Tutorial: DOI. No. 10.1109/MAES.2020.

3015537

Counter-Unmanned Aircraft System(s) (C-UAS):

State of the Art, Challenges, and Future Trends
Jian Wang, Yongxin Liu, and Houbing Song, Embry-Riddle Aeronautical
University, Daytona Beach, FL 32114 USA

Unmanned aircraft systems (UAS), or unmanned C-UAS technologies, thus fostering a research
aerial vehicles, often referred to as drones, have been community committed to the safe integration of UAS
experiencing healthy growth in the United States and into the airspace system.
around the world. The positive uses of UAS have the
potential to save lives, increase safety and efficiency,
and enable more effective science and engineering INTRODUCTION
research. However, UAS are subject to threats
An unmanned aircraft system (UAS) is an unmanned aircraft
stemming from increasing reliance on computer and (an aircraft that is operated without the possibility of direct
communication technologies, which place public human intervention from within or on the aircraft) and asso-
safety, national security, and individual privacy at ciated elements (including communication links and the
risk. To promote safe, secure, and privacy-respecting components that control the unmanned aircraft) that are
UAS operations, there is an urgent need for innovative required for the operator to operate safely and efficiently in
technologies for detecting, tracking, identifying, and the airspace system. Over the last five years, UAS, or
mitigating UAS. A Counter-UAS (C-UAS) system is unmanned aerial vehicles (UAVs), often referred to as
defined as a system or device capable of lawfully and drones, have been experiencing healthy growth in the United
safely disabling, disrupting, or seizing control of an States and around the world [1]. According to the Federal
unmanned aircraft or UAS. Over the past five years, Aviation Administration (FAA) aerospace forecast fiscal
significant research efforts have been made to detect, years 2019–2039, the model UAS fleet is set to grow from
and mitigate UAS: detection technologies are based on the present 1.25 million units to around 1.39 million units by
acoustic, vision, passive radio frequency, radar, and 2023 and the nonmodel UAS fleet is set to grow from the
data fusion; and mitigation technologies include present 277 000 aircraft to over 835 000 aircraft by 2023 [2].
physical capture or jamming. In this tutorial, we The positive uses of UAS have the potential to save lives,
provide a comprehensive survey of existing literature increase safety and efficiency, and enable more effective sci-
in the area of C-UAS, identify the challenges in ence and engineering research [3]. These uses may include
countering unauthorized or unsafe UAS, and evaluate modelers experimenting with small UAS, performing
the trends of detection and mitigation for protecting numerous functions including aerial photography and per-
against UAS-based threats. The objective of this sonal recreational flying, commercial operators experiment-
tutorial is to present a systematic introduction of ing with package and medical supply delivery, and
providing support for search and rescue missions.
While the introduction of UAS in the airspace system
Authors’ current addresses: Jian Wang, Yongxin Liu,
has opened up numerous possibilities, UAS can also be
and Houbing Song are with the Security and Optimi-
zation for Networked Globe Laboratory (SONG Lab, used for malicious schemes by terrorists, criminal organi-
www.SONGLab.us), Department of Electrical Engi- zations (including transnational organizations), and lone
neering and Computer Science, Embry-Riddle Aero- actors with specific objectives. UAS-based threats stem
nautical University, Daytona Beach, FL 32114 USA from increasing reliance on computer and communication
(e-mail: [email protected]; [email protected]. technologies, placing public safety, national security, and
edu; [email protected]). individual privacy at risk [4].
Manuscript received November 15, 2019, revised
March 19, 2020; accepted July 23, 2020, and ready for
Safety: Unsafe UAS operations involve operating
publication August 7, 2020. UAS near other aircraft, especially near airports;
Review handled by William Dale Blair. over groups of people, public events, or stadiums
0885-8985/21/$26.00 ß 2021 IEEE full of people; near emergencies such as fires or

4 IEEE A&E SYSTEMS MAGAZINE MARCH 2021

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
 Image credit: Image licensed by Ingram Publishing

hurricane recovery efforts; or under the influence of owners and operators may want to take, include: 1) the pilots
drugs or alcohol [5], [6]. Reports of UAS sightings of the UAS are not able to receive commands from airspace
from pilots, citizens, and law enforcement have authorities; 2) the pilots of the UAS commonly take actions
increased dramatically over the past five years [7]. according to video streams and global positioning system
A recent notable UAS incident was Gatwick Airport (GPS) trajectories; 3) the communication links between
UAS incident: Between December 19 and 21, 2018, pilots and UAS are vulnerable to interference.
hundreds of flights were cancelled at Gatwick Air- Over the past five years, a lot of research efforts have been
port near London, England, following reports of made to detect, track, identify, and mitigate UAS: detection
UAS sightings close to the runway. The reports technologies are based on acoustic [15], vision [16], passive
caused major disruption, affecting approximately radio frequency (RF) [17], radar [18], and data fusion [19];
140 000 passengers and 1000 flights [8]. and mitigation technologies include physical capture (contain-

Security: Unsecure UAS operations involve operat- ment netting) [20], jamming (RF command and control (C2)
ing UAS over designated national security sensitive jamming and spoofing, or GPS jamming and spoofing) [21],
facilities, such as military bases, national landmarks and destruction (RF C2 intercept and control) [22]. However,
(such as the Statue of Liberty, Hoover Dam, Mt. these efforts are not mature: lack of scalability, modularity, or
Rushmore), and certain critical infrastructure (such affordability. Innovative technologies toward relatively
as nuclear power plants), among others [9], [10]. mature scalable, modular, and affordable approaches to detec-
tion and negation of UAS are desired.

Privacy: Privacy-invading UAS operations involve In this article, we provide a comprehensive survey of
operating UAS with their camera on when pointing existing literature in the area of UAS detection and negation,
inside a private residence [4], [11], [12]. identify the challenges in countering adversary UAS, and
To promote safe, secure, and privacy-respecting UAS evaluate the trends of detection and negation for protecting
operations, there is an urgent need for innovative technol- against UAS-based threats. The objective of this tutorial is
ogies for detecting, tracking, identifying, and mitigating to present a systematic introduction of C-UAS technologies,
UAS. A counter-UAS (C-UAS) system is defined as a sys- thus fostering a research community committed to the safe
tem or device capable of lawfully and safely disabling, integration of UAS into the airspace system.
disrupting, or seizing control of an unmanned aircraft or The remainder of this article is structured as follows.
UAS [13]. Typically such a system is comprised of two The “Background” section presents various UAS-based
subsystems: one for detection and the other for mitiga- threats and introduces restrictions that commonly affect
tion [5], [14]. The ideal UAS detection subsystem, UAS flights. “State of the Art UAS Detection” and “State
will detect, track, identify an unmanned aircraft, or UAS, of the Art Mitigation” present the state of the art detection
have a small footprint, and support highly automated oper- and mitigation, respectively. “Challenges in UAS Detec-
ations. The ideal UAS mitigation subsystem [13] will law- tion and Mitigation” identifies the challenges in countering
fully and safely disable, disrupt, or seize control of an unauthorized or unsafe UAS. “Future Trends” evaluates
unmanned aircraft or UAS [13], while ensuring low collat- the trends of detection and mitigation for protecting against
eral damage and low cost per engagement. UAS-based threats. Finally, we conclude the tutorial.
Due to the fact that UAS are aircraft without a human
pilot onboard that are controlled by an operator remotely or
programmed to fly autonomously, protecting against UAS- BACKGROUND
based threats is very challenging. The challenges, which In this section, we discuss why we need UAS detection and
UAS can present to critical infrastructure and several courses mitigation, different types of UAS-based threats, and the
of action that law enforcement and critical infrastructure common airspace restrictions applicable to UAS flights.

MARCH 2021 IEEE A&E SYSTEMS MAGAZINE 5

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
Counter-Unmanned Aircraft System(s) (C-UAS): State of the Art, Challenges, and Future Trends

from which we could observe that unsafe and unautho-

rized UAS operations have been increasing dramati-
cally. It is interesting that the most UAS sightings
occur in the summer months. The UAS sightings in
each state are shown in Figure 2, from which we could
observe that unsafe and unauthorized UAS operations
occur in most populated US states.

UAS-BASED THREATS
Figure 1.
Temporal distribution of UAS sightings. In this tutorial, we classify the UAS-based threats into
three categories: public safety, national security, and
individual privacy, as shown in Table 1. UAS-based
public safety threats are due to operating UAS near
REPORTED UAS SIGHTINGS
other aircraft, especially near airports; over groups of
Over the past five years, reports of UAS sightings from people, public events, or stadiums full of people [6];
pilots, citizens, and law enforcement have increased near emergencies such as fires or hurricane recovery
dramatically. Each month the FAA receives more than efforts; or under the influence of drugs or alcohol;
100 such reports. The monthly UAS sightings between UAS-based national security threats are due to operat-
January 2014 and December 2019 are shown in Figure 1, ing UAS over designated national security sensitive

Figure 2.
Spatial distribution of UAS sightings.

Table 1.

Classification of UAS-Based Threats

Threats Threatened entities Threat mode Consequences

Safety Human, facilities, and Collisions, indirect hazards, Injuries or damage of properties.
high value targets and controlled attacks
Security High value targets Aerial imaging and posterior Disclosure of sensitive information
reconstruction and national security issues
Privacy Human Aerial imaging or real-time Privacy invasion
video stream

6 IEEE A&E SYSTEMS MAGAZINE MARCH 2021

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
 Wang et al.

Table 2.

Some Recent UAS-Based Threats

Event Description Time Category Location Influence

Drone There were 54 drone April 10, Safety Mercedes- Causing special
intrusion in incursions in Super Bowl [23] 2019 Benz attention of law
nonflight Stadium, enforcement
zone USA agencies
Drones Two drones crashed into April 21, Safety Heathrow RAF have spent £5
threatening landing zone when a B787-9 2019 Airport, million to prevent
aviation was landing [24] UK future attacks
safety
Drones Spy drones hacked wireless July 29, Security Las Vegas, Loss of information
stealing networks with software- 2011 USA security.
information defined radios [25]
Drones Border patrol spotted drones April 19, Security US- Increasing border
threatening trying to help migrants 2019 Mexico protection difficulty
national illegally enter America [26] border,
security USA
Drones Illegal activities obtained October Privacy Orlando, Police taking part in
threatening individual’s privacy using 28, 2018 USA investigation
personal drones [27]
privacy
Drones Residents disturbed by February Privacy Upper Privacy invasion
threatening peeping drone outside 23, 2018 Hutt, New concerns
personal bedroom window [28] Zealand
privacy

facilities, such as military bases, national landmarks, SECURITY THREATS

and certain critical infrastructure, among others [9],
UAS-related national security threats may include the
[10]; UAS-based privacy threats are due to operating
following [30].
UAS with their camera on when pointing inside a pri-
vate residence [4], [11].
Weaponized or Smuggling Payloads: Depending on
power and payload size, UAS may be capable of
transporting contraband, chemical, or other explo-
SAFETY THREATS sive/weaponized payloads.

The unmanned nature of UAS operations raises two
Prohibited Surveillance and Reconnaissance: UAS
unique safety concerns that are not present in manned- are capable of silently monitoring a large area from
aircraft operations: the pilot of the small UAS, who is the sky for nefarious purposes.
physically separated from it during flight, may not
Intellectual Property Theft: UAS can be used to per-
have the ability to see manned aircraft in the air in form cyber crimes involving theft of trade secrets,
time to prevent a mid-air collision, and the pilot of the technologies, or sensitive information.
small UAS could lose control of it due to a failure of
the communications link between the small UAS and
the pilot’s handset for controlling the UAS [29]. Safety
PRIVACY THREATS
risks related to the use of UAS include the potential UAS-based privacy threats lie in intentional disruption or
for unintentional collisions between a small UAS and harassment. UAS may be used to disrupt or invade the pri-
a manned aircraft or other objects, causing damage to vacy of other individuals [30].
property, or injury, or death to persons [29]. Operating Some recent UAS-based threats are given in Table 2,
UAS around airplanes, helicopters, and airports is dan- which shows the time and location of threat occurrence,
gerous and illegal. threat category, and corresponding consequences.

MARCH 2021 IEEE A&E SYSTEMS MAGAZINE 7

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
Counter-Unmanned Aircraft System(s) (C-UAS): State of the Art, Challenges, and Future Trends

Table 3.

Airspace Restrictions Applicable to UAS by FAA (March 2020)

Stadiums Operations are prohibited within a radius of three nautical miles. Operations
are prohibited starting one hour before and ending one hour after scheduled
time of important events.

Airports 1. One must have a Remote Pilot Certificate and get permission from Air
traffic control (ATC). 2. Model aeroclub organization must notify the airport
operator and control tower to fly within 5 miles. 3. Public entity (law
enforcement or government agency) may apply for special permission.
Security sensitive Operations are prohibited from the ground up to 400 feet above ground level.
airspace
Capital areas Varying according to the policies of government, in Washington, DC.,
airspace is governed by a Special Flight Rules Area (SFRA), UAS operations
are restricted within a 30-mile radius of Ronald Reagan Washington National
Airport (15-mile inner ring: prohibitted without specific permission from
FAA; 15 to 30 miles outer ring: registered, light and small UAV can fly lower
than 400 feet and visual range in clear weather).
Restricted or Special Use Certain areas where drones and other aircraft are not permitted to fly without
Airspace special permission, or where limitations must be imposed for any number of
reasons.
Temporary Flight A TFR defines a restricted airspace due to a hazardous condition or specific
Restriction (TFR) events. List of TFR can be found at https://tfr.faa.gov/tfr2/list.html
Emergency and Rescue FAA prohibits drones over any emergency or rescue operations. It’s a federal
Operations crime to interfere with firefighting aircraft regardless of whether restrictions
are established.

over the past five years, as shown in Figures 1 and 2.

AIRSPACE RESTRICTIONS APPLICABLE TO
Therefore, there is an urgent need for promoting safe,
UAS FLIGHTS secure, and privacy-respecting UAS operations. We envi-
sion that an integrated system capable of detecting and
In the United States, there are several types of airspace
negating UAS will be essential to the safe integration of
restrictions that commonly affect UAS flights [31], as shown
UAS into the airspace system. Such a system needs novel
in Table 3. From the table, the most stringent restrictions are
technologies in the following two main areas.
security sensitive airspace restrictions which prohibit UAS
operations from the ground up to 400 feet above ground level,
Detection: The ideal UAS detection will detect,
and apply to all types and purposes of UAS flight operations; track, and identify an unmanned aircraft or UAS,
for stadiums and sporting events, restrictions are valid during have a small footprint, support highly automated
gathering of the crowds; Washington DC has the largest spa- operations and location functions.
tial restriction circle (30 miles).

Mitigation: The ideal UAS mitigation system will
The FAA established a series of rules and regulations
lawfully and safely disable, disrupt, or seize con-
which apply to UAS operations, based on the type of UAS
trol of an unmanned aircraft or UAS, while
flier: recreational fliers and modeler community-based
ensuring low collateral damage and low cost per
organizations, certificated remote pilots including com-
engagement.
mercial operators, public safety and government, and edu-
cational users [1]. UAS that weigh more than 0.55 pounds A survey of UAS detection and mitigation is the first
must be registered with the FAA. In addition, flying UAS step in leveraging communications and signal processing
that are less than 55 pounds for work or business requires to develop low footprint UAS detection solutions, in terms
remote pilot certificates [1]. of size, weight, power, and manning, as well as varied and
Although the FAA has established guidelines and reg- low collateral damage UAS mitigation techniques and
ulations, and reports of UAS sightings from pilots, citi- effectors, towards safe integration of UAS into the air-
zens, and law enforcement have increased dramatically space system.

8 IEEE A&E SYSTEMS MAGAZINE MARCH 2021

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
 Wang et al.

Figure 3.
Evolution of UAS detection technologies.

proposed to deploy acoustic sensor array which acquires

STATE OF THE ART UAS DETECTION
the sound of engine in UAS. The acoustic sensor array
Since 2014, five technologies have been proposed for consists of 24 custom-built microphones which locate and
UAS detection, including acoustic, vision, passive RF, track the UAS collaboratively. They calibrated each sen-
radar, and data fusion. We summarized the evolution of sor with time delay of arrival (TDOA) and predicted the
UAS detection technologies in Figure 3, where different UAS flight path with beamforming. They could track the
technologies are ranked based on the popularity for each flight path well but this approach could not work well in a
year. From Figure 3, we know that data fusion approaches large scale space, and the accuracy of the sensor is depen-
have been the most popular technology for UAS detection dent seriously on calibration [34]. Two arrays consists of
while acoustic-based approaches have been the least popu- four microphone sensors to improve the capability of the
lar. In this section, we will introduce each UAS detection UAS localization. Due to the multipath effect, they pro-
technology and discuss its advantages and disadvantages. vided a Gauss prior probability density function to
improve the TDOA estimation. Their arrays could be
deployed in specific area efficiently and achieved good
ACOUSTIC-BASED UAS DETECTION performance to track the UAS. However, their system is
Acoustic-based UAS detection leverages acoustic sensors not stable when it works for a long time [35]. Advanced
to capture the sound of UAS, identify and track the UAS acoustic cameras are leveraged to detect and track the
with audio. Acoustic sensor arrays, which are deployed UAS. To be specific, they used 2 to 4 acoustic cameras to
around the restricted areas, record the audio signal period- capture the strength distribution of sound and fused the
ically and deliver the audio signal to the ground stations. strength distribution to compute the location of the UAS
The ground stations extract the features of the audio signal indoors and outdoors [36]. An audio-assisted camera array
to determine whether the UAS are approaching. is deployed to detect the UAS, which captured the video
Conventionally, after receiving the audio signal of and audio signals at the same time and classified the object
UAS, the power spectrum or frequency spectrum will be with histogram of oriented gradient feature and mel fre-
analyzed to identify the UAS. Vil
ımek et al. adopted the quency cepstral coefficients feature [37].
linear predictive coding to distinguish the sound of UAS Different from the above conventional methods, sig-
engine from the sound of car engine but the performance nificant research efforts have been made to leverage
is subject to weather conditions [32]. Kim et al. designed machine learning (ML) to classify the UAS from audio
a real-time UAS sound detection and analysis system data. Support vector machine (SVM) is implemented to
which could acquire the real-time sound data from the sen- analyze the mid term signal of UAS engine and con-
sor and recognize the UAS [15]. Jang et al. applied the structed the signal fingerprint of UAS. Their results
Euclidean distance and scale-invariant feature transform showed that the classifier could precisely distinguish the
(SIFT) to distinguish the UAV engine sound from the UAS in some scenarios [38]. An approach is proposed to
background sound and demonstrated their effectiveness, transform the detection of the presence of UAS to a binary
even though the power spectrum of the noise is larger than classification problem, and used Gaussian mixture model
that of the UAS sound. However, in practice, their proc- (GMM), convolutional neural network (CNN), and recur-
essing efficiency is poor [33]. rent neural network to detect UAS. Their results showed
Due to light weight, low-cost and easy assembly, that could work well with the short input signal within
acoustic sensors could be used to construct acoustic 240 ms [39].
acquiring array and deployed in the target area to locate The current acoustic-based UAS detection technolo-
and track the trajectories of the UAS. An approach is gies can recognize and locate the UAS precisely to meet

MARCH 2021 IEEE A&E SYSTEMS MAGAZINE 9

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
Counter-Unmanned Aircraft System(s) (C-UAS): State of the Art, Challenges, and Future Trends

to identify UAS operating with unknown telemetry

protocols. Therefore, some researches focus on the use
of the coupled kinetics motion patterns of flying UAS
from their radio signal to specify their presence.
In [42] and [43], Matthan, a cost-effective and passive
RF-based UAS detection system was introduced. Their
system detected the presence of UAS by identifying
the unique signatures of its coupled vibration and shift-
Figure 4. ing patterns in the transmitted wireless signals. The
Categorization of passive RF approaches. joint detector integrated evidence from both a fre-
quency-based detector that indicated UAS body vibra-
the accuracy requirement of UAS detection. However, the tion as well as a wavelet-based detector that captured
nature of acoustic approaches limits the deployment and the sudden shifts of drone frame by computing wave-
detection of UAS in a large scale. ML presents a tremen- lets at different scales from the temporal RF signal. A
dous opportunity of integrating with the acoustic sensing similar approach was also presented in [44].
into acoustic-based UAS detection for improved UAS Localization is also an important part of passive RF-
detection performance. based UAS detection. In [45], the authors divided the
complete detection procedure for UAS detection into
RF spectrum sensing and the direction of arrival (DoA)
PASSIVE RF-BASED DETECTION estimated. The work in [46] presented a useful experi-
UAS usually maintain at least one RF communication ment, which used commercial off the shelf field pro-
data link to its remote controller to either receive control grammable gate array (FPGA) based SDR system for
commands or deliver aerial images. In this case, the spec- detecting and locating small UAS. Their results demon-
tral patterns of such transmission are used as an impor- strated that it is possible to develop a UAS detection
tant evidence for the detection and localization of UAS. system capable of detecting small UAS with error of
Different passive RF technologies are shown in Figure 4. 50–75 m using commercial FPGA-based SDR system.
In most cases, software-defined radio (SDR) receivers are The SDR system with optimized clock synchronization
employed to intercept the RF channels. would radically reduce the measurement error in dis-
To utilize the spectrum patterns of UAS, in [40], an tance. How to implement robust localization algorithms
artificial neural networks (ANN) detection algorithm for a with acceptable accuracy on ubiquitous hardware is still
UAS RF signal was proposed which employs three signal an open problem in passive RF detection of UAS. How
features: improved slope, improved skewness, and to deploy such passive RF-based systems on the ground
improved kurtosis. It was shown that the proposed algo- stations or other UAS platforms is another open prob-
rithm based on ANN outperforms other recognition tech- lem in the detection of UAS.
nologies of the improved slope, skewness, and kurtosis of
signal spectrum.
VISION-BASED UAS DETECTION
Data traffic patterns are also an important feature to
specify UAS. In [41], a UAS detection and identification Vision-based UAS detection technologies mainly focus on
system, which utilized commercial off-the-shelf hardware image processing. Videos and cameras are adopted to cap-
to passively listen to the wireless signal between UAS ture the images of trespassing UAS. The ground stations
and their controllers for packet transmission characteris- figure out the appearance of UAS from the videos and pic-
tics, was proposed. They mainly extracted the packet tures with computational methods. Conventional methods
length distribution of UAS and evaluated the prototype mainly rely on the methods of image segmentation. The
system with three types of UAS. Their experiment results differential of UAS and environment in images is used to
demonstrated the feasibility of using the data frame determine whether the restricted areas have the UAS. A
length to identify different UAS within 20 s. The increas- vision-based UAS detection approach is presented in [47],
ing amount of commercial UAS, using WiFi as control which could separate the UAS from background effi-
and first person view (FPV) video streaming protocol, ciently. Similar work was reported in [16] and[48] . The
motivated the method in [17], a UAS detection approach common challenges for their approaches are how to sepa-
based on WiFi fingerprint. The method identified the rate UAS from background images and how to distinguish
presence of unauthorized UAS in a nearby area by moni- UAS from flying birds. Typical vision-based UAS detec-
toring the data traffic. tion technologies are summarized in Figure 5.
Data traffic based methods or pure spectrum pattern In contrast, state-of-art image segmentation methods
methods depends highly on the telemetry protocol or make use of neural networks to directly identify the
RF front-ends of UAS. These methods may not be able appearance of UAS. An approach leverages the thermal

10 IEEE A&E SYSTEMS MAGAZINE MARCH 2021

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
 Wang et al.

applied to capture the rotating frequency of the propeller

to distinguish the UAS from birds efficiently [57].
Currently, the vision-based approaches can be imple-
mented in some specific scenarios to recognize the fea-
tures of UAS from the environment. The evolution of
deep neural networks stimulated the processing of image
processing which could have multiple positive effects on
the UAS detection in vision field. The real time attempts
showed that the vision approaches have the potentials of
Figure 5. efficiency. However, how to implement the recognized
Categorization of vision-based approaches. algorithms in multiple and variable environments is chal-
lenging. The novel approaches are supposed to be robust,
camera to detect the UAS and neural network to identify adjustable, and precise. The vision-based approaches need
the UAS [49]. An outstanding research is presented to be robust to the quick variation of the environment. The
in [50]. A lightweight and fast algorithm which could image distortion caused by weather change could be miti-
operate on embedded system (Nvidia Jetson TX1) and gated by the multiple level image processors which cap-
identify the UAS in movement. ture the features of UAS in different spectrum. The
A real-time vision-based UAS detection system is mobility of UAS poses a challenge to the vision-based
designed which is based on two vision processing plat- approaches, i.e., the images are supposed to be captured
forms: FPGA-based platform, which can operate below and recognized in different levels of mobility of UAS.
10 W (i.e., power saving), and graphics processing unit The bioinspired robots are limited in the detection accu-
(GPU)-based platform, which is able to process more racy, which poses the risk of distinguishing of UAS and
frames. However, for FPGA, it is impossible to change birds mistakenly. The enhanced accuracy of the distin-
algorithms in real time [51]. Muhammad et al. compared guishing can improve the efficiency of UAS detection and
different CNNs’ performance in detecting UAS, and their mitigation greatly.
results showed that the visual geometry group (VGG 16)
network with Faster R-CNN achieved outstanding perfor-
mance [52]. An approach is proposed to combine different RADAR-BASED UAS DETECTION
pictures to generate synthetic images to extend the image Radars have several advantages in detecting airborne
data set to train the CNN to enhance the performance of objects compared with other sensors in terms of day and
the UAS detection [53]. night operating capability, weather independency, and
Birds are a serious factor which downgrades the ability to measure range and velocity simultaneously.
identification of UAS from the images. Significant However, regular radar systems focus on fighting air tar-
research efforts have been made to use CNNs to gets of medium and large size with radar cross-section
enhance the identification of UAS. Survey about the larger than 1 m2 , which makes it infeasible to detect
challenges of detection of UAS and birds is presented small-size and low-speed UAS in [18] and [58]. It is diffi-
in [54]. The survey concluded that the neural network cult to detect UAS due to slow speed, since Doppler proc-
algorithms are promising in identifying UAS and birds. essing is typically used. Therefore, efforts are in needed to
It compared policy-based approaches and neural net- either develop new radar models or increase the detective
work based algorithms for the recognition of birds and resolution of conventional systems. In this section, we
UAS using datasets of videos and pictures. The results will discuss three categories of radar-based UAS detection
showed that the neural network based approaches can technologies: Active detection, passive detection, and pos-
outperform the policy-based approaches over 100 times terior signal processing.
in terms of accuracy and efficiency. A UAS detection
framework is presented in [55], which is based on video
streams and classified the objects into different types
ACTIVE DETECTION
with CNN. The work mainly focused on distinguishing Typically, there are two ways to increase the resolution of
the birds and UAS in different scenarios. conventional radar detection systems for UAS surveillance:
At the same time, some attempts were made to apply utilizing higher frequency carriers and using multiple input–
infrared cameras to identify the UAS. Infrared sensors are multiple output (MIMO) beamforming radio front-ends, as
leveraged to detect small variations of UAS in heat to shown in Figure 6.
identify the UAS. The drawback of this approach is that To utilize shorter wave length, in [59] and [60], X-
the heat from batteries has significant effects on result band and W-band frequency modulated continuous
detection [56]. Different from other research on classify- wave (FMCW) radars are designed for UAS detection.
ing the frame of images, dynamic vision sensors are Their solutions use bistatic antenna and finally convert

MARCH 2021 IEEE A&E SYSTEMS MAGAZINE 11

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
Counter-Unmanned Aircraft System(s) (C-UAS): State of the Art, Challenges, and Future Trends

Figure 6. Figure 7.
Categorization of active radar-based approaches. Categorization of passive radar-based approaches.

received signals into digital quadrature stream for pos- The apparent drawback of active radars is that they
terior processing. The feasibility of using ultra wide- need specially designed transmitters which might not be
band signals with 24 GHz carrier was demonstrated easy to deploy and are vulnerable to antiradioactive attacks.
in [61]. The selection of carrier frequency for UAS
detection radar should be higher than 6 GHz (K-band)
in [62] and [63].
PASSIVE RADAR
Other approaches utilize multiple antennas to form Passive radars do not require specially designed transmit-
MIMO front-ends. The benefit of such approach lies in ter. Existing radioactive sources such as cellular signals
its applicability to radar system with lower carrier can be leveraged to illuminate the space. In this section,
frequencies. The work in [64] demonstrated the detection we classify passive radars into two categories: single sta-
of a small hexacopter using 32 by 8 element L-band tion passive radar and distributed synthetic passive radar,
receiver array, which achieved good detection sensitivity as shown in Figure 7.
against micro-UAS. Similar research was presented

Single Station Passive Radar: This type of passive
in [65], a ubiquitous FMCW radar system working at
radar exploits only one illumination source. The
8.75 GHz (X-band) with PC-based signal processor is pre-
variation of received signals can be analyzed to
sented. The results indicated that it has the capability to
specify the appearance of UAS. In [74], a WiFi-
detect a micro-UAS at a range of 2 km with an excellent
based passive radar was presented for the detection
range-speed association. In [66], a Ka-band radar system
and two-dimensional (2-D) localization of small air-
which uses 16 transmit and 16 receive antennas to form
craft. Obviously this is the most direct adaptation of
256 virtual antenna elements is presented. Their experi-
active radars.
ment proves that even in a nonstationary clutter environ-
ment, the UAS could be clearly detected at a range of about
Distributed Synthetic Passive Radar: Distributed
150 m. Similar work was presented in [67]. Another impor- station leverages the existing telecommunication
tant characteristic of MIMO systems is that they generate infrastructures as illumination sources to enhance the
large quantity of data for further processing in [68]. The UAS detection. There are mainly two approaches:
authors use the concept of data cube and classifier to assure cellular system based solutions and digital video
the presence and location on an incoming UAS. A simpli- broadcasting (DVB) system-based solutions.
fied approach, multiple input single output, was utilized for
Cellular system based passive radars: An
UAS detection in [69]. approach is proposed to enhance the detection
Noise radar is considered to be an efficient way to system which could locate and track the UAS
detect the slow moving UAS and its benefit is that UAS by using reflected global system for mobile
can be detected by using simple antenna components and communications (GSM) signal [75]. An
lower carrier frequency. In [70] and [71], the feasibility of approach is proposed to receive the 3 G cellular
using random sequence radar for UAS detection was dem- reflecting signal from UAS for track UAS. He
onstrated in sub X-band, and their results indicate that leveraged the Doppler features of 3 G cellular
such radar can be the future of cost-efficient UAS detec- signal to monitor the target area and tracked the
tion solutions. trajectory of UAS. The results showed that
The advances in computation enable another radar, the UAS could be tracked obviously in the water-
SDR-based multimode radar [72]. Such radar is small-size fall data. The drawback of this approach is that
and highly configurable. However, the operational perfor- it needs a reference receiver to calibrate the
mance of SDR relies highly on the back-end processor. received signal. The accuracy of detection is
In [73], two different implementations of FMCW radar dependent on the calibration accuracy seri-
and an implementation of continuous wave noise radar are ously [76]. 5 G mm-wave radar deployment
presented to test their feasibility for UAS detection. And infrastructure is constructed to detect amateur
their findings indicate that the analog implementation has UAS. The deployed radars capture the signal of
higher updating rate and signal noise ratio (SNR). amateur UAS and upload it to cloud which

12 IEEE A&E SYSTEMS MAGAZINE MARCH 2021

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
 Wang et al.

method is able to detect and recognize UAS simul-

taneously. The micro-Doppler effect caused by fly-
ing UAS can also be used for detection of multiple
UAS in [86], where the time–frequency spectro-
gram is converted into the Cadence-velocity dia-
Figure 8. gram (CVD). And then the Cadence frequency
Categorization of radar signal processing.
spectrum (CFS), as the basis of training data from
each class, is extracted from CVD. K-means classi-
could analyze whether there exist hazards. This fier is used to recognize the component of multiple
approach could be an excellent approach in pro- micro-UAS based on the CFS. Their experimental
tecting safety of the city if the challenges of results on real radar data demonstrated that their
resource management, nonline of sight radar method is capable of handling multiple UAS with
operation, noise mitigation, and big data man- satisfactory classification accuracy.
agement could be addressed [77]. Similar
implementation was presented in [78], where a 2) Learning-Based Pattern Recognition: Learning-
passive radar array system is proposed to based pattern recognition methods is capable of
receive and process the orthogonal frequency classifying various types of objects. An example of
division multiplexing (OFDM) echoes of UAS, conventional classification methods was presented
which is originally transmitted by the nearby in [87]. This research demonstrated a classification
base stations. experiment of UAS versus non-UAS tracks, based
on a mixture of bird, aircraft, and simulated UAS

DVB system based passive radars: The perva- tracks, which is mainly based on the statistical fea-
sive digital television signals are considered as tures of the tracks and yields a high accuracy rate.
an efficient illumination source for passive Neural network based methods employ deep learn-
drone detection radars. In [79]–[81], passive ing techniques to automate the feature engineering
drone detection radars were designed and tested. process of conventional ML. In [88], the UAS
Similar to active radar approaches, the micro-Doppler detection is carried by a CNN which learns the char-
effect could be employed. In [82], UAS classification tests acteristic features in the 2-D distribution of the
were conducted through propeller-driven micro Doppler Doppler spectrogram with high classification accu-
signature and ML. Their experiment reveals that the micro racy. In [89], the deep belief networks are applied to
Doppler signature of a plastic propeller is much less visi- characterize the features embodied by generated
ble than carbon fiber propeller. spectral correlation functions patterns to detect and
An apparent drawback for passive radar is that large identify different types of UAS automatically. The
amount of post processing efforts or multiple receivers are results of experiment illustrate that the proposed
needed to achieve acceptable detection accuracy. system is able to detect and classify micro UAS
effectively. The validation of their approach using
cognitive radio is given in [90]. The benefit of learn-
POSTERIOR SIGNAL PROCESSING
ing-based pattern recognition approach is that such
In UAS detection, efforts are needed to derive weak and systems are programmable and trainable to adapt to
sparse reflection signals of targets from noisy output of various scenarios.
RF front-ends. Researches in this domain can be classified
Radar-based UAS detection can achieve better detec-
into two categories: conventional signal feature based
tion performance than the current other sensors. The
detection and learning-based pattern recognition. The gen-
antennas and signal processors were always considered as
eral categorization of radar signal is given in Figure 8.
expensive options for the implementations. Though radar-
1) Signal Feature Based Detection: The micro-Dopp- based approaches can achieve the better performance of
ler effect of propellers of UAS are proved to be a the UAS detection (long distance and short distance), their
useful feature of UAS detection. In [83] and [84], cost is high in terms of deployment, calibration, and main-
methods for estimation of small-size UAS are dis- tenance. The manipulation of the radar-based approaches
cussed, with focus on the micro-Doppler signatures requires the technicians to have the relative background of
of rotating rotor blades. Their experiments proved radar operations which poses a significant challenge to
that such feature can be used to distinguish UAS ubiquity in a large scale. The light, energy-saving, afford-
from other flying objects. In [85], a method based able, and easy assembling radar element is desired which
on hough transform was proposed to improve the is supposed to be capable of easy deployment and mainte-
detection and tracking performance. By making use nance. The posterior signal processing algorithms fueled
of the linear distributed micro Doppler features, the by ML show great potentials to improve the efficiency and

MARCH 2021 IEEE A&E SYSTEMS MAGAZINE 13

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
Counter-Unmanned Aircraft System(s) (C-UAS): State of the Art, Challenges, and Future Trends

spectrum [91]. The researchers extracted the phrase

difference in the sound to locate the UAS. In the sig-
nal processing, the researchers need to adjust the
weight of each sensor to obtain the higher accuracy
of the location. Apart from the accuracy of detection,
the data fusion can improve the functions of UAS
detection. The RF-based detection can acquire the
Figure 9. RF signal with omnidirectional antennas. The
Data fusion based UAS detection.
researchers could adjust the detection phrase of the
signals in the receiving processing with multiple
the accuracy of UAS detection. The improved capacity of omnidirectional antennas. Thereafter, the system
reduced overhead of portable signal processing algorithms with omnidirectional antennas could acquire signals
can expand the implementations of radar-based UAS in the specific direction [17], [92]. This function
detection. could enable the ground stations to track trajectories
of UAS and determine whether the UAS has the
malicious intentions. The combined function also
DATA-FUSION-BASED UAS DETECTION could enlarge the range of UAS detection which can-
not be gained by the single antenna. The multiple
Data fusion, which is the process of integrating multiple
sensor approaches combine multiple sensors with
data sources to produce more consistent, accurate, and
the same type to obtain the better accuracy or addi-
useful information than that provided by any individual
tional functions to improve the performance of the
data source, has the potential to generate fused data which
UAS detection. The combination of the multiple sen-
is more informative and synthetic than the original inputs.
sors extends the capacity of the single sensor and
The data fusion approaches can leverage the advantages
maximize the range of detection geographically.
of each method to acquire a combined result that is more
robust, accurate, and efficient than the single approaches. 2) Multiple-Type Sensor Data Fusion: In some scenar-
For the UAS detection, data fusion can be used to improve ios, the improvement on the amount cannot miti-
the performance of the UAS detection system to overcome gate the disadvantages of the single sensors.
the disadvantages of the single approach which exist in Different UAS detection approaches are tested
some specific scenarios. in [48]. The acoustic sensors are sensitive to the
Based on the above discussion, the disadvantages and humidity, the temperature, and the vibration in the
the advantages of each approach have been presented. The environment. The cameras are invalid when the
first general problem of the single approach is that limita- sunlight project on the lens directly. And the RF
tion of the detection range and accuracy. The second gen- antennas are hard to recognize the target signal
eral problem of the single approach is the vulnerability to when it is buried in the white Gaussian noise envi-
detection scenarios. The third general problem of the sin- ronment. More specifications can be found in[48].
gle approach is the efficiency of the computation. Based They concluded that the fusion of acoustic and
on the above general problems, the research on the data radar could give more precise detection than other
fusion based UAS detection (shown as Figure 9), can be approaches. The conventional single sensors cannot
classified into three categories: 1) multiple-sensor data achieve outstanding performance on a variable
fusion; 2) multiple-type sensor data fusion; 3) multiple environment, and meet multiple requirements of
sensing algorithm fusion. UAS detection ranges.
1) Multiple-Sensor Data Fusion: Each type of sensors Concurrently, the cost of developing high quality
have their own advantages and disadvantages in the functional sensors is prohibitively high. The
UAS detection scenarios. The general problem of researchers resort to the combination of different
the single approach is the limited detection range. types of sensors. Based on the disadvantages and
The more straight and efficient approach to improve the advantages of each type of sensors, the
the performance of the single approach is designing researchers could combine different types of sen-
a specific type of sensors to avoid the drawbacks of sors to achieve the accuracy and long distance
nature materials. detection. Long and short range detection technolo-
A classical example is acoustic sensor array. Dis- gies are combined, where the passive RF receivers
tributed acoustic sensors are deployed in the detec- detect UAS’s telemetry signals while video and
tion areas. Each sensor can record the audio and acoustic sensors are used to increase the detection
deliver the record to the ground stations to make a accuracy in the near field. Different range sensor
combination evaluation of the environment in sound systems including acoustic, optic, and radars are

14 IEEE A&E SYSTEMS MAGAZINE MARCH 2021

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
 Wang et al.

integrated for UAS detection in target area. They thunderstorm, rain, wind, and UAS). Based on the
deployed a 120-node acoustic array which use recognition of scenarios, the system, proposed in
acoustic camera to locate and track the UAS, and this research, triggers SVM and K nearest neigh-
16 high revolution optical cameras to detect the bor, separately, to detect the amateur drones in the
UAS in the middle distance. In the long distance, restricted areas. To achieve the tracking efficiency,
they adopted MIMO radar to operate three different the authors, in [94]–[97], implemented multiple
band radars to achieve remote detection [93]. The sensing algorithms on the passive mm-wave radar
resulting combination overcomes the drawbacks of system to achieve the different accuracy of track-
each type of sensors on the UAS detection and ing according to the requirement of the recognition.
maximize the advantages of each type of sensors. The combination of the multiple sensing algorithms
Simultaneously, the combination reduced the cost could achieve advantages of the efficiency, the accu-
of deploying sensors in a large scale. In [93], the racy, less overhead of system, etc., according to the
deployment of cameras can meet the middle dis- requirement of the UAS detection. However, how to
tance detection requirement and reduce the cost on make a reasonable arrangement for the sensing algo-
the deployment of acoustic sensors and radars. rithms still needs more efforts. And the reasonable
The combination of different types of sensors could schedules of sensing algorithms can be a stimulation
achieve an outstanding performance for the for the UAS detection system in the future.
restricted areas. However, the deployment and the
configuration of this approach require much more Other data fusion schemes can be based on different
specific and professional technologies and techni- platform integration. The researchers deploy multiple sen-
cians with related backgrounds to maintenance. sors into different platforms to leverage the mobility of dif-
The different type sensor combination is a promis- ferent platforms, thus maximizing the sensors’ capacities.
ing approach to maximize the capacity of sensors The authors deployed the cameras on the surveillance UAS
in the physical levels. In the future, investigation to make sure the amateur drones entering the restricted areas
should be focused on exploring more combinations after the deployed acoustic sensors, in the sensing areas,
and characteristics of different sensors for more send alarms to the ground stations [98]. In this research, they
robust, affordable solutions. could recognize different targets such as birds.
The data fusion approaches combine the advantages of
3) Multiple Sensing Algorithm Fusion: The conven- each approach in the detection. The attempts show that the
tional approaches of data fusion fuse multiple data data fusion approaches have obvious advantages com-
acquired from the sensors. Many data fusion pared with single methods. According to the characteris-
approaches had maximized the capacity of sensors tics of each type of approaches, the detection deployment
greatly. However, for UAS detection, the efficiency could contain multiple schemes in different areas which is
and the accuracy still cannot meet the requirement. apart from the center restricted areas in distance differ-
The novel approaches are needed to combine the ently. Thereafter, how to implement the data fusion algo-
multiple sensing algorithms to achieve the effi- rithms to achieve the consistency of the detection system
ciency and the accuracy required by UAS detection. on the results will be next challenge. Another challenge of
The sensing algorithms could be triggered accord- the data fusion approaches is how to balance the weight of
ing to the status of detection. The activated sensors each approach in the final decision to achieve optimal
deliver the information to the ground stations. detection results.
Thereafter, the ground stations set the status of the The distinction of “capture and retrieve” and “disable
detection, and the relevant algorithms will be and drop” is important. Most malicious UAS are captured
swapped into the processing to extract the features by the defenders with physical capture, directional EMP,
of the target information. The UAS detection system RF jamming, and hacking. However, only the technolo-
will generate the threat outcome according to the gies of RF jamming and hacking can realize the function
result of sensing algorithms. of retrieve. The retrieve function is supposed to be robust,
In this part, the sensing algorithms receive the data accurate, and efficient. The defenders are supposed to be
delivered from the sensors and extract the target fea- confident that their systems have high probabilities of
tures according to the types of sensors. The UAS retrieving the malicious UAS again with protection of the
detection system can adjust the sensing algorithm property and the public. For the “disable and drop,” only
accuracy to meet the requirement of detection once the physical capture methods just drop the UAS from the
the abnormal signals are detected. In [91], the flight. The technologies of directional EMP, RF jamming,
researchers leverage the unsupervised approaches to and hacking have the both capacities of disabling and
extract the features of signal from various acoustic dropping. The directional EMP, RF jamming, and hacking
sensors under different scenarios (bird, airplanes, can disable the UAS sensors, circuit, control system, and

MARCH 2021 IEEE A&E SYSTEMS MAGAZINE 15

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
Counter-Unmanned Aircraft System(s) (C-UAS): State of the Art, Challenges, and Future Trends

Figure 11.
Categorization of Jamming.
Figure 10.
Categories of UAS negation. 2) Directional Electromagnetic Pulse: Electromag-
netic pulses have been mainly used to counter
illegal electronic facilities in the car which could
communication devices to disable the control from remote
restart or disable the operation of control system.
attackers. However, these technologies can go deeper, like
Based on the function of electromagnetic pulse,
damaging control circuit and control algorithms, that can
Gomozov et al. [22] focus on the functional neu-
drop the UAS from the flight directly.
tralization of onboard radio electronic system on
UAS, and they adopted spatio-temporal pulse of
the wavelength
¼ 2:5 cm to neutralize the UAS
STATE OF THE ART MITIGATION
and their results showed that their approach could
The technologies of detection and mitigation are still provide an aimed impact on UAS in the range
immature. The research on UAS mitigation is limited. from 0.5 to 1 km and no harm to biological
David et al. [99] developed an architecture of UAS protection.
defense system. In this architecture, they specified the
The physical capture mainly focuses on disabling the
effective engagement range, including initial target range,
mobility of drone and control system. The physical captur-
detection range, and neutralization range which is domi-
ing approaches have advantages of easy manipulation,
nant for response. And their report showed that when the
light weight, quick assembling, etc. Once the drone is cap-
range is over 4000 feet, the hardware based reaction and
tured by the physical capturing approaches, the drone will
neutralization could operate efficiently. Based on the
experience damages at different levels. The physical cap-
architecture, the approaches could be classified into three
turing approaches are efficient and low cost, but not
main categories, as shown in Figure 10. The first one is
friendly to pilots.
the physical capture which focuses on capturing UAS
with physical methods. The second is to leverage the
noise generator to jam the systems or sensors, thus ren-
dering the UAS inoperable by the UAS controller. The JAMMING
third is to exploit vulnerabilities of system or sensors to
Jamming is the most popular method used in neutraliz-
acquire priority of control.
ing UAS entering restricted areas. The defenders lever-
age noise signal to interfere operation of UAS sensors
PHYSICAL CAPTURE or systems for neutralization. In this section, we classify
jamming into three main categories, as shown in
1) Nets Capture: Net capture is a physical method to Figure 11. Among these attack methods, the main tar-
negate the UAS. The defenders adopt guns or some gets are UAS sensors and systems. Zhao et al. [101]
specific weapons to trigger the net to catch UAS. proposed an approach to leverage a team of UAS to
The net is stretched when the net is shot, and closed form an air defense radar network which could jam the
to disable the mobility of drone when the net targets’ sensors. This approach could detect and negate
touches the drone. Kilian et al. [100] invented a the unauthenticated UAS and their experimental results
deployable net capture system which could be showed that they could track and jam the phantom
installed in the airplane or authenticated UAS. made by DJI to leave the restrict areas, and proofed that
When the unauthorized or unsafe UAS are located, the N UAS, in a team, could negate at most N  ðN 
the system could capture the unauthorized or unsafe 1Þ targets. Li et al. used the direct track deception and
UAS. Practical approaches to neutralize UAS are fusion to invade the control priority of navigation sys-
attracting attentions of the military. In [20], a spin tem and trajectory control system. Based on the GPS
launched UAS projectile is developed. This projec- deception jamming theory, they leveraged the trajectory
tile aims to launch a net to capture a flying UAS. cheating to lead the unauthenticated UAS to fly out
The net is stored in the warhead of projectile which from the restricted areas. Their results showed that both
allow soldiers to shoot it by regular guns. the direct track and the fusion track could make UAS

16 IEEE A&E SYSTEMS MAGAZINE MARCH 2021

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
 Wang et al.

drift off the restricted areas. P€arlin et al. [21] proposed

an approach to use the SDR to realize a protocol-aware
UAS jamming system. They used an SDR to achieve
remote controller’s signal and recognized the communi-
cation protocol with analysis. The SDR generates com-
mand to control the UAS to fly away restricted areas.
They compared three different approaches (tone, sweep,
and protocol-aware) to evaluate the performance of
their approach which showed that the protocol-aware is
more efficient than tone and sweep jamming.

Tone: a narrow band signal at the center of a single
channel.
Figure 12.

Sweep: a linear chirp swept across the entire Catergorization of vulnerabilities.
2.4 GHz ISM band.

Protocol-aware: a signal imitating either Futaba

restricted areas. The jamming approaches can provide
advanced spectrum spread technology (FASST) and
the neutralizing effects on the drones in different levels
advanced continuous channel shifting technology
(from hardware to software). The jamming can be
(ACCST).
deployed in a large scale and take effect for a long
Li et al. [102] proposed an approach to leverage the time. However, the current jamming cannot make a
UAS with jammer to neutralize other UAS eavesdroppers. directional effect which can be controlled by the
They used the mobility of UAS to get close to malicious defenders. The effects of jamming is omnidirectional
UAS and the jammer installed on the UAs could impact which could affect the devices in the restricted areas,
the malicious UAS’s trajectories. Sliti et al. [103] pre- and energy consuming. The jamming needs a long time
sented different attacks which focus on UAS network for to take effects when the drone receives enough jamming
neutralization. signals. In the coming future, the jamming is supposed
to be controlled, directional, and quickly reactions.

Jamming attack: Transmitting a jamming signal to
disrupt communications between a drone and the
pilot, forcing the drone to return to “home” location,
VULNERABILITIES
i.e., where it took off.
There are three main methods to exploit the UAS vulnerabil-

Black hole attack: a type of denial-of-service attack
ities, as shown in Figure 12. Most vulnerabilities exploitation
which discards the incoming or outgoing traffic of
work focus on GPS control using sensors and communica-
communication.
tion protocol. The defenders leverage the spoofing methods

Replay attack: a network attack that vises to mali- to GPS and control using sensors, and adopt modification
ciously repeat a valid communication so that the com- and invading to control using sensors and communication
munication of the UAS could be analyzed and invaded. protocol. Rodday et al. [108] demonstrated an approach to
exploit the identified vulnerabilities of the UAS control sys-
Curpen et al. [104] focus on neutralizing the UAS
tems and performed Man-in-the-Middle attack to inject the
which is based on long term evolution (LTE) network. The
control commands to interact with the UAS. Dey et al. [109]
spectrum analysis in two different cell networks showed
presented cracking SDK, reversing engineering, and GPS
that the efficient jamming range for LTE UAS is approxi-
spoofing to hijack the UAS. They compared the DJI and Par-
mated 60 m. Mototolea et al. [105] leveraged the SDR to
rot drone performances under the exploitation attack. The
analyze and hijack the small UAS. In this work, the decod-
results showed that the DJI is more secure than Parrot, which
ing protocol of DSM2 could get the fingerprinting and pair-
means that the DJI is hard to rush into the restricted area.
ing process. Willner [106] invented a system which could
Chen et al. [110] analyzed the popular altitude estimation
neutralize remotely explosive UAS in a combat zone. This
algorithms utilized in navigation system of UAS and pro-
system could be equipped on the ground or installed on the
posed several effective attacks (shown as “Vulnerabilities”
authenticated UAS which transmit the jamming signal
section) to the vulnerabilities.
once the target is locked. Bhattacharya et al. [107] devel-
oped a game theoretic approach to optimize the jamming
KF-based sensor fusion
method on the expelling the UAS attacker. 1) Maximum false data injection: modifying the
Jamming approaches can provide friendly and zero- calculation functions of GPS and barometer
damage schemes to neutralize the drone entering the measurements.

MARCH 2021 IEEE A&E SYSTEMS MAGAZINE 17

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
Counter-Unmanned Aircraft System(s) (C-UAS): State of the Art, Challenges, and Future Trends

Altitude estimation based on accurate sensor noise energy consumption cameras are suitable for the deploy-
models ment of vision-based detection. The current radar-based
1) Blocking GPS: Disable the GPS readings. detection has the disadvantages of heavy weights which is
a challenge for the mobile UAS platform. The payload and
2) Modifying barometer readings: Manipulating
power supply of the UAS are limited which cannot satisfy
the barometer and inject bad data.
the requirement of radar. The most cases of the deployment

First-order low-pass filter: Inject the barometer of radar-based UAS detection are ground-based stations
readings and influence the accuracy of estimation. and manned aircraft. The most fancy approaches are the
data fusion based detection which are just constrained by
Esteves et al. [111] demonstrated a simulation on a computation capacity. Concurrently, the data fusion based
locked target to gain access to the internal sensors for neu- detection approaches are mainly implemented in the
tralizing the UAS from restricted areas. Melamed et al. ground-based stations and manned aircraft. Apart from the
filed a patent on how to utilize the SDR with antenna array above, the combination of different platforms also has
to detect the UAS and create override signal to link of com- promising potentials to improve the capacity of detection
munication to neutralize the UAS. Marty [112] presented for big properties. The different deployment on the ground-
an approach to hijack the MAVLink protocol on the Ardu- based stations, UAS and manned aircraft could achieve
Pilot Mega 2.5 autopilot. Katewa et al. [113] proposed a scalable detection to malicious UAS.
probabilistic attack model to neutralize the UAS which
executed denial of service attack against a subset of sensors
based on Bernoulli process. They also described the vulner-
CHALLENGES IN UAS DETECTION AND MITIGATION
abilities on the sensors of the UAS and strategies to negate Tables 4 and 5 compare various UAS detection and mit-
the UAS via jamming sensors or access control systems. igation technologies, respectively. On one hand, milli-
Huang et al. [114] proposed a spoofing attack based on the meter wave radar along with data fusion methods are
physical layer to utilize the angle of arrival, distance-based considered as the most promising trends for UAS detec-
path loss, and the Rician  k factor to recognize the UAS tion in the future, on the other hand, physical capture is
and source where the signal comes from. regarded as the most practical and reliable approach to
Penetrating vulnerabilities of system has affected neutralize unwelcome UAS. Hacking and spoofing have
the processing of security in computer field for a long emerged as a promising negation solution with low
time. The evolution of the amateur drones enables the footprint and low collateral damage. However, there are
drones have their own operation system (OS) which a lot of challenges which must be addressed to develop
gives the defenders a great chance to exploit the sys- mature scalable, modular, and affordable approaches to
tem via the vulnerabilities of OS. The integration of UAS detection and negation. In this section, we will
embedded system and sensors extends the vulnerabil- identify the challenges of each UAS detection or nega-
ities of the OS of drones. The vulnerabilities releasing tion technology.
of the OS for the drones could improve the success of
exploit the unauthorized drones with malicious inten-
tions in the future. UAS DETECTION
For the deployment of detection technologies, the
ACOUSTIC-BASED UAS DETECTION
deployment decides the capacity of each type of approach
significantly. According to the nature of sensors, the The piezoelectric substrate materials are the core of the
deployment on ground-based stations, UAS or manned air- acoustic sensors. These materials could generate the elec-
craft are varying. The acoustic sensors are sensitive to the tricity according to the strength of vibration on the surface
sound which needs the environment noise keeps stable and which is very smart characteristics. But these materials
quiet. This means the acoustic sensors are not suitable for also could be affected by the temperature, humidity, and
the mobile platforms. The passive RF-based detection has light intensity. In practice, it is hard to keep a stable per-
specific requirement of the antennas distance between each formance of the detection when the scenario is full of mul-
other which is important to achieve accurate result for pas- tiple variable physical parameters. At the same time, the
sive RF signal detection. The passive RF-based detection acoustic sensors are sensitive to the vibration of the air
needs the platform have powerful computation capacity caused by wind. The real UAS signal with the multiple
which are just for the ground based stations and manned fading effect may be buried in the wind. The acoustic sen-
aircraft. The current UAS cannot provide suitable computa- sor array can improve the accuracy of detection, and mul-
tion and power supply. The vision-based detection can be tiple sensors in different places could locate the position
deployed on the ground-based stations, UAS, and manned of the UAS, but there is limited research which takes into
aircraft. The key sensor of vision-based detection are account Doppler effect generated by the movement of the
mainly cameras. Meanwhile, a number of light weight, low UAS, and the combination effect of the movement of

18 IEEE A&E SYSTEMS MAGAZINE MARCH 2021

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
 Wang et al.

Table 4.

Comparison of UAS Detection Technologies

Methods Principles Enabling Advantages Disadvantages

technologies

Acoustic Unique acoustic Pattern recognition, Low-cost, Affected by weather

Insert pattern of drones’ Digital signal light-weight condition, such as
Graphic rotating motors. processing, TDoA, wind or vibration.
Realtime Limited range of
embedded system detection. Lacks of
identification
capacity.
Vision The unique pattern Digital image Low-cost, High computational
Insert of drones’ processing, Pattern light-weight, capacity
Graphic appearance. recognition, small-size. consumption.
Realtime Airborne Affected by
embedded system, environmental
Night vision illumination
monocular. Limited range of
detection. Limited
identification
capacity
Passive Distinguishable Data traffic pattern Low-cost. Not applicable to DSSS
RF Insert pattern of drones’ recognition (FPV long-range. private or encrypted scheme.
Graphic radio signals and telemetry); Identification data protocol. Confusion
against SDR; Digital signal capacity. Limited capacity to with other
background processing; TDoA Airborne FHSS ISM band
spectrums. IoT
devices.
Active Micro doppler Antenna Long-range. High-cost. High
Radar effects of rotating technologies. High- power
Insert propellers. Unique Pattern recognition. precision consumption. Lacks
Graphic pattern in reflected Digital signal of identification
millimeter wave processing capacity. Easy to be
signals. interfered.
Passive Moving objects Pattern recognition Long-range Needs high
Radar cause changes to Digital signal Low-cost computation
Insert the spectrum processing capacity. Lacks of
Graphic identification ability
Easy to be
interfered.
Data Hybrid decision Data fusion and High Needs high
fusion making. pattern recognition reliability computation
Insert capacity. High-cost
Graphic

wind when the speed of wind is over 5 m/s, because the methods are highly dependent on the telemetry protocol and
effect caused by wind is not negligible. RF front-ends. A novel approach which could recognize
multiple protocols simultaneously are needed. And such an
approach is supposed to be efficient, stable, and easily
PASSIVE RF-BASED UAS DETECTION deployed in mobile embedded devices. To our knowledge,
The passive RF needs multiple antennas to form an antenna there are no SDR specifically designed for UAS detection
array and recognize the UAS according to the combination available on the market. Currently SDR devices suffer
of each antenna’s detection results. The passive RF detection from heavy weight, high energy consumption, and poor

MARCH 2021 IEEE A&E SYSTEMS MAGAZINE 19

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
Counter-Unmanned Aircraft System(s) (C-UAS): State of the Art, Challenges, and Future Trends

Table 5.

Comparison of UAS Mitigation Technologies

Methods Principles Enabling Advantages

technologies Disadvantages

Physical Physically disable or Target High success rate for Destructive, high-cost,
capture block a flying object. tracking, non-protected targets.
Insert motion
Graphic control.
High- Disrupt the logic of High power Low-cost, light-weight, Leakage of
power circuits, so as to disable microwave, small-size. electromagnetic
EMP a flying drone. directional energy. Destructive
Insert antenna,
Graphic target
tracking
RF Block the telemetry or RF power Low-cost, light-weight, Can only deal with
jamming GPS signals of drones amplifier. RF small-size. drones operating on
Insert and trigger their spectrum Nondestructive. ISM band. Interfere
Graphic evacuation or landing recognition Applicable to multiple other ISM band
strategies. drones simultaneously. devices.
Hacking Seize the root System Low-cost, Only deals with
Insert privileges of drones’ penetration. Nondestructive specific operation
Graphic operation system and System system and network
issue appropriate vulnerability based protocols.
operations. analysis. Interfere other ISM
band devices.
Spoofing Use fake positioning Signal Guidance and eviction Interfere other devices.
Insert signals or simulated analysis, capacity, Limited capacity to
Graphic control commands to Datapacket Nondestructive. deal with encrypted
redirect drones. analysis and channels
decoding.

portability, which limits the use of SDR in UAS detection. very similar to some fixed wing UAS, and many bionic
How to design a light-weight, small-size, and low cost RF robots could fly like birds. It is urgent to classify these two
analysis device with comparable or better performance scenarios in the mobile devices, especially these devices
could be a significant improvement for the UAS detection. could be installed in the surveillance UAS. So combining
At last, the artificial intelligence (AI), like deep learning, some additional biological signals in the detection process is
could be a good approach to improve accuracy and robust- needed to recognize these two different situations. Third, the
ness of the RF-based detection methods. With the plenty of deep learning has been applied in vision detection for many
signal data generated per second feeding, the deep learning years, but low size, weight, and power-consumption (SWaP)
has the potential to improve the accuracy and efficiency. deep learning algorithms are still needed. The portable deep
The emergency of the RF signal in different small scales is learning algorithms could be implemented into a new sce-
very important for efficient and accurate recognition for nario without too much time training. In the computation
UAS. field, this function of deep leaning is called transfer learning.
The mature and outstanding models of deep learning can be
VISION-BASED UAS DETECTION implemented to multiple scenarios to achieve excellent per-
formances with few training episodes.
Although vision detection has been investigated for a long
time, research efforts are still needed to improve their perfor-
RADAR-BASED UAS DETECTION
mance. First, there is an urgent need for a vision device
designed for UAS detection to be light-weight, small-size, There have been significant research efforts made on the
and low cost. The most important issue is that how to adjust radar-based UAS detection. The grounded radar could
the aperture of the camera to avoid the fading effect caused well meet the requirement of UAS detection in military.
by sunlight in different angles. Second, the shape of birds is But for the civilian usage of the UAS detection in the

20 IEEE A&E SYSTEMS MAGAZINE MARCH 2021

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
 Wang et al.

scenarios like stadiums and residential areas, the current interconnected cords which is efficient when the tar-
schemes are not easy and fast to be deployed in the places get has frames or propellers. The power system of
with the crowds. Most radars are highly dependent on the the most UAS is based on the propeller. The bolas
antennas which are very huge and lack of flexibility. The could be a powerful tool to stop the UAS working if
light-weight, small-size, and mobile-based phrase array the rope is strong enough. The net and bolas both
radars for UAS usage have potentials to meet the require- need to be designed in variable scale so that they
ment of deploying in the civilian scenarios. This radar could be used to capture UAS in different sizes.
also needs to be low-cost, because UAS does not allow

The physical capture should be easy to master. The
the energy consuming equipment to be onboard. Another
physical capture tools like nets and bolas, could be
challenge for the radar is how to solve the interference
loaded into an easy trigger platform like bullets so
between the radar and other communication equipment on
that a proper shooting gun could trigger the bullets
the UAS. Based on advanced radar detection equipment,
to the target and capture UAS. The surveillance
there also needs an optimized deployment approach to
UAS just carry a light and simple trigger platform
change the radar deployment according to the change of
without much energy consumption. To improve the
scenarios in real time. Of course, the fast and accurate
successful capturing rate, the physical capture meth-
radar analysis approaches also can make a great improve-
ods needs assistance of the navigation and tracking
ment of the detection of the UAS.
systems like missiles.

DATA FUSION BASED UAS DETECTION

DIRECTIONAL EMP
Multiple signals fuse from different detection devices
which have different data format. The traditional methods The directional EMP is a very efficient weapon to counter
perform well on the different data in the same acquisition the UAS which could navigate itself by inertial measure-
accuracy, but cannot fuse well in different acquisitions. ment unit (IMU) without any communication with outer
To detect unauthorized or unsafe UAS, the surveillance facilities. The main challenges of the directional EMP are
needs to fuse multiple data from different devices includ- no harm to human being, low divergence angle, and long
ing video, radio, and audio, and so on. A novel fusion distance effect.
approach could input the data in different formats simulta-
Most EMP shooting contains too much electromag-
neously and easily port in different embedded systems. netic energy so that it also has damage effect on the
other nearby facilities or human beings. To make
the EMP shooting approaches more practical, the
UAS MITIGATION researchers need to find a proper frequency that take
efficient effect on UAS and zero damage to organ-
PHYSICAL CAPTURE ism. Because the organism obtains the electromag-
The physical capture is the most direct way to counter netic materials, if the EMP frequency is similar to
unauthorized or unsafe UAS, which is easy to be the respond frequency to the humans or animals, the
deployed. So the requirement of the physical capture organism will take response to the specific frequen-
methods is light-weight, variable scale, and easy to master. cies. We need to make sure that the EMP weapons
work in a different frequency from human and ani-

The physical capture needs to be light so that the
mal response.
human being could carry it and get ready to take
down the UAS once they make sure that the UAS is
The EMP is high power weapon which needs much
unauthorized or unsafe. Also, the surveillance UAS electricity to drive the EMP. However, the divergence
could install the equipment and the surveillance angle of the EMP wave will cause the much fading of
UAS could counter the intrusion UAS when they energy when it has effects on the target. So the research
are patrolling. needs to reconsider how to minimize the divergence
angle of the wave. The narrow divergence angle of the

The physical capture needs to be variable scale so
wave could improve the success rate of countering
that it could work when there are many different
UAS while saving power.
styles of UAS in size. In the current, the most effect
approach is net, but the net also needs to be opti-
There is a disadvantage of the EMP. Different frequen-
mized which should be light, firm, and recyclable. cies have different transmission distances. The higher
The benefit of the net is that it could capture all the frequency EMP will disappear more quickly in the air
things in the capacity of net. Apart from the net, while the higher frequency EMP obtains more energy
the bolas also works well when capture the UAS. which improves the counter successful rate. So there is
The bolas is made of weights on the ends of an embarrassing problem: when the target is detected

MARCH 2021 IEEE A&E SYSTEMS MAGAZINE 21

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
Counter-Unmanned Aircraft System(s) (C-UAS): State of the Art, Challenges, and Future Trends

but the counter distance is limited. How to balance the correct data from the sensors. The remote interfer-
frequency and distance to achieve a better performance ence devices and approaches are needed to lead the
of the counter purpose is desired. UAS to fly away from the sensitive area in a long
distance.
RF JAMMING
The network using on UAS are mainly WiFi and
The current RF Jamming approaches have the following cellular networks. One way is to attack the WiFi
disadvantages. First, the RF jamming consumes too much and cellular networks of the UAS to obtain the pri-
power which is not practical for surveillance UAS to carry ority of the UAS, and send command to autopilot to
to execute the RF jamming precisely. Second, RF jam- lead the UAS to launch off in a safe place. Also, the
ming works for an area, however the RF jamming could defenders could leverage network to access the link
not jam a specific target in a desired point. The RF jam- and play the man-in-the-middle, decrypt the com-
ming only takes effect when the UAS are communicating munication packet of the command, and then mod-
with outer devices. The RF jamming does not work if the ify the command to control the target to launch or
UAS navigates itself with inner global navigation system fly back where it comes. There are also another
(GPS). approach to allow the cellular operators to request
authentication on the cellular link to check the com-

The current RF jamming devices are deployed on mands on the flight randomly.
the ground or mobile vehicles which have limited
movement space to generate the efficient jamming
There are some other sensors that the researches
signals to interfere the UAS communication system. have not tried yet like optical flow, camera, and
Also these devices are too heavy to be loaded on the laser. The optical flow sensors are leveraged to
surveillance UAS. So the light-weight, high effi- locate the position of the UAS, so that the UAS
cient and low cost RF jamming devices are required, could navigate itself to the destination. The image
especially the UAS oriented RF jamming devices. matching technology enable the UAS to leverage
the cameras to get the destinations. The laser sen-

RF jamming has similar characteristics to elec-
sors are also a powerful tool to get location and nav-
tromagnetic wave. So once the target is deter-
igation for the UAS. Novel approaches are needed
mined, how to transmit the jamming signal to a
to interfere and spoof these sensors or invalid these
specific position is an issue. The directional
functions to force the UAS to get back.
antenna could send the electromagnetic wave
into the specific area, but it still needs to realize
The UAS controlled by radio communicate with
the specific points attack. The phase array radar each end device with protocols. There is a need
maybe a good direction for further investigation. to recognize the protocol used by the autopilot
A RF jamming array may include the omnidirec- and communication devices. Because this
tional antennas and directional antennas. The approach could enable the officials to determine
control ends change the RF jamming transmitting the parameters of the communication and attack
power and phase to realize a combination of RF the command link used by the pilots on the
jamming in a specific position. ground. At the same time, the recognition

The RF jamming is easy to realize but how to recog- approach could be executed in the SDR so that
nize the target communication channels is also an the surveillance UAS could leverage the SDR to
open problem. The surveillance officials could eaves- decrypt the communication packet and modify the
drop the target UAS communication and determine flight configuration to return.
its communication channels. With the recognition of
communication, the defenders just execute the RF
jamming in the specific channels and the energy on FUTURE TRENDS
the invalid jamming channels will be saved.

HACKING TECHNICAL ADVANCEMENTS

As discussed above, simple detection approaches cannot
Hacking UAS has been investigated for many years. The
get a reliable rate of detecting malicious UAS success-
hacking methods mainly focus on outer interference, net-
fully. On one hand, each simple approach has its disadvan-
working, and spoofing.
tages so the simple detection sensor could not meet all the

The current outer interference requires that the inter- detection requirements in a variable environment; on the
ference devices are very close to the UAS sensors other hand, the UAS designed with different materials and
(like IMU, GPS) so that the UAS could get the configurations also pose a big challenge for simple

22 IEEE A&E SYSTEMS MAGAZINE MARCH 2021

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
 Wang et al.

Figure 13.
Unified framework for drone safety management.

detection sensors to capture. The future UAS detection the UAS are instructing in restricted areas, the
approaches would be more mature, practical, and efficient. defenders could be able to access the system in phys-
The detection schemes need to be combined from multiple ical level to drive away or stop the UAS remotely.
sensors and fused with ground data and aerial data collab-

The basic training for pilots needs to include the
oratively. The diversity and the amount of data in types of
safety operations and security knowledge. The spec-
detection and acquisition space could be a trend to
ification of UAS operation training and certification
improve. Similar to detection schemes, simple negation
could help pilots avoid basic mistakes.
approaches also could not satisfy negation requirement,
especially the countermeasures could not damage the
property of the pilots. Future negation schemes should
focus on navigating the intruding UAS to fly away the sen- UNIFIED AND SECURED COORDINATION STRATEGIES
sitive areas and no harm to the property of pilots. Of
The discussion above shows that it is hard to protect the
course, the different countermeasures are promising to
public from unsafe and unauthorized drone operations by
achieve better performance when they are implemented
using one single approach. Therefore, we propose a uni-
collaboratively. That means how to construct a unified
fied framework of collaborative UAS safety management,
and systematic framework for the UAS safety defense
as shown in Figure 13. The collaborated entities for UAS
also is a challenge in the following stages. What’s more, a
safety management are:
UAS safety defense system includes detection and nega-
tion, so how to balance the two parts in a collaborative 1) Local UAS Coordinator: The local UAS authori-
and unified framework is also a research focus. ties [2] are responsible for making use of all interfa-
ces provided by UAS manufactures to secure the
operation of UAS and handing over UAS within
coordinator when necessary. Obviously this is a dis-
INDUSTRIAL STANDARDS
tributed management paradigm [10].
Most security and safety problems are caused by the peo-
2) Existing Airspace Authorities: Existing airspace
ple’s mistaking operation. The detailed operation and
management authorities are not required to deal
management policy could be helpful to people to avoid
with UAS directly. It is desired for airspace man-
the mistakes and reduce the burden of the defenders.
agement authorities to interact with the regulation
These policies not only serve as guidance to the pilots, but
database and release information to the shared data
also standards to the industries. A guidance allows the
cube[1], [7], which is supposed to shared with local
pilots to make awareness of flight of UAS in safety and
UAS coordinators friendly. The key information
security and avoid the mistaking operations when the
can be visualized to figure out the status of UAS
UAS is on the flight. The industrial standards make it pos-
management quickly and effectively. Apart from
sible to stop the UAS when it is out of control.
the data sharing, the authorities are supposed to

The industrial standards need the market entrance maintenance the security and the integrity of the
standards which limit the UAS on the market to be releasing data in case modified by attackers. Right
controllable and identified in a physical level. Once before the publication of this article, we do notice

MARCH 2021 IEEE A&E SYSTEMS MAGAZINE 23

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
Counter-Unmanned Aircraft System(s) (C-UAS): State of the Art, Challenges, and Future Trends

that FAA has released a mobile App to graphically and mitigating UAS will be essential to the safe integra-
display where drone operation is allowed as well as tion of UAS into the airspace system.
specific rules to follow [115].
3) UAS Manufacturers: Manufacturers are responsible
ACKNOWLEDGMENTS
for specifying the minimal environmental require-
ment for proper manipulation [9] of UAS. Mean- This work was supported through Embry-Riddle Aeronau-
while, they are required to provide privileged tical University’s Faculty Innovative Research in Science
control interfaces for local UAS coordinators [29] and Technology (FIRST) Program and the National Sci-
to interrupt and reaccommodate UAS when neces- ence Foundation under Grant No. 1956193.
sary. Detect and avoid technologies will play an
important role in overcoming barriers to UAS
integration [116]–[121]. REFERENCES
4) Local UAS Coordinators: This entity interacts as an [1] Federal Aviation Administration, “Unmanned aircraft
agent between UAS users and airspace authorities, systems,” Apr. 2019. [Online]. Available: https://www.faa.
they are responsible for providing proper guidelines gov/uas/
for UAS users and make use of privileged control [2] Federal Aviation Administration, “FAA Aerospace Fore-
interfaces to assure that UAS operations comply cast 2019–39,” Apr. 2019. [Online]. Available: https://
with issued regulations. www.faa.gov/aerospace/forecasts
[3] H. Song, R. Srinivasan, T. Sookoor, and S. Jeschke, Smart
5) Trusted Information Provider: The information Cities: Foundations, Principles and Applications. Hobo-
providers are responsible for: a) providing the ken, NJ, USA: Wiley, 2017.
whole framework with safety and security related [4] H. Song, G. A. Fink, and S. Jeschke, Security and Privacy
information. b) reviewing the report submitted from in Cyber-Physical Systems: Foundations, Principles and
residents [6]. Applications. Chichester, U.K.: Wiley, 2017.
Within our proposed framework, UAS operations are [5] H. Song, Y. Liu, and J. Wang, “UAS detection and neg-
within the supervision of local UAS coordinators and ation,” U.S. Patent 62 833 153, 4 12, 2019.
UAS manufactures. The residents could achieve a more [6] X. Yue, Y. Liu, J. Wang, H. Song, and H. Cao, “Software
secure and safe life under a harmony management of the defined radio and wireless acoustic networking for amateur
UAS traffic environment. drone surveillance,” IEEE Commun. Mag., vol. 56, no. 4,
pp. 90–97, Apr. 2018.
[7] Federal Aviation Administration, “UAS sightings report,”
2019, Accessed: April 29, 2019. [Online]. Available:
CONCLUDING REMARKS https://www.faa.gov/uas/resources/uas-sightings-report/
In addition to recreational use, UAS, also known as UAV [8] L. Josephs, “Deliberate’ drone flights shut down London Gat-
or drones, are used across our world to support firefighting wick Airport, stranding thousands of travelers,” 2019,
and search and rescue operations, to monitor and assess Accessed: April 29, 2019. [Online]. Available: https://www.
critical infrastructure, to provide disaster relief by trans- cnbc.com/2018/12/20/drone-sightings-shut-down-britains-
porting emergency medical supplies to remote locations, gatw ick-airport.html
and to aid efforts to secure our borders. However, UAS [9] Federal Aviation Administration, “Security sensitive air-
can also be used for malicious schemes by terrorists, crim- space restrictions,” 2019, Accessed: April 29, 2019.
inal organizations, and lone actors with specific objec- [Online]. Available: https://www.faa.gov/uas/recreational-
tives. To promote safe, secure, and privacy-respecting fliers/where-can-i-fly/airspace-re strictions/security-sensitive/
UAS operations, there is an urgent need for innovative [10] U.S. Department of Homeland Security, “Countering
technologies for detecting and mitigating UAS. Over the unmanned aircraft systems,” 2019, Accessed: April 29,
past five years, significant research efforts have been 2019. [Online]. Available: https://www.dhs.gov/publication/
made to counter UAS: detection technologies are based on countering-unmanned-aircraft-systems
acoustic, vision, passive RF, or data fusion; and mitigation [11] E. B. Carr, “Unmanned aerial vehicles: Examining the
technologies include physical capture or jamming. In this safety, security, privacy and regulatory issues of integra-
article, we provided a comprehensive survey of existing tion into US airspace,” Nat. Centre Policy Anal., vol. 23,
literature in the area of UAS detection and mitigation, 2013, Art. no. 2014.
identified the challenges in countering unauthorized or [12] B. Jiang, J. Yang, and H. Song, “Protecting privacy from
unsafe UAS, and evaluated the trends of detection and aerial photography: State of the art, opportunities, and
mitigation for protecting against UAS-based threats. We challenges,” in Proc. IEEE Conf. Comput. Commun.
envision that an integrated system capable of detecting Workshops, Jul. 2020, pp. 1–6.

24 IEEE A&E SYSTEMS MAGAZINE MARCH 2021

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
 Wang et al.

[13] Department of Homeland Security & Department of [25] M. Humphries, “WASP: The Linux-powered flying spy
Defense & Federal Aviation Administration, “Federal drone that cracks Wi-Fi & GSM networks,” Jul. 2011.
Contract Opportunity for Air Domain Awareness and [Online]. Available: https://www.geek.com/geek-pick/
Protection in the NAS 2019–2020 Equipment Demon- wasp-the-linux-powered-flying-spy-drone- that-cracks-wi-
stration and Evaluation,” 2019–2020. [Online]. Avail- fi-gsm-netwokrs-1407741/
able: https://govtribe.com/opportunity/federal-contract- [26] G. Norman, “Border patrol spots drone trying to help
opportunity/air-domai n-awareness-and-protection-in- migrants illegally enter America,” Apr. 2019. [Online].
the-nas-70rsat20rfi000001#. Available: https://www.foxnews.com/us/border-patrol-
[14] Defense Advanced Research Projects Agency, “RFI— foils-drone-trying-to-help-mig rants-illegally-enter-
Detection and negation ideas for protecting against small america
unmanned air systems,” 2017, Accessed: March 19, 2020. [27] L. Seabrook and M. Valdes, “Florida peeping Tom uses
[Online]. Available: https://fbo.gov.surf/FBO/Solicitation/ drone to spy on women in high rise, police say,” Oct.
DARPA-SN-17-77 2018. [Online]. Available: https://www.ajc.com/news/
[15] J. Kim, C. Park, J. Ahn, Y. Ko, J. Park, and national/florida-peeping-tom-uses-drone-spy-wo men-high-
J. C. Gallagher, “Real-time UAV sound detection and rise-police-say/hDUjd4fP7QhwkQo6ReNlpO/
analysis system,” in Proc. IEEE Sensors Appl. Symp., [28] M. Margaritoff, “Woman confronted by peeping drone
Mar. 2017, pp. 1–5. outside bedroom window,” Feb. 2018. [Online]. Available:
[16] F. Christnacher et al., “Optical and acoustical UAV https://www.thedrive.com/aerial/18752/woman-confronted-
detection,” in Electro-Optical Remote Sensing X, by-peeping-drone -outside-bedroom-window
vol. 9988. SPIE: Bellingham, WA, USA, 2016, Art. no. [29] U.S. Government Accountability Office, “Small unmanned
99880B. aircraft systems: FAA should improve its management of
[17] I. Bisio, C. Garibotto, F. Lavagetto, A. Sciarrone, and S. safety risks,” 2018.
Zappatore, “Unauthorized amateur UAV detection based [30] U.S. Department of Homeland Security, “Unmanned
on WiFi statistical fingerprint analysis,” IEEE Commun. aircraft systems (UAS)—Critical infrastructure,” 2019,
Mag., vol. 56, no. 4, pp. 106–111, Apr. 2018. Accessed: Apr. 29, 2019. [Online]. Available: https://
[18] M. E. Rovkin et al., “Radar detection of small-size www.dhs.gov/cisa/uas-critical-infrastructure
UAVs,” in Proc. Ural Symp. Biomed. Eng., Radioelec- [31] Federal Aviation Administration, “Airspace restrictions,”
tronics Inf. Technol., May 2018, pp. 371–374. 2019, Accessed: Apr. 29, 2019. [Online]. Available:
[19] J. Sander, A. Kuwertz, D. M€uhlenberg, and W. M€
uller, https://www.faa.gov/uas/recreational-fliers/where-can-i-fly/
“High-level data fusion component for drone classification airspace-re striction/

and decision support in counter UAV,” in Open Architec- [32] J. Vil
ımek and L. Bua
zita, “Ways for copter drone acustic
ture/Open Business Model Net-Centric Systems and detection,” in Proc. Int. Conf. Mil. Technol., May 2017,
Defense Transformation, vol. 10651. SPIE: Bellingham, pp. 349–353.
WA, USA, 2018, Art. no. 106510F. [33] B. Jang, Y. Seo, B. On, and S. Im, “Euclidean
[20] B. Tomasz, F. Richard, and T. LaMar, “Scalable effect net distance based algorithm for UAV acoustic detection,”
warhead,” Feb. 5, 2019, U.S. Patent 10 197 365. in Proc. Int. Conf. Electron., Inf. Commun., Jan. 2018,
[21] K. P€arlin, M. M. Alam, and Y. L. Moullec, “Jamming of pp. 1–2.
UAV remote control systems using software defined [34] E. E. Case, A. M. Zelnio, and B. D. Rigling, “Low-cost
radio,” in Proc. Int. Conf. Mil. Commun. Inf. Syst., May acoustic array for small UAV detection and tracking,” in
2018, pp. 1–6. Proc. IEEE Nat. Aerosp. Electron. Conf., Jul. 2008,
[22] A. V. Gomozov, D. V. Gretskih, V. A. Katrich, and pp. 110–113.
M. V. Nesterenko, “Functional neutralization of small-size [35] X. Chang, C. Yang, J. Wu, X. Shi, and Z. Shi, “A sur-
UAVs by focused electromagnetic radiation,” in Proc. veillance system for drone localization and tracking
XXIInd Int. Seminar/Workshop Direct Inverse Problems using acoustic arrays,” in Proc. IEEE 10th Sensor
Electromagn. Acoust. Wave Theory, Sep. 2017, Array Multichannel Signal Process. Workshop, 2018,
pp. 187–189. pp. 573–577.
[23] A. Giaritelli, “Super Bowl saw 54 drone incursions: [36] J. Busset et al., “Detection and tracking of drones using
Homeland security,” Apr. 2019. [Online]. Available: advanced acoustic cameras,” in Unmanned/Unattended Sen-
https://www.washingtonexaminer.com/news/super-bowl- sors and Sensor Networks XI; and Advanced Free-Space
saw-54-drone-incursi ons-homeland-security Optical Communication Techniques and Applications, vol.
[24] L. Mowat, “Virgin flight comes within ‘seconds’ of 9647. SPIE: Bellingham, WA, USA, 2015, Art. no. 96470F.
crashing into drones at Heathrow,” Apr. 2019. [37] H. Liu, Z. Wei, Y. Chen, J. Pan, L. Lin, and Y. Ren,
[Online]. Available: https://news.yahoo.com/virgin- “Drone detection based on an audio-assisted camera
flight-comes-within-seconds-crashing-dron es-heathr array,” in Proc. IEEE 3rd Int. Conf. Multimedia Big Data,
ow-082129196.html Apr. 2017, pp. 402–406.

MARCH 2021 IEEE A&E SYSTEMS MAGAZINE 25

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
Counter-Unmanned Aircraft System(s) (C-UAS): State of the Art, Challenges, and Future Trends

[38] A. Bernardini, F. Mangiatordi, E. Pallotti, and L. Capodi- [52] M. Saqib, S. D. Khan, N. Sharma, and M. Blumenstein,
ferro, “Drone detection by acoustic signature identification,” “A study on detecting drones using deep convolutional
Electron. Imag., vol. 2017, no. 10, pp. 60–64, 2017. neural networks,” in Proc. IEEE Int. Conf. Adv. Video Sig-
[39] S. Jeon, J.-W. Shin, Y.-J. Lee, W.-H. Kim, Y. Kwon, and nal Based Surveillance, 2017, pp. 1–5.
H.-Y. Yang, “Empirical study of drone sound detection in [53] C. Aker and S. Kalkan, “Using deep networks for drone
real-life environment with deep neural networks,” in Proc. detection,” 2017, arXiv:1706.05726.
IEEE 25th Eur. Signal Process. Conf., 2017, pp. 1858–1862. [54] A. Coluccia et al., “Drone-vs-bird detection challenge at
[40] H. Zhang, C. Cao, L. Xu, and T. A. Gulliver, “A UAV IEEE AVSS2017,” in Proc. 14th IEEE Int. Conf. Adv.
detection algorithm based on an artificial neural network,” Video Signal Based Surveillance, Aug. 2017, pp. 1–6.
IEEE Access, vol. 6, pp. 24 720–24 728, 2018. [55] A. Schumann, L. Sommer, J. Klatte, T. Schuchert, and
[41] P. Kosolyudhthasarn, V. Visoottiviseth, D. Fall, and J. Beyerer, “Deep cross-domain flying object classifica-
S. Kashihara, “Drone detection and identification by using tion for robust UAV detection,” in Proc. 14th IEEE
packet length signature,” in Proc. 15th IEEE Int. Joint Int. Conf. Adv. Video Signal Based Surveillance, Aug.
Conf. Comput. Sci. Softw. Eng., Jul. 2018, pp. 1–6. 2017, pp. 1–6.
[42] T. Miquel, J. Condomines, R. Chemali, and N. Larrieu, [56] P. Andrasi, T. Radisi
c, M. Mustra, and J. Ivosevi
c, “Night-
“Design of a robust controller/observer for TCP/AQM net- time detection of UAVs using thermal infrared camera,” in
work: First application to intrusion detection systems for Proc. Int. Conf. Air Transp. INAIR, 2017, pp. 183–190.
drone fleet,” in Proc. IEEE/RSJ Int. Conf. Intell. Robots [57] S. Hoseini, G. Orchard, A. Yousefzadeh, B. Deverakonda,
Syst., Sep. 2017, pp. 1707–1712. T. Serrano-Gotarredona, and B. Linares-Barranco, “ Pas-
[43] P. Nguyen, H. Truong, M. Ravindranathan, A. Nguyen, sive localization and detection of quadcopter UAVs by
R. Han, and T. Vu, “Matthan: Drone presence detection using dynamic vision sensor,” in Proc. 5th Iranian Joint
by identifying physical signatures in the drone’s RF Congr. Fuzzy Intell. Syst., 2017, pp. 81–85.
communication,” in Proc. Int. Conf. Mobile Syst., Appl. [58] J. Ochodnickỳ, Z. Matousek, M. Babjak, and J. Kurty,
Serv., 2017, pp. 211–224. “Drone detection by ku-band battlefield radar,” in Proc.
[44] P. Nguyen, M. Ravindranatha, A. Nguyen, R. Han, and IEEE Int. Conf. Mil. Technol., 2017, pp. 613–616.
T. Vu, “Investigating cost-effective RF-based detection of [59] S. Park and S. Park, “Configuration of an X-band FMCW
drones,” in Proc. 2nd Workshop Micro Aerial Veh. Netw., radar targeted for drone detection,” in Proc. IEEE Int.
Syst. Appl. Civilian Use, 2016, pp. 17–22. Symp. Antennas Propag., Oct. 2017, pp. 1–2.
[45] S. Basak and B. Scheers, “Passive radio system for real- [60] M. Caris, W. Johannes, S. Sieger, V. Port, and S. Stanko,
time drone detection and doa estimation,” in Proc. Int. “Detection of small UAS with W-band radar,” in Proc.
Conf. Mil. Commun. Inf. Syst., May 2018, pp. 1–6. 18th Int. Radar Symp., Jun. 2017, pp. 1–6.
[46] D. Mototolea and C. Stolk, “Detection and localization of [61] R. Nakamura and H. Hadama, “Characteristics of ultra-
small drones using commercial off-the-shelf FPGA based wideband radar echoes from a drone,” in Proc. IEICE
software defined radio systems,” in Proc. IEEE Int. Conf. Commun. Express, vol. 6, no. 9, pp. 530–534, 2017.
Commun., Jun. 2018, pp. 465–470. e atk
[62] M. KrA ˛ A e and L. Fuxa, “Mini UAVs detection by
[47] Q. Dong and Q. Zou, “Visual UAV detection method with radar,” in Proc. Int. Conf. Mil. Technol., May 2015,
online feature classification,” in Proc. IEEE 2nd Inf. Technol., pp. 1–5.
Netw., Electron. Automat. Control Conf., 2017, pp. 429–432. [63] D. Shin, D. Jung, D. Kim, J. Ham, and S. Park, “A distrib-
[48] S. Hengy et al., “Multimodal UAV detection: Study of uted FMCW radar system based on fiber-optic links for
various intrusion scenarios,” in Electro-Optical Remote small drone detection,” IEEE Trans. Instrum. Meas., vol.
Sensing XI, vol. 10434. SPIE: Bellingham, WA, USA, 66, no. 2, pp. 340–347, Feb. 2017.
2017, Art. no. 104340P. [64] M. Jahangir and C. Baker, “Robust detection of micro-
[49] V. M. Sineglazov, “Multi-functional integrated complex of UAS drones with l-band 3-D holographic radar,” in Proc.
detection and identification of UAVs,” in Proc. IEEE Int. Sensor Signal Process. Defence, Sep. 2016, pp. 1–5.
Conf. Actual Problems Unmanned Aerial Veh. Develop., [65] A. D. de Quevedo, F. I. Urzaiz, J. G. Menoyo, and
Oct. 2015, pp. 320–323. A. A. L
opez, “Drone detection with X-band ubiquitous
[50] C. Briese, A. Seel, and F. Andert, “Vision-based detection radar,” in Proc. 19th Int. Radar Symp., Jun. 2018, pp. 1–10.
of non-cooperative UAVs using frame differencing and [66] J. Klare, O. Biallawons, and D. Cerutti-Maori, “UAV
temporal filter,” in Proc. Int. Conf. Unmanned Aircraft detection with MIMO radar,” in Proc. 18th Int. Radar
Syst., Jun. 2018, pp. 606–613. Symp., Jun. 2017, pp. 1–8.
[51] T. M. T. Perschke and K. Moren, “Real-time detection of [67] F. Hoffmann, M. Ritchie, F. Fioranelli, A. Charlish, and
drones at large distances with 25 megapixel cameras,” in H. Griffiths, “Micro-Doppler based detection and tracking
Proc. SPIE, vol. 10799, 2018, pp. 10 799–10 799–15. of UAVs with multistatic radar,” in Proc. IEEE Radar
[Online]. Available: https://doi.org/10.1117/12.2324678 Conf., May 2016, pp. 1–6.

26 IEEE A&E SYSTEMS MAGAZINE MARCH 2021

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
 Wang et al.

[68] M. Jian, Z. Lu, and V. C. Chen, “Drone detection and [82] H. Fu, S. Abeywickrama, L. Zhang, and C. Yuen, “Low-
tracking based on phase-interferometric Doppler radar,” in complexity portable passive drone surveillance via SDR-
Proc. IEEE Radar Conf., Apr. 2018, pp. 1146–1149. based signal processing,” IEEE Commun. Mag., vol. 56,
[69] G. Sacco, E. Pittella, S. Pisa, and E. Piuzzi, “A MISO no. 4, pp. 112–118, Apr. 2018.
radar system for drone localization,” in Proc. 5th IEEE [83] Y. Zhao and Y. Su, “Cyclostationary phase analysis on
Int. Workshop Metrology AeroSpace, 2018, pp. 549–553. micro-Doppler parameters for radar-based small UAVs
[70] M. Zywek, G. Krawczyk, and M. Malanowski, detection,” IEEE Trans. Instrum. Meas., vol. 67, no. 9,
“Experimental results of drone detection using noise pp. 2048–2057, Sep. 2018.
radar,” in Proc. 19th Int. Radar Symp., Jun. 2018, [84] M. Jian, Z. Lu, and V. C. Chen, “Experimental study
pp. 1–10. on radar micro-Doppler signatures of unmanned aerial
[71] S. J. Lee, J. H. Jung, and B. Park, “Possibility verification vehicles,” in Proc. IEEE Radar Conf., 2017, pp. 0854–
of drone detection radar based on pseudo random binary 0857.
sequence,” in Proc. Int. SoC Des. Conf., Oct. 2016, [85] C. Zhou, Y. Liu, and Y. Song, “Detection and tracking of
pp. 291–292. a UAV via hough transform,” in Proc. CIE Int. Conf.
[72] Y. Kwag, I. Woo, H. Kwak, and Y. Jung, “Multi- Radar, Oct. 2016, pp. 1–4.
mode SDR radar platform for small air-vehicle drone [86] W. Zhang and G. Li, “Detection of multiple micro-drones
detection,” in Proc. IEEE CIE Int. Conf. Radar, Oct. via cadence velocity diagram analysis,” Electron. Lett.,
2016, pp. 1–4. vol. 54, no. 7, pp. 441–443, 2018.
[73] K. Stasiak, M. Ciesielski, A. Kurowska, and W. Przybysz, [87] N. Mohajerin, J. Histon, R. Dizaji, and S. L. Waslander,
“A study on using different kinds of continuous-wave “Feature extraction and radar track classification for
radars operating in c-band for drone detection,” in Proc. detecting UAVs in civillian airspace,” in Proc. IEEE
22nd IEEE Int. Microw. Radar Conf., May 2018, Radar Conf., May 2014, pp. 0674–0679.
pp. 521–526. e
[88] J. Martinez, D. Kopyto, M. SchAijtz, and M. Vossiek,
[74] T. Martelli, F. Murgia, F. Colone, C. Bongioanni, and “Convolutional neural network assisted detection and
P. Lombardo, “Detection and 3D localization of ultralight localization of UAVs with a narrowband multi-site radar,”
aircrafts and drones with a wifi-based passive radar,” in in Proc. IEEE MTT-S Int. Conf. Microw. Intell. Mobility,
Proc. Int. Conf. Radar Syst., Oct. 2017, pp. 1–6. Apr. 2018, pp. 1–4.
[75] B. Knoedler, R. Zemmari, and W. Koch, “On the [89] G. J. Mendis, T. Randeny, J. Wei, and A. Madanayake,
detection of small UAV using a GSM passive coherent “Deep learning based Doppler radar for micro UAS detec-
location system,” in Proc. IEEE 17th Int. Radar Symp., tion and classification,” in Proc. IEEE Mil. Commun.
2016, pp. 1–4. Conf., 2016, pp. 924–929.
[76] A. D. Chadwick, “Micro-drone detection using software- [90] G. J. Mendis, J. Wei, and A. Madanayake, “Deep learning
defined 3G passive radar,” in Proc. Int. Conf. Radar Syst., cognitive radar for micro UAS detection and classi-
Oct. 2017, pp. 1–6. fication,” in Proc. Cogn. Commun. Aerosp. Appl. Work-
[77] D. Solomitckii, M. Gapeyenko, V. Semkin, S. Andreev, shop, Jun. 2017, pp. 1–5.
and Y. Koucheryavy, “Technologies for efficient amateur [91] Z. Uddin, M. Altaf, M. Bilal, L. Nkenyereye, and
drone detection in 5G millimeter-wave cellular infra- A. K. Bashir, “Amateur drones detection: A machine
structure,” IEEE Commun. Mag., vol. 56, no. 1, pp. 43–50, learning approach utilizing the acoustic signals in the pres-
Jan. 2018. ence of strong interference,” Comput. Commun., vol. 154,
[78] X. Yang, K. Huo, W. Jiang, J. Zhao, and Z. Qiu, “A pas- pp. 236–245, Mar. 2020. [Online]. Available: http://dx.doi.
sive radar system for detecting UAV based on the OFDM org/10.1016/j.comcom.2020.02.065
communication signal,” in Proc. Progress Electromagn. [92] B. Nuss, L. Sit, M. Fennel, J. Mayer, T. Mahler, and
Res. Symp., Aug. 2016, pp. 2757–2762. T. Zwick, “MIMO OFDM radar system for drone
[79] Y. Liu, X. Wan, H. Tang, J. Yi, Y. Cheng, and X. Zhang, detection,” in Proc. 18th Int. Radar Symp., Jun. 2017,
“Digital television based passive bistatic radar system for pp. 1–9.
drone detection,” in Proc. IEEE Radar Conf., May 2017, e uniger et al., “ Detection of mini-UAVs in the presence
[93] U. BA€
pp. 1493–1497. of strong topographic relief: A multisensor perspective,” in
[80] G. Fang, J. Yi, X. Wan, Y. Liu, and H. Ke, “Experimental Proc. SPIE, vol. 9997, 2016, pp. 999 702–999 702–8.
research of multistatic passive radar with a single antenna [94] V. K. Klochko, V. V. Strotov, and S. A. Smirnov,
for drone detection,” IEEE Access, vol. 6, pp. 33 542– “Multiple objects detection and tracking in passive scan-
33 551, 2018. ning millimeter-wave imaging systems,” in Millimetre
e
[81] C. SchAijpbach, C. Patry, F. Maasdorp, U. B€uniger, and Wave and Terahertz Sensors and Technology XII, N. A.
P. Wellig, “Micro-UAV detection using DAB-based pas- Salmon and F. Gumbmann, Eds., vol. 11164, Bellingham,
sive radar,” in Proc. IEEE Radar Conf., May 2017, WA, USA: SPIE, 2019, pp. 117–124. [Online]. Available:
pp. 1037–1040. https://doi.org/10.1117/12.2532546

MARCH 2021 IEEE A&E SYSTEMS MAGAZINE 27

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
Counter-Unmanned Aircraft System(s) (C-UAS): State of the Art, Challenges, and Future Trends

[95] M. Martinez, “UAS detection, classification, and tracking [109] V. Dey, V. Pudi, A. Chattopadhyay, and Y. Elovici,
in urban terrain,” in Proc. IEEE Radar Conf., Apr. 2019, “Security vulnerabilities of unmanned aerial vehicles and
pp. 1–6. countermeasures: An experimental study,” in Proc. 31st
[96] B. M. Sazdi
c-Joti
c, D. R. Obradovi
c, D. M. Bujakovi
c, Int. Conf. VLSI Des. 17th Int. Conf. Embedded Syst., Jan.
and B. P. Bondzuli
c, “Feature extraction for drone classi- 2018, pp. 398–403.
fication,” in Proc. 14th Int. Conf. Adv. Technol., Syst. Serv- [110] W. Chen, Y. Dong, and Z. Duan, “Attacking altitude esti-
ices Telecommun., Oct. 2019, pp. 376–379. mation in drone navigation,” in Proc. IEEE INFOCOM
[97] T. B. Sarikaya, D. Yumus, M. Efe, G. Soysal, and T. Kiru- Conf. Comput. Commun. Workshops, Apr. 2018, pp. 888–
barajan, “Track based UAV classification using surveil- 893.
lance radars,” in Proc. 22th Int. Conf. Inf. Fusion, Jul. [111] J. L. Esteves, E. Cottais, and C. Kasmi, “Unlocking the
2019, pp. 1–6. access to the effects induced by IEMI on a civilian UAV,”
[98] J. Wang et al., “Integrating ground surveillance with aerial in Proc. Int. Symp. Electromagn. Compat., Aug. 2018,
surveillance for enhanced amateur drone detection,” in pp. 48–52.
Disruptive Technologies in Information Sciences, M. [112] J. A. Marty, “Vulnerability analysis of the mavlink protocol
Blowers, R. D. Hall, and V. R. Dasari, Eds., vol. 10652, for command and control of unmanned aircraft,” Air Force
Bellingham, WA, USA: SPIE, 2018, pp. 101–110. Institute of Technology, Tech. Rep. AFIT-ENG-14-M-50,
[Online]. Available: https://doi.org/10.1117/12.2304531 2013.
[99] D. Arteche et al., “Drone defense system architecture for US [113] V. Katewa, R. Anguluri, A. Ganlath, and F. Pasqualetti,
navy strategic facilities,” Naval Postgraduate School Mon- “Secure reference-tracking with resource-constrained
terey United States, Tech. Rep. AAT-0704-0188, 2017. UAVs,” in Proc. IEEE Conf. Control Technol. Appl., Aug.
[100] J. C. Kilian, B. J. Wegener, E. Wharton, and D. R. Gave- 2017, pp. 1319–1325.
lek, “Counter-unmanned aerial vehicle system and meth- [114] K. Huang and H. Wang, “Combating the control sig-
od,” Jul. 21 2015, U.S. Patent 9 085 362. nal spoofing attack in UAV systems,” IEEE Trans.
[101] Z. C. Zhao, X. S. Wang, and S. P. Xiao, “Cooperative Veh. Technol., vol. 67, no. 8, pp. 7769–7773, Aug.
deception jamming against radar network using a team of 2018.
UAVs,” in Proc. IET Int. Radar Conf., Apr. 2009, pp. 1–4. [115] K. FAA, “B4ufly mobile app,” May 2019. [Online].
[102] A. Li, Q. Wu, and R. Zhang, “UAV-enabled cooperative Available: https://www.faa.gov/uas/recreational_fliers/
jamming for improving secrecy of ground wiretap where_can_i_fly/b4ufly/
channel,” IEEE Wireless Commun. Lett., vol. 8, no. 1, [116] G. Fasano, D. Accado, A. Moccia, and D. Moroney,
pp. 181–184, Feb. 2019. “Sense and avoid for unmanned aircraft systems,” IEEE
[103] M. Sliti, W. Abdallah, and N. Boudriga, “Jamming attack Aerosp. Electron. Syst. Mag., vol. 31, no. 11, pp. 82–110,
detection in optical UAV networks,” in Proc. 20th Int. Nov. 2016.
Conf. Transparent Opt. Netw., Jul. 2018, pp. 1–5. [117] A. D. Zeitlin, “Sense avoid capability development
[104] R. Curpen, T. Balan, I. A. Micloş, and I. Comanici, challenges,” IEEE Aerosp. Electron. Syst. Mag., vol. 25,
“Assessment of signal jamming efficiency against LTE no. 10, pp. 27–32, Oct. 2010.
UAVs,” in Proc. IEEE Int. Conf. Commun., 2018, [118] R. Opromolla, G. Fasano, and D. Accardo, “Perspectives
pp. 367–370. and sensing concepts for small UAS sense and avoid,” in
[105] D. Mototolea and C. Stolk, “Software defined radio for Proc. IEEE/AIAA 37th Digit. Avionics Syst. Conf.,
analyzing drone communication protocols,” in Proc. Int. Sep. 2018, pp. 1–10.
Conf. Commun., Jun. 2018, pp. 485–490. [119] X. Prats, L. Delgado, J. Ramirez, P. Royo, and E. Pastor,
[106] B. J. Willner, “Methods and apparatuses for detecting and “Requirements, issues, and challenges for sense and avoid
neutralizing remotely activated explosives,” Nov. 26, in unmanned aircraft systems,” J. Aircraft, vol. 49, no. 3,
2009, U.S. Patent App. 12/126,570. pp. 677–687, 2012.
[107] S. Bhattacharya and T. Başar, “Game-theoretic analysis of [120] B. Korn and C. Edinger, “UAS in civil airspace: Demon-
an aerial jamming attack on a UAV communication strating ‘sense and avoid’ capabilities in flight trials,” in
network,” in Proc. Amer. Control Conf., Jun. 2010, Proc. IEEE/AIAA 27th Digit. Avionics Syst. Conf., Oct.
pp. 818–823. 2008, pp. 4.D.1–1–4.D.1–7.
[108] N. M. Rodday, R. D. O. Schmidt, and A. Pras, “Exploring [121] Y. Lin and S. Saripalli, “Sense and avoid for unmanned
security vulnerabilities of unmanned aerial vehicles,” in aerial vehicles using ADS-B,” in Proc. IEEE Int. Conf.
Proc. IEEE/IFIP Netw. Oper. Manage. Symp., Apr. 2016, Robot. Automat., May 2015, pp. 6402–6407.
pp. 993–994.

28 IEEE A&E SYSTEMS MAGAZINE MARCH 2021

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.
 Wang et al.

Jian Wang is currently Science, Embry-Riddle Aeronautical University, Daytona

working toward the Ph.D. Beach, FL, USA, where he is currently an Assistant Pro-
degree with the Depart- fessor and the Director of the Security and Optimization
ment of Electrical Engi- for Networked Globe Laboratory (SONG Lab, www.
neering and Computer SONGLab.us). He served on the faculty of West Virginia
Science, Embry-Riddle University from August 2012 to August 2017. In 2007, he
Aeronautical University, was an Engineering Research Associate with the Texas
Daytona Beach, FL, USA. A&M Transportation Institute. In 2019, he served as a
He is also a Graduate subject matter expert on AI and counter cyber for autono-
Research Assistant with mous unmanned collective control for the U.S. Special
the Security and Optimi- Operations Command (USSOCOM). He has served as an
zation for Networked Associate Technical Editor for the IEEE COMMUNICATIONS
Globe Laboratory (SONG MAGAZINE (2017–present), an Associate Editor for the
Lab, www.SONGLab.us). His major research interests IEEE INTERNET OF THINGS JOURNAL (2020–present), and a
include wireless networks, unmanned aerial systems, and Guest Editor for the IEEE JOURNAL ON SELECTED AREAS IN
machine learning. He received the B.S. degree from COMMUNICATIONS (J-SAC), IEEE INTERNET OF THINGS JOUR-
Nanyang Normal University, Nanyang, China, in 2014, NAL, IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, IEEE
and the M.S. degree from South China Agricultural Uni- SENSORS JOURNAL, IEEE TRANSACTIONS ON INTELLIGENT
versity, Guangzhou, China, in 2017. He was a recipient of TRANSPORTATION SYSTEMS, and IEEE NETWORK. He is the
the Best Paper Award from the 12th IEEE International Editor of six books, including Big Data Analytics for
Conference on Cyber, Physical and Social Computing Cyber-Physical Systems: Machine Learning for the
(CPSCom-2019). Internet of Things (Elsevier, 2019), Smart Cities:
Foundations, Principles and Applications (Wiley,
YongXin Liu received the 2017), Security and Privacy in Cyber-Physical Systems:
B.S. and M.S. degrees Foundations, Principles and Applications (Wiley, 2017),
from South China Agricul- Cyber-Physical Systems: Foundations, Principles and
tural University, Guangz- Applications (Academic Press, 2016), and Industrial
hou, China, in 2011 and Internet of Things: Cybermanufacturing Systems
2014, respectively, and the (Springer, 2016). He is the author of more than 100
Ph.D. degree from the articles. His research interests include cyber-physical
School of Civil Engineer- systems, cybersecurity and privacy, Internet of Things,
ing and Transportation, edge computing, AI/machine learning, big data analyt-
South China University of ics, unmanned aircraft systems, connected vehicle,
Technology, Guangzhou, smart and connected health, and wireless communica-
China. His major research tions and networking. His research has been featured
interests include data min- by popular news media outlets, including IEEE Global-
ing, wireless networks, the Internet of Things, and unmanned Spec’s Engineering360, USA Today, U.S. News &
aerial vehicles. He was a recipient of the Best Paper Award World Report, Fox News, the Association for
from the 12th IEEE International Conference on Cyber, Unmanned Vehicle Systems International, Forbes,
Physical and Social Computing (CPSCom-2019). WFTV, and New Atlas.
Dr. Song is a Senior Member of ACM. He was a recipi-
Houbing Song (Senior ent of the Best Paper Award from the 12th IEEE Interna-
Member, IEEE) received tional Conference on Cyber, Physical, and Social
the M.S. degree in civil Computing (CPSCom-2019), the Best Paper Award from
engineering from the Uni- the 2nd IEEE International Conference on Industrial
versity of Texas, El Paso, Internet (ICII 2019), the Best Paper Award from the 19th
TX, USA, in December Integrated Communication, Navigation and Surveillance
2006, and the Ph.D. Technologies (ICNS 2019) Conference, and the prestigious
degree in electrical Air Force Research Laboratory’s Information Directorate
engineering from the Uni- (AFRL/RI) Visiting Faculty Research Fellowship in 2018.
versity of Virginia, Char-
lottesville, VA, USA, in
August 2012. In August
2017, he joined the
Department of Electrical Engineering and Computer

MARCH 2021 IEEE A&E SYSTEMS MAGAZINE 29

Authorized licensed use limited to: Rutgers University. Downloaded on May 18,2021 at 10:40:09 UTC from IEEE Xplore. Restrictions apply.