30
A concise overview | Gerald Poppinga, October 3 rd 2019 C-UAS Detection, Tracking and Intent

A concise overview | Gerald Poppinga, October 3 rd 2019

  • Upload
    others

  • View
    1

  • Download
    0

Embed Size (px)

Citation preview

Page 1: A concise overview | Gerald Poppinga, October 3 rd 2019

A concise overview | Gerald Poppinga, October 3rd 2019

C-UAS Detection, Tracking and Intent

Page 2: A concise overview | Gerald Poppinga, October 3 rd 2019

Drone on the Rise as a ThreatCivil Military

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 2The Guardian, 14 sept 2019

BBC News,27 Sept 2019

Spiegel Online, 16 Sept 2013

Reuters, 3 July2018

Al Jazeera, 10 Jan 2019

CNBC, 11 Jan 2018

CBS news, 21 Feb 2017

DefenseOne, 2 July 2019

Washington Post, 23 Sept. 2016

USA today, 7 Aug 2018New York Times, April 13 2017

Page 4: A concise overview | Gerald Poppinga, October 3 rd 2019

The Counter-UAS Challenge

Counter measures against Unmanned Aircraft Systems (UAS) are often technologically immature, scarce and expensive. Therefore research and development is required for effective and affordable counter measures for both civil and military use.

Countermeasures should consist of layered capabilities for rapid (real-time) detection, tracking, classification, identification, and neutralization of hostile UAS with minimal collateral damage.

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 4

Page 6: A concise overview | Gerald Poppinga, October 3 rd 2019

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 6

Agenda for the remainder of this Presentation

• The C-UAS protection model• Detection, Tracking and Intent• Additional considerations• Conclusion

Page 7: A concise overview | Gerald Poppinga, October 3 rd 2019

The C-UAS protection model

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 7

Page 8: A concise overview | Gerald Poppinga, October 3 rd 2019

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 8

The Netherlands’ Counter UAS Protection Model

1 • Prevention

2 • Detection/Tracking

3 • Classification/Identification/Intent

4 • Decision making

5 • Neutralization

6 • Forensics

Page 9: A concise overview | Gerald Poppinga, October 3 rd 2019

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 9

Prevention

1 • Prevention

2 • Detection/Tracking

3 • Classification/Identification/Intent

4 • Decision making

5 • Neutralization

6 • Forensics

Page 10: A concise overview | Gerald Poppinga, October 3 rd 2019

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 10

Detection, Tracking, Classification, Identification, Intent

1 • Prevention

2 • Detection/Tracking

3 • Classification/Identification/Intent

4 • Decision making

5 • Neutralization

6 • Forensics

Page 11: A concise overview | Gerald Poppinga, October 3 rd 2019

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 11

Decision Making

Options: • Real time decision support• Operator support• Interoperability• Common operational picture• Rules of engagement

1 • Prevention

2 • Detection/Tracking

3 • Classification/Identification/Intent

4 • Decision making

5 • Neutralization

6 • Forensics

Page 12: A concise overview | Gerald Poppinga, October 3 rd 2019

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 12

Neutralization

Options: • Jamming/Spoofing• Kinetic

• Capture• Collision• Destruction

• Directed Energy• High Power Microwave• ElectroMagnetic Pulse• Laser

1 • Prevention

2 • Detection/Tracking

3 • Classification/Identification/Intent

4 • Decision making

5 • Neutralization

6 • Forensics

Photo courtesy of TNO

Page 13: A concise overview | Gerald Poppinga, October 3 rd 2019

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 13

Forensics

Aspects: • Content / Log file analysis• Mode of operation research• Decryption• ...

DNA Analysis, DoD photo by Fred W. Baker III

1 • Prevention

2 • Detection/Tracking

3 • Classification/Identification/Intent

4 • Decision making

5 • Neutralization

6 • Forensics

Page 14: A concise overview | Gerald Poppinga, October 3 rd 2019

Detection, Tracking and Intent

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 14

Page 15: A concise overview | Gerald Poppinga, October 3 rd 2019

“Drone” is a collective name

Many additional differentiating factors :- type: Helicopter, Multi-rotor, Fixed

Wing,...- Materiel: plastic, metal, wood, ...- Propulsion: None (glider), Electric

Engine, Piston Engine, Turbine,...- Control/Datalink: None, ISM specific

band, 3G, 4G, 5G, other... - Navigation: None, Magnetic compass,

Camera image processing, INS, GPS, ..

- Numbers: single, multiple, swarm,..EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 15

Page 16: A concise overview | Gerald Poppinga, October 3 rd 2019

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved

Detect and Track Modalities

Various Detection/Tracking modalities:• Human observers• Acoustics• Electro Optics• ESM• Radar• Laser• ...

Many aspects influence the detect- & track-ability per modalities for each different type of drone

OnyxStar HYDRA-12 by Cargyrak, used under CC BY-SA 4.0, adapted version.

Dr.One VTOL drone

Lammergier Fixed Wing

1616

Page 17: A concise overview | Gerald Poppinga, October 3 rd 2019

Detection & Tracking – Human Observers

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 17

Britain's Home Front 1939 - 1945- Roof Spotters An acoustic aircraft detection apparatus in the USA in 1921

Depends on many aspects:• UAV: Size, Shape, Colors,

Sound,...• Observer: Eyesight, Hearing,

Search technique, ...• Conditions: Weather, Time of

Day, Time of Year, ...Training might help to improve detection qualities

ROTARY WING FIXED WINGNano ( < 0.5 kg) 100 100Micro (0.5 - 2 kg) 200 500Mini (2 - 20 kg) 300 1000Small (20 - 150 kg) 800 1200Notional maximal visual detection ranges (m)** Based on a literature study, assessments and experience, for a generic

“representative” type of drone under “normal” conditions.

Page 18: A concise overview | Gerald Poppinga, October 3 rd 2019

Parrot Anafi(320 gr.) (100 m)foto by KlausFoehl, used under CC BY-SA 4.0, clipped version.

Yuneec Typhoon H(1.7 Kg) (200 m)foto by SkylarkCoder, used under CC BY-SA 4.0, clipped version.

DJI M600pro (MTOW 15.5 kg) (300m))foto by Mc clapurhands, used under CC BY-SA 4.0, clipped version.

1

2

3

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 18

Human Observers – Notional Visual Detection Ranges

The notional maximal visual detection ranges visualized in Google Earth

Example systems

Page 19: A concise overview | Gerald Poppinga, October 3 rd 2019

Detection & Tracking – Acoustics, E/O and ESM

Acoustic• Single/multiple microphone(s)• Acoustic radarElectro Optics (E/O)• IR • Visual Range• UVElectromagnetic Spectrum Monitoring• specific bandS (Wi-Fi, RC, ...)• 3G, 4G, 5G• other frequencies

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 19

Acoustics Electro Optics ESMNano ( < 0.5 kg) 150 100 1500Micro (0.5 - 2 kg) 250 250 1500Mini (2 - 20 kg) 350 400 1500Small (20 - 150 kg) 500 1000 1500Notional maximal detection ranges (m)*

Depending on many conditions, e.g. weather, time of day/year, EMS noise levels, transmission power, # of transmitters

* Based on a literature study, assessments and experience, for a generic “representative” type of drone under “normal” conditions.

Page 20: A concise overview | Gerald Poppinga, October 3 rd 2019

Notional Detection Ranges For A Micro Drone

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 20

Example micro drone system

DJI Mavic Pro(~750 gr) foto by thoroughlyreviewed.com, used under CC BY 2.0, clipped version.

1. Acoustics& E/O (250 m)2. ESM(1500 m)*, **

* If the signal is know, the maximal detection range can be much larger** if the drone flies autonomously, there might not be a RF signature

1

2

Page 21: A concise overview | Gerald Poppinga, October 3 rd 2019

Detection & Tracking – Radar and Laser

• Active Radar– Pulse Doppler– Continuous-Wave– Active Phased Array– MIMO Phased Array or UWB– Other

• Passive Radar• Laser Enabled

– LIDAR/LADAR

Active RadarROTARY

WINGActive RadarFIXED WING Laser Enabled

Nano ( < 0.5 kg) 3000 6000 300Micro (0.5 - 2 kg) 3000 6000 300Mini (2 - 20 kg) 3000 6000 300Small (20 - 150 kg) 10000 10000 2000

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 21

Some remarks:• Radar signatures of drones can be very small

and are therefore often ignored (like birds) • A secondary radar requires a drone to have a

transponder in order for the radar to be able to detect the drone

Notional maximal detection ranges (m)** Based on a literature study, assessments and experience, for a generic “representative” type of drone under “normal” conditions.

Page 22: A concise overview | Gerald Poppinga, October 3 rd 2019

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 22

The Detect and Track Optimization Puzzle

Many aspects influence the detectability/track-ability • Drone type, shape, materials, ...• Detection/tracking modalities• Landscape & Buildings• Weather conditions• ...

Goal: to maximally mitigate drone related risks for an airport with respect to flight safety and economical concerns, with available and affordable means, within the operational constraints.

+

Page 23: A concise overview | Gerald Poppinga, October 3 rd 2019

E.g. E/O: Various orientations, weather conditions, camera blurs, etc.

DNN’s are a Black box• Generalization• Impressive results on test sets• Results on real life data may vary• Potentially susceptible to Adverserial

Attacks

Classification/Identification

Sensor Fusion Deep Neural Networks (DNNs)

Requires large data/training sets: images, sound (profiles), radar (Doppler) signatures, ESM characteristics, etc.

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 23

Visual

Infrared

Supervised Learning

Classifier

Page 24: A concise overview | Gerald Poppinga, October 3 rd 2019

Intent

Correlate specific characteristics with other data (e.g. Classification/Identification data) and detect Discrepancies.

Options: • Object/payload recognition• Flight Path Analysis• ...

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 24

From: Drone classification by payload-speed analysis, from: Conceptual study of an Anti-Drone Drone through the coupling of design process and interception strategy Simulations, Onera 2016

Fox News, January 25, 2017

Page 25: A concise overview | Gerald Poppinga, October 3 rd 2019

Additional Considerations and Conclusion

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 25

Page 26: A concise overview | Gerald Poppinga, October 3 rd 2019

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 26

No Silver Bullet - Consideration

Many different partial solutions, cherry picking might speed things up: • Sensor X of Vendor C• Sensor Y of Vendor D• Intervention Solution P of Vendor H• Intervention Solution Q of Vendor I

Required: a Common (Open Source) Backbone Inspired by the Robot Operating System (ROS): “a collection of tools, libraries, and conventions that aim to simplify the task of creating complex and robust C-UAS behavior across a wide variety of C-UAS elements”

10 oz silver bullet by Money Metals,Used under CC-BY-2.0, cropped version

The Dutch C-UAS Nucleus’ view on C-UAS

Page 27: A concise overview | Gerald Poppinga, October 3 rd 2019

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 27

Test (operationally) before you buy

Test for some of the operationally highly relevant aspects, e.g.:• Detection probability

– False Negatives– False Positives

Challenge: - Simulation of representative

malicious behavior requires support of the responsible CAAand National Telecom Authority

- Realistic testing could result in collateral damage

Page 28: A concise overview | Gerald Poppinga, October 3 rd 2019

There’s more - DOTMLPFI-P

• Doctrine• Organization• Training• Material• Leadership and Education• Personnel• Facilities• Interoperability• Policy

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 28

Tip of the iceberg by Uwe KilsUsed under CC BY-SA 3.0, original version

Page 29: A concise overview | Gerald Poppinga, October 3 rd 2019

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 29

Conclusions & Recommendations

There’s no silver bullet, not expected for quite some time. Each location/setting requires a complex detection means puzzle to be completed for optimal risk mitigation

Much to do by many• Science – Fill in the Gaps in Knowledge• Developers – Keep Developing & Improving C-UAS Systems• Overarching Party – Create a common (open source) Backbone • Operations - Make clear what you do and do not need, help by facilitating operational tests and share the results

1• Prevention

2• Detection/Tracking

3• Classification/Identification/Intent

4• Decision making

5• Neutralization

6• Forensics

Fully engaged in C-UAS!

Page 30: A concise overview | Gerald Poppinga, October 3 rd 2019

NLR AmsterdamAnthony Fokkerweg 21059 CM AmsterdamThe Netherlands

p ) +31 88 511 31 13 e ) [email protected] i ) www.nlr.org

NLR MarknesseVoorsterweg 318316 PR MarknesseThe Netherlands

p ) +31 88 511 44 44 e ) [email protected] i ) www.nlr.org

Fully engagedRoyal Netherlands Aerospace Centre

EUROCONTROL "Drones’ incursions at the airports" workshop - © NLR 2019 All Rights Reserved 30