAD– HELO & BAT– COUNTER SNIPER– RFPI

• CURRENT TECHNOLOGY/PROGRAMS– NINOX– RAPTOR

• CLASSIFIER• TARGET COUNTER

– TECH BASE (6.2)• ACOUSTIC COUNTER BATTERY SYSTEM (ACBS)• ACOUSTIC/SEISMIC MODELING• NETWORKED DISTRIBUTED SENSORS

Army Benefits

• Passive• Day/Night/Adverse Weather• NLOS Threat Target Detection• NC-IFF, PHID (Avoids Fratricide)• Acquire Threats at Stand-off Ranges• Support Shoot-on-the-Move• Range to Target

BATTLEFIELD SENSOR COMPARISON

ESMSYSTEMSInfra-Red

Search and Track

FLIR

Light SHORAD Radar

Acoustic

5 10 15 20 25 30 35 40 45 50 55 60 65 70

1,000

900

800

700

600

500

400

300

200

10050

UN

IT C

OS

T IN

TH

OU

SA

ND

S O

F D

OLL

AR

S Heavy Division SHORAD Radar

RANGE in Km

Concept Definition• System Description

– Acoustic Sensors for Target Detection,Tracking and Location

• Unique Capabilities– All weather, Day/Night, All Terrain Target

Tracking– Provide Situational Awareness– Low Cost– BCID (Battlefield ID/Classification)– Passive and Resists CM– Promotes Fratricide Avoidance

• Operational CapabilityRequirements (OCRs) Addressed

– BC01, BC09, DSA06, DSA12, DSA13,DBS01A, DBS03, DBS04A, DBS05A,DBS10, DBS12, MTD04, MTD14, MTD22,EEL13

Operational BenefitLow cost, passive acoustic sensorsystems provide non-line-of-sightsituational awareness and targetacquisition and handoff to weaponsystems fire control. New integratedwarfighting capabilities are provided through sensor fusion and battlefielddigitization.

BBN-12 Channel Acoustic HelmetBBN-12 Channel Acoustic HelmetHeading SensorHeading Sensor

BCD/JWS/RFPI-2 10WHERE TOMORROW'S VICTORIES BEGIN

Increase Early Entry Force Survivability, Lethality, Target Acquisition,and Control of Battle Tempo against Heavy Armored Threat

RFPI ACTD HUNTER/STANDOFF KILLER CONCEPT

HIMARS155MM HOWITZERAUTOMATED FCS

Digital C4Battle

Command

FOFAC

FEBA

FLOT

FLOTFEBA

Extended Battlespace

Stand-OffKillers

AdvancedForwardSensors

MAIN BATTLE AREA SECURITY AREA

EFOGM

DEPLOY FORWARDSENSORS

TARGET ACQUISITIONWITH FORWARD

SENSORS

ENGAGEMENT

ARMY AVN

DRB (-) LIFT CONSTRAINEDEARLY ENTRY TASK FORCE

HASTY DEFENSE

ARMORED THREAT

UAV

IREMBASS

REAL TIMETARGETING

PROCESS

INTEGRATEDACOUSTICSYSTEM

FREE-PLAYOPFOR

LD TOC REMOTE SENTRY

HUNTER SENSORSUITE

IAS Array Configuration

Air Deliverable AcousticSensor

- Detect, track, and classifyground/air vehicles- 4’ aperture, 5 mic array, DSP- Hand emplace or air deploy w/ optional parachute- Self mapping via GPS- Separate long haul and short haul data radios

IAS System Components

CONTROLSTATION

TOC LAN

ADASAir Deliverable

Acoustic Sensor

HE-ADASHand Emplaced

ADAS

CommRelay

<75 km

S

S

S

S

S

S

S

S

S

<20 km

S

S

S

R

Hand-Off ToTOC LAN

Fused Sensor Data - Coordinates - Classification

Acoustic CRADA (TSD & ARDEC)ARDEC to develop improved air-acoustic signal processing techniques for IAS/ADAS

•Advanced detection & classification methods

•Field test facility support (ADAS units, site, drivers, etc.)

•GFE ADAS units for Operational Testing

•Textron to support & implement

•Tech support & consultation to above tasks

•Provide GDAS to ARDEC for Development Testing

•Implement ARDEC algo improvements in ADAS S/W

•Field test support (personnel, met, truth, etc.)

Compelling Australian Need

* Alice Springs

* Sydney

* Brisbane

* Melbourne

Adelaide * * Canberra

Perth *

Australian Army<25,000 Total

Ninox UGS

70 sites in the Northfunded in current program~$20-33M US

Eventual requirementmay exceed 1000 sites >$200M US

* Darwin

Pop. >150M

Pop. 0.7M

Development Plan

ADAS Prototype OASIS - ADAS H/W & S/W modifications funded by Contractors

Confirmatory Demo Nov,1999

OASIS Deliverables - Development completed under NINOX UGS contract

Current Hardware 1998 Deliveries 2001-2002

Some Key Features• Beamforming Acoustic Array (TSD)

• Long Range Discrimination & Tracking of Motor Vehicles

• Distributed Mini-Sensors (RACAL->Thompson->THALES)

• Seismic, Magnetic, & Passive Infrared• Personnel Detection & Back-Up for Acoustics

• Precision Cued Day/Night Electro-Optics (TENIX)

• Operator in the Loop Target Recognition

• Satellite Based Long Haul Communications• Operation in Remote Areas - Unlimited Range

• Advanced Integrated Control Station• Remote Situational Awareness

Terrain Commander

OASIS - Optical Acoustic SATCOM Integrated Sensor

- OASIS Day/Night Electro-Optics Head

- Rucksack

- Central Monitoring Facility (CMF)

- OASIS Base Unit w/ 5 Mic Beamforming Acoustic Sensor & Satellite Comms.

- CLASSIC 2000 Seismic, Magnetic, Passive Infrared, & Monitor

WHAT IS RAPTOR?

• A smart, autonomous, anti-armor/vehicle systemwhich increases lethality of its own Wide AreaMunitions and other weapon systems through thesynergistic effects of its munitions and sensors.

A Network of:• Sensors• Gateways• Munitions• Control Station

CORE RAPTOR

• Remote Employment⇒ Up to 50 Kilometers from Control

Station⇒ Delivered by Helicopter, Hand

Emplaced

• Extended Communications⇒ Multiple Ground and/or Aerial

Communication Relay

• Targets (detect, classify, track/locate,attack) – MUTIPLE TARGETS⇒ Heavy Wheeled and Tracked⇒ Light Wheeled and Tracked

An Early Operational Capabilityfor the Brigade Commander

X

III

•

GCS

Hornet PIP

30 - 50 km

ADAS

•

Gateway

A Force XXI System

Current UGS Functions/Features• Autonomous sensor networks

deployed in clusters• GPS, Compass, Radios• DSP hardware/software• Detection, Multiple Target

Tracking, Classification• Master/Slave Data Fusion• Early Warning for Munitions &

TOC• Target Info for Long Range

Shooters/Hunters

Committed to Excellence

IASFT SENSOR LAYOUT

Spesutie Island, APG, MD

N1

N2

N3

N4

North

Current UGS Systems

Light Tracked Target

DetectionM60 – 4km

Tracking4 Targets

Simultaneously

ClassificationM60 – 3km

Target ClassHeavy TrackedHeavy WheeledLight TrackedLight Wheeled

• Two Areas of development:§ Classification - Cylinder Counting Algorithm§ Template Based Approach – Using HLA information§ Statistically Enhanced using naïve Bayesian classifier

§ Tracking - Target Counting Algorithm§ Requires Enhanced Directivity Using Adaptive Beamforming§ Null – Steered Response useful§ Minimum Variance Distortionless Response

Algorithm Development for RAPTOR

RAPTOR Vehicle Classifier

WE

IGH

T A

TT

RIB

UT

E

TRACKED ATTRIBUTE

T-72

T-62

ZIL

M 60

BRDM

BTR 80

BMP

MAZ

WH

EELE

D O

R T

RA

CK

ED

LIGHT OR HEAVY

LW

HW

LT

HT

M 1

2-1/2 TM 113

V-813,600kg

Weight 17,300kg(chassis), payload15,000kg V-12

V-8 5,800 -10,525kg

V-12 44,500 kg

V-12 40,000 kg

12 Cyl52,617 kg

6 cyl 14,061 kg

V-87,000 kg

6 cyl 14,300 kg

Turbine 61,700 kg

Classification Algorithm Development

20 40 60 80 100 120 140 160 180 20030

40

50

60

70

80

90

100

28

47

56

78

94

109

112125

141

156

EFR(9) th

Frequency (Hz)

SP

L re

20u

Pa

(dB

)

• Frequency domain features– spectral content

• Harmonically associatedspectral components

• Clustered according tonumber of cylinders &target type

• Statistical propertiestabulated

• Bayesian statistics used forclassification algorithms

• The identification of HLA templates relies on the 3 tallest harmonics(e.g. 9 – 7 – 5 for T72) 6 %

20 40 60 80 100 120 140 160 180 20030

40

50

60

70

80

90

100

28

47

56

78

94

109

112125

141

156

EFR(9) th

Frequency (Hz)

SP

L re

20u

Pa

(dB

)

(8)

(7)

(6)

(5)

(4)

(3)

(2)

Signature for a HEAVY TRACKED target

Classifier Improvements for RAPTOR

“tread slap”

Cylinder Counting Algorithm

Load signaturefrom file

FFT

ObtainSpectrum

20 40 60 80 100 120 140 160 180 20030

40

50

60

70

80

90

100

2 8

47

56

78

94

1 0 9

112125

141

1 5 6

EFR(9) th

Frequency (Hz)

SP

L re

20u

Pa

(dB

)

Median Filter

Identify Peaks > 10dBCalculate

Noise Floor

Determine EFRPerform HLA

Tallest PeakGroup

HarmonicallyRelated Peaks

Retain 3Tallest Peaks

EstimateNumber ofCylinders

PerformTemplate

Match

TemplateDatabase

DatabaseCompare

Extract Features

CylinderAttribute

Discrimination Classification

Classification Algorithm Results

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Training Set All Runs Runs not in Set

Cylinder Classification over all WAM data

92

85

79

89 89 91

86

92

83

99 99

6 8 12 Turbine

Classification Algorithm Results

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1Target Identification over WAM all data sets

Training Set All Runs Runs not in Set

BMP Sov. T 2-Ton ZIL M1 HMV BTR M60

92 88 88

90

84 82

75 74

70

89

95 98 97 96 93 93 93

86 86

99 95 93

Target Counting Algorithms

• Threat tracked as a “target mass” at long ranges• Decomposed into list of individual targets at closer ranges• Target tracking maintained throughout scenario

LW #1

LW #2

HT #1

HT #2

HT #3HW #3

AOS 1 AOS 2

AOS 3AOS 4

• Preliminaries§ Requires superior bearing resolution§ MATLAB program to test beams for different

array geometries and apertures§ Try adaptive beamforming methods to check the

feasibility of assumptions made§ Nullsteering, Optimal Beamformer response is

determined as weights are obtained§ MVDR solution

Target Counting Algorithm

Target Counting Algorithm

Loadmicrophone

data from file

FFT

ObtainSpectrum

20 40 60 80 100 120 140 160 180 20030

40

50

60

70

80

90

100

2 8

47

56

78

94

1 0 9

112125

141

1 5 6

EFR(9) th

Frequency (Hz)

SP

L re

20u

Pa

(dB

)

Calculate Power in BeamMVDR

Beamformer

CurrentPower

SpectralDensity

Compare withPSD at t-1

ScenarioAssessment

Fuse Resultsfrom allsensors

Target Counting Algorithm

12345

12345

time

LW

HW

HT

Total Energy

Multiple Target Tracking & Counting

Summary

• Algorithm Development usingMATLAB/SIMULINK

• Extensive Signature Databases w/ GroundTruth

• Sensor Hardware / MATLAB models– IAS Overwatch Sensor– Wide Area Munition (WAM)Sensor

• Currently working on– Target Classification– Multiple Target Tracking