1

Key Technologies and System Trade-Offs for

Detection and Localization of Amateur Drones
Mohammad Mahdi Azari, Hazem Sallouha, Alessandro Chiumento,
Sreeraj Rajendran, Evgenii Vinogradov, and Sofie Pollin
Email: [email protected]

Abstract—The use of amateur drones (ADrs) is expected image to identify whether a drone has entered the protected
to significantly increase over the upcoming years. However, space, the drone is then identified and tracked.
regulations do not allow such drones to fly over all areas, in
addition to typical altitude limitations. As a result, there is an Passive methods, on the other hand, listen to either the
urgent need for ADrs surveillance solutions. These solutions ADr emitted audio or to the transmitted RF signal (i.e.
should include means of accurate detection, classification, and control or downlink to the ADr’s base station). Sound-
localization of the unwanted drones in a no-fly zone. In based solutions are able to identify the presence and model
this paper, we give an overview of promising techniques for of commercial ADrs by listening for specific motor sound
modulation classification and signal strength based localization
of ADrs by using surveillance drones (SDrs). By introducing signatures but require distributed microphone arrays to track
a generic altitude dependent propagation model, we show the drone [3]. Passive RF solutions listen instead for the
how detection and localization performance depend on the ADrs downlink transmission of, for example, a video stream
altitude of SDrs. Particularly, our simulation results show a and are able to localize the drone by determining the source
25 dB reduction in the minimum detectable power or 10 times point of the transmission [5]. The main advantages and
coverage enhancement of an SDr by flying at the optimum
altitude. Moreover, for a target no-fly zone, the location disadvantages of such methods are listed in Table I.
estimation error of an ADr can be remarkably reduced by The ADrs detection and tracking solutions described
optimizing the positions of the SDrs. Finally, we conclude the above assume the presence of a ground infrastructure. Hence,
paper with a general discussion about the future work and
in densely built-up areas, the number of deployed sensor
possible challenges of the aerial surveillance systems.
nodes must be dramatically increased to maintain the re-
quired sensor system performance in challenging propaga-
I. I NTRODUCTION tion environments. However, an alternative solution is to
Detecting the presence of amateur drones (ADrs) in a no- detect and localize the ADrs by using surveillance drones
fly zone is an arduous task as the ADrs are usually small (SDrs) with passive RF sensing ability. Using SDrs flying at
objects flying at low altitudes. These ADrs can pose great higher altitudes than the ADrs allows for a flexible solution
safety and security problems in critical locations such as able to be deployed quickly, it can be used to cover a wide
power plants, military zones, densely populated areas and range of ground surfaces and subsequently, is able to detect
private residences [1]. Such a variety of no-fly zone locations and localize the ADrs with high accuracy due to better
requires a flexible surveillance solution. The research com- propagation conditions at high altitudes (i.e. higher signal-
munity is very active in developing techniques to determine to-noise ratio and line-of-sight probability).
the location and subsequently track unidentified drones [2]– Received signal strength (RSS) localization has been used
[5]. successfully for the positioning of ground nodes [6]. A
Current solutions can be divided into active methods, good link between the receiving sensor and the node to be
in which a ground infrastructure actively scans the no-fly localized is necessary to allow RSS-based methods. In fact,
zone for intruders using video or radar techniques [2,4], or it has been shown that the presence of an aerial based sensor,
passive methods in which an ADr is detected by its RF together with an appropriate channel model, can greatly
transmission or audio signature [3,5]. Active Radar-based benefit the connection to and thus the localization of target
methods require large mono- or multi- static RF nodes used nodes [7,8]. The usage of aerial RF monitoring devices can
to scan the no-fly zone. Frequency sweeping is used to scan then make use of better channels between ADrs flying at
for the presence of drones and classification techniques are low altitudes and surveillance drones (SDrs) for detection
employed to determine the nature of the detected drone and and localization to improve current passive RF solutions.
to track it. Such solutions are very powerful but require the In this paper, we present a framework for the detection and
purchase of large devices with fixed coverage radius and the localization of ADrs using SDrs. The proposed framework
small size of the ADrs poses great detection limitations [4]. is based on the received signal by at least three SDrs,
Solutions based on video detection require the presence of see Figure 1. As the SDrs are assumed to be flying at
either distributed camera equipment or 360◦ video recording higher altitudes than the ADrs, this gives them a better
devices [2]. Video processing is then applied on the recorded view of the no-fly zone and places them much further
 TABLE I. Drone Detection Technologies 2

Technology Advantages Disadvantages

Distributed high resolution camera network or
Mature technology 3D cameras needed
Video [2]
Accessible equipment Subject to poor visibility problems
Fixed installation
Sensitive to ambient noise
Cheap sensors Limited range
Audio [3] Limited processing necessary Distributed microphone array necessary
Accessible equipment Large datasets necessary for training
Fixed installation
Either mono-static with limited resolution or
Multi-static and distributed
Radar [4] Easily installable
Small drone cross-section can be challenging to detect
Expensive
Requires good SNR to perform detection
Easily installable
RF [5] Susceptible to interference
Cheap sensors
Range is limited by link quality

away from buildings and people. Moreover, they could also to rely on a very dense infrastructure of RF sensors that
be tethered to guarantee increased safety and operated by are constantly monitoring the radio spectrum. To facilitate
certified pilots or authorities. The intercepted drone needs detection, we foresee passive RF sensing as a simple yet
to be first identified as such, then by employing a realistic robust monitoring technique to protect the target no-fly zone.
propagation model, the approximate location of the ADr is Continuous passive radio spectrum monitoring should be
obtained by multilateration based on the RSS at the SDrs. enabled in SDrs to allow effective detection of ADrs. For
The main contributions of this work can be summarized simplifying the analysis, ADrs are assumed to have omni-
as follows: directional antennas enabling a LoS component to the SDrs
• An overview of the passive RF scanning for detec- as shown in Figure 1. Accordingly, the ADr’s antenna gain
tion solutions, relying on recent innovations in deep is identical for any elevation angle θ.
learning, as a possible solution for the detection and A few recent approaches try to tackle the RF spectrum
localization of an ADr entering the no-fly zone, monitoring problem in a crowd-sourced fashion [9]. Wireless
• a characterization of the propagation channel seen by an spectrum sensing at high altitude is less challenging as more
SDr in urban environments, necessary to determine the LoS channels are available when compared to sensing at
density of the surveillance sensors for meeting detection ground stations. Even with a perfect RF spectrum scanning
and localization constraints, architecture in place, huge effort is involved in analyzing,
• an investigation on the impact of SDr altitude on the detecting and locating transmissions or anomalies in the
coverage area for ADr detection and the range of its sensed RF spectrum. Automated systems should be in place
detectable transmit power, to detect authorized or unauthorized transmissions. The de-
• a study of the optimal SDrs positioning for better tected signals should be then classified to understand the type
sensing and higher localization accuracy over a target of transmission. Subsequently, this technology classification
zone based on a range of possible ADr’s transmit can further aid signal power estimation and RSS based
power, localization algorithms.
• and an overview of the future research directions and
Accurate technology classification can be achieved to
challenges.
some extent using state-of-the-art (SoA) machine learning
classifier models [10,11]. To validate technology classifi-
II. PASSIVE RF SENSING , T ECHNOLOGY DETECTION cation, a few deep learning based time domain models
AND L OCALIZATION employing convolutional neural networks (CNN) and long
In this section, an overview of passive RF sensing and short term memory (LSTM) units are tested. The deep
localization techniques is presented. For each task, we learning models take IQ samples as input giving out the
illustrate a promising solution: deep learning for detection probability of the data belonging to a particular technology
and RSS based distance estimation for localization. Analyses class. Analyses show that the proposed models [10,11]
show that detection and localization performance strongly yields an average classification accuracy close to 90% for
depends on the received signal strength. 11 different technologies, at varying signal-to-noise ratio
(SNR) conditions ranging from 0 dB to 20 dB, independent
of channel characteristics. It was also noticed that most
A. Passive RF sensing and detection of the technology classification algorithms including deep
A first step in the surveillance of ADrs is detecting learning solutions require an SNR above 0 dB for accurate
their presence in the no-fly zone. When detection of small classification which is used as a required SNR threshold in
ADrs with low transmit powers is of interest, it is needed our simulation.
 3

Fig. 1. Due to the high LoS probability PLoS at high altitudes, the use of SDrs guarantees higher SNR for better detection and less
shadowing for accurate localization compared to terrestrial surveillance. As the ADrs’ transmit power is unknown the estimated location
of the ADr is represented by a sphere of radius δ.

B. Localization challenges with uncertainty. As shown in Figure 1 the uncertainty can

be modeled as a sphere of radius δ. This uncertainty can
The multilateration process is the most prominent method be minimized by classifying the technology being used by
for accurately determining the position of a transmitting the ADrs due to the fact that known technologies have
source. It is basically a process that uses the estimated a standard range of transmit powers (e.g., wifi limited to
distances from such source to at least three different re- 20 dBm, LTE limited to 24 dBm in uplink and 43 dBm–
ceivers in order to perform localization. Time-based or RSS- 48 dBm in downlink, etc). A thorough discussion of SDr
based techniques are used to estimate the distance between altitude’s effect on the channel characteristics including path
a transmitter and a receiver [6]. Time-based techniques loss and shadowing effects is presented in the following
estimate the distance by multiplying the time of flight by section.
the speed of light. However, defining the time of flight is
the bottleneck of these techniques as a very accurate time III. W HY F LY H IGHER : A B ETTER L O S E XPERIENCE
synchronization is required between the transmitter and the To analyze the deployment of SDrs over urban environ-
receivers. Certainly, such technique is not feasible for the ments, as our main focus, and to characterize the received
SDrs since the time of flight cannot be accurately defined SNR, a comprehensive understanding of the communication
as there is no cooperation with the ADr. links’ channel characteristics is needed. Although air-to-air
In contrast to time-based techniques, RSS-based solutions (A2A) links are dominated by Line-of-Sight (LoS) propaga-
are known for their computational simplicity and for not tion, the impact of multipath fading due to ground/buildings
requiring any time synchronization. RSS-based techniques reflections cannot be ignored. In [12], the Rician model
estimate the distance by using a deterministic function that with an altitude-dependent K-factor was used to model A2A
represent RSS as a function of distance. In fact, assuming a channels. Naturally, the influence of LoS grows with the
known transmit power, the path loss model is a well known increasing ADr’s altitude as well as the Doppler frequency
representation for the RSS-distance relation. Generally, the due to the higher relative velocity.
major issue limiting the accuracy of RSS-based techniques In general, a ground-to-air (G2A) link encounters obstruc-
is the presence of shadowing between the transmitter and the tions between the terrestrial and aerial nodes which limits
receiver which causes either over- or under-estimation of the the LoS and lowers the quality of the channel. Therefore,
distance. However, this drawback can be overcome by using a G2A link represents a worst case scenario which occurs
SDrs combined with their altitude optimization which is when the ADrs fly in very low altitudes.
thoroughly discussed in the next sections. Another challenge A G2A channel is observed to be significantly different
that SDrs have to tackle is the unknown transmit power than A2A and ground-to-ground (G2G) communication due
of the ADrs (PTx ) due to the use of different standard of to the high impact of altitude on the channel parameters
communication (LTE, WiFi, ...) or power control in the ADr. including path loss exponent, small-scale fading and shad-
The unknown transmit power leads to an estimated location owing effect. To clarify this fact, we consider two extreme
 4
cases for a given ground terminal that aims to communicate exponent is proportional to the number of obstacles
with a drone seen by an elevation angle of θ: between a transmitter and receiver. Accordingly, for
1) θ → 0: In this case, which is equivalent to h → 0 (for larger elevation angle the path loss exponent is smaller
r 6= 0), the channel behavior follows G2G models due to the presence of less obstacles between a ground
where the presence of many obstacles results in a transmitter and an aerial receiver. The reduction of path
dramatic drop for the received power. This significant loss in Figure 2b is due to a decrease in α(θ) and the
power decay is reflected onto the channel model by increase is because of an increase in the link length d
proposing a large path loss exponent α and severe while the altitude increases.
shadowing and small-scale fading effects. It is worth • Small-Scale Fading: a G2A link is likely to experience
noting that for this case the channel between a trans- LoS condition and hence Rician fading is an adequate
mitter and receiver is roughly always non-line-of-sight choice for such channel that reflects the combination
(NLoS) as the probability of line-of-sight (LoS) PLoS of LoS and multipath scatters [14]. In this model, the
converges to zero1 . In fact the LoS probability PLoS fading power is determined by the Rician factor K
can be obtained as follows [7] characterized as the ratio between the power of LoS
and multipath components. In fact, the Rician factor
1
PLoS (θ) = , (1) represents the severity of fading such that a smaller K
1 + a0 exp (−b0 θ) corresponds to a more severe fading. Due to a higher
where a0 and b0 are environment dependent constants. LoS probability and the presence of fewer obstacles
2) θ → 90o : In this case, which is equivalent to h → ∞ and scatters at higher altitudes, the average Rician
(for r 6= 0), the probability of LoS PLoS converges to factor could be characterized as a function of θ, i.e.
one and the channel adopts roughly free space char- K = K(θ) [8]. Investigating a functional form for k(θ)
acteristics. Accordingly, a lower path loss exponent at different urban environment is an open question.
and a lighter small-scale fading and shadowing effects • Shadowing: the shadowing effect is studied in [13]
are experienced since the environment between the where a log-normal distribution is considered separately
transmitter and receiver becomes less obstructed [7]. for each LoS and NLoS component. The standard
The above-mentioned intuition encourages to model the deviation of each group σLoS (θ) and σNLoS (θ) is char-
drone communication channel dependent on the elevation acterized using a negative exponential dependency with
angle as this, easily observable variable, presents a strong the elevation angle in which a lower elevation angle
correlation with the link quality. To this end, the authors and hence altitude leads to a larger variation around the
in [13] studied a statistical propagation model by con- average path loss. Following [13], the overall average
sidering two major groups of received power and their shadowing effect in the links can be represented by the
probability of occurrence, namely LoS and dominant non- standard deviation written as
LoS (NLoS) components. This model captures different σ 2 (θ) = PLoS
2 2
(θ) · σLoS (θ) + [1 − PLoS (θ)]2 · σNLoS
2
(θ),
urban environment properties and proposes a θ-dependent (3)
path loss and shadowing prediction of the communication which is illustrated in Figure 2b. From the figure, as
channel between a terrestrial and an aerial node. the drone goes higher the shadowing effect gradually
To extend this G2A model we refer to the work pre- diminishes due to the presence of fewer obstacles
sented in [7,8] where we include the small-scale fading and between the transmitter and receiver.
elevation angle dependent path loss exponent. This model By relying on this altitude-dependent shadowing model,
unifies a widely used G2G channel model with that of it becomes possible to determine the SDr detection coverage
G2A that enables us to study the co-existence of drones and localization accuracy as function of SDr height. These
with the existing terrestrial networks. In the following, we results include an optimization of SDrs network in order
briefly discuss the dependency of the main components to to provide larger coverage, to detect low power ADrs and
the elevation angle: subsequently maximize the localization accuracy. Finally,
• Path Loss: path loss exponent is linked to the LoS please note that the above-mentioned channel characteristics
probability PLoS in [7] by proposing a negative linear are environment dependent [13], however in the sequel we
dependency as follows have examined an Urban environment and focus on the
detailed analysis of the performance for this scenario only,
α(θ) = −a1 PLoS (θ) + b1 , (2)
rather than quantifying the values for different environments.
where a1 and b1 are environment dependent parameters.
Such dependency, illustrated in Figure 2a for Urban IV. S UPERIOR S URVEILLANCE D RONES
environment, is motivated by the fact that the path loss We deploy an SDr system for detection and localization,
and study its performance as function of altitude. This
1 Please note that for a short distance between a transmitter and receiver,
will give insights into the possibility and benefits of using
the LoS probability PLoS exponentially decreases as the link length in-
creases. This impact, however, is approximately neglected for long G2A SDrs for ADrs surveillance. In this section we discuss the
communication links. efficient positioning of the SDrs to optimally localize an
 5
3 1 110
Shadowing
Average Path Loss
2.8 0.8
Path Loss Exponent, α(θ)

105

LoS Probability, PLoS (θ)

2.6 0.6

Loss [dB]
100

2.4 0.4

95
2.2 α(θ) 0.2
PLoS (θ)

2 0 90
0 10 20 30 40 50 60 200 500 800 1100 1400 1700
o
Elevation Angle, θ [ ] Altitude, h [m]
(a) (b)
Fig. 2. (a) For an urban environment the LoS probability PLoS (θ) is highly dependent on the elevation angle θ. (b) The altitude dependent
path loss and shadowing effect for an specific ground node located at r = 400 m.

ADr considering the scenario in which the ADr is very of SDrs increases. Please note that if the ADr flies higher
close to the ground. In the following, by considering an over the ground, then the link to the SDr will become LoS
Urban environment, we link the altitude of the flying SDr to and hence the channel quality will increase resulting in a
its coverage area, to the ADr’s transmit power, and finally lower detectable power and a higher classification accuracy
to the localization accuracy as the ultimate goal. Note that of the technology used in the ADr.
the detection of ADrs can be done using one SDr whereas
localizing them in the no-fly zone requires the cooperation
of all three SDrs. B. Coverage Extension
In this subsection we present an efficient deployment of an
A. A Target Control Zone SDr to maximize the covered region assuming a minimum
Considering a target control zone for an SDr flying at ADr PTx . Therefore, the result can be used to find the optimal
altitude h, the fact that the ADr’s transmit power is unknown number of surveillance drones in order to cover a larger
means that several flight levels have to be defined based on target region. Figure 3b illustrates the impact of altitude
the range of possible transmit powers rather than a single for different minimum transmit powers. The figure shows
optimal altitude. In any case, in order to detect the drone, that as the drone goes higher the coverage increases such
it should be within the coverage of the SDr. Figure 3a that at an optimum altitude the coverage is maximized. For
illustrates a lower bound of the range of ADr’s transmit instance, an SDr can fly at an altitude of 900 m to maximize
power that is detectable by an SDr at each altitude. This the region of control assuming a minimum ADr’s transmit
lower bound is obtained by comparing the received SNR power of -10 dBm. More technical discussion for the optimal
and a certain threshold. For instance, if the SDr flies at deployment of the drones and the trade-offs can be found in
200 m above a target region of radius 500 m, only an ADr [7,8,15].
transmitting higher than 0 dBm can be sensed within the
whole target zone. This figure also proposes an optimum
C. Localizing Amateur Drones
altitude at which the lowest possible transmit power is
detectable. As a matter of fact, if the SDr goes higher the Once the ADr has been detected and its RSS has been
LoS probability increases resulting in better channel quality measured, we can proceed to estimate its location. In our
as explained in the previous section, however the link length simulation we assume 3 SDrs positioned as vertices of
also increases deteriorating the channel quality due to path an equilateral triangle with sides of length l and equal
loss. These opposite effects are balanced at the optimum adjustable altitude h. Each SDr has its own coverage radius
altitude shown in the figure. From this figure it can be and is able to detect any ADr separately. The intersection of
seen that, as the target region becomes larger the minimum the three coverage areas produces the no-fly zone associated
required transmit power of ADrs and the optimum altitude with the three SDrs combined.
 6
20 2500
Coverage Radius = 500 m
Coverage Radius = 750 m
15
Coverage Radius = 1000 m
2000
ADr’s Transmit Power [dBm]

10 ADr’s PTx = -10 dBm

Coverage Radius [m]

ADr’s PTx = -5 dBm
1500 ADr’s PTx = 0 dBm
5

0
1000

−5
500
−10

−15 0
0 500 1000 1500 2000 0 1000 2000 3000 4000
Altitude, h [m] Altitude, h [m]
(a) (b)
Fig. 3. (a) For a target zone indicated by its radius, the range of sensing powers are influenced significantly by the altitude of flying SDrs.
(b) For an assumed minimum ADr’s transmit power an SDr can fly at an optimum altitude to extend its monitoring region compared to
ground surveillance.

Unlike the G2G RSS-based localization scenarios which shadowing effect will decrease, concurrently, the resolution2
suffer from shadowing as the main source of error, intro- will also decrease. In the case of low resolution, any small
ducing the altitude in G2A and A2A scenarios as a third variation will bring a large localization error. Therefore,
dimension promises to overcome the shadowing due to the based on the behavior of the path loss model, the existence
high PLoS and hence minimize the localization error. In this of an optimal altitude is investigated as shown in Figure 4.
subsection, we present the localization of ADrs by using Our results show that for a targeted no-fly zone, an optimum
SDrs as shown in Figure 1. As illustrated in the figure, we altitude h minimizing the localization error is present. This
target localizing any ADr that would fly in the defined no- optimal altitude is shown in Figure 4a. As one can see,
fly zone by means of the RSS-based distance estimation. In an optimal altitude for minimizing the estimated location
order to define the position of the ADr, distances to three error exists at h = 800 m for a coverage radius of 1000 m.
different SDrs need to be estimated using RSS. However, as Moreover, it can be seen that when the SDrs are at the low
the transmit power from the ADr is unknown, one can define altitude of < 800 m, the RSS is exposed to high shadowing
(min) (max)
a constant PTx ≤ C ≤ PTx that represents the possible resulting a relatively high estimation error assuming the
transmit power. Accordingly, the estimated distance between same coverage. On the other hand, high altitude means
the ADr and the SDr is equal to dˆ+ δ where δ is a constant low resolution in the RSS-distance curve where any low
that represents the uncertainty due to the unknown transmit shadowing causes a relativity high estimation error. It is
power. Subsequently, after estimating the distance between worth noting that, the coverage radius here can be connected
the ADr and three SDrs, we will end up with a sphere of to the PTx , since the lower the PTx of ADr, the smaller the
radius δ in which the ADr is located. A representation of coverage radius of the SDr at which ADrs are detectable.
δ is shown in Figure 4a which can cause under- or over- In addition to the altitude dependence, the localization
estimation of the distance. accuracy is affected by the distance l between the SDrs.
Figure 4b illustrates the location estimation error against
the distances l under an assumption of the equidistant SDrs
As shown in previous sections, the altitude of the drones positioning. As shown in the figure, when l is relatively
has a significant influence on the model representing the small, the intersection zone between the estimated distances
received power and hence, the accuracy of the estimated of the three SDrs will be large leading to a relatively
location of the ADr. In fact, for any SDr, the variations high localization error. Moreover, increasing l improves
of both the the path loss exponent and shadowing standard the localization accuracy where an accuracy of 100 m is
deviation with the altitude are modeled based on statistical achieved at l = 300 m for the same h of 1000 m. Eventually,
representations given in equations (2) and (3), respectively.
The shadowing effect at low values of h will be relatively 2 The resolution is the ability to distinguish two different distances from
high causing large localization errors. As h increases, the two different RSS measurements
 7
500 300
Coverage Radius = 500 m
Coverage Radius = 750 m
Coverage Radius = 1000 m 250
400
Location estimation error [m]

Location estimation error [m]

200
300

150

200 δ
100

100
50
h = 1000 m
h = 500 m
0 0
500 1000 1500 2000 200 400 600 800 1000 1200
Altitude, h [m] Distance between SDrs, l [m]
(a) (b)
Fig. 4. Localization of ADrs is affected by the altitude h and the distance l between SDrs: (a) Localization error as a function of the
SDrs altitude with the uncertainty constant δ. (b) Localization error as function of distance l between SDrs.

as l keeps increasing, the distance will become too large to C. RF fingerprinting for ADr identification
give an acceptable resolution for distance estimate, making Even though we have explained various state-of-the-art
a low localization error impossible. techniques for ADr detection and technology classification,
V. R ESEARCH D IRECTIONS AND F UTURE W ORK identifying a drone with passive RF monitoring is quite chal-
lenging. For detecting ADrs, temporal and spatial wireless
Gradually, drones are gaining a lot of momentum as
transmission statistics should be derived from the received
an ingrained part of future wireless technologies making
detected signal which should be further associated with a
surveillance of amateur drones very important. In this section
particular drone. Detailed studies should be done to enable
we address some open problems as future work towards
and improve RF fingerprinting for drone surveillance. We
auxiliary reliable surveillance of ADrs.
believe drone identification with low false identification
A. Experimental channel models validation rates can be achieved by combining RF localization and
fingerprinting.
Although, some interesting works have been carried out to
model the G2A channel characteristic using measurements
or simulations, they only considered either non-urban en- D. Mobility aided surveillance
vironments or relatively high altitudes. In fact, a generic In order to decrease the number of surveillance drones in
channel model that reflects characteristics of both A2A a given region, a mobile drone can be deployed. A mobile
and G2A channels and dependency on the elevation angle drone will not only increase the number of accessible ground
including the shadowing effect is required. To this end, a station but it can also localize other drones by collecting
measurement campaign in order to validate the different measurements at different locations. However, the cost to be
proposed simulation-based models and define a generic one payed is the delay although the speed and the trajectory of
is still one of the future plans. movement can be optimized for the minimum delay.
B. Tracking of ADrs
In this work we consider localizing the ADrs at a given E. Network lifetime
time, however, as the ADrs are mobile in nature, localiza- In order to increase network lifetime we need to optimize
tion must be considered over time in order to track the the power consumption. Since mechanical power is the main
ADrs’ movement. To this end, the processing time between source of the energy cost, power consumption can be im-
detecting the ADr, defined a valid RSS measurement and proved by optimizing the flying trajectory. Considering the
estimating its position need to be minimized. In fact, this trajectory of movement, the duration of communication, the
time has to be lower than the time needed for the ADr to payload weight and the battery size, the lifetime of a drone
move a certain distance that defined the required localization is limited to less than one hour. Therefore, an alternative
accuracy. solutions such as solar cells (e.g., Facebook Aquila Drone)
 8
for providing the required mechanical energy is of high [12] N. Goddemeier and C. Wietfeld, “Investigation of Air-to-Air Channel
importance. Nevertheless, adding solar cells also increases Characteristics and a UAV Specific Extension to the Rice Model,” in
IEEE Globecom Workshops (GC Wkshps), Dec. 2015, pp. 1–5.
the rate of energy consumption as more weight has to be [13] A. Al-Hourani, S. Kandeepan, and A. Jamalipour, “Modeling air-to-
carried on the drone. ground path loss for low altitude platforms in urban environments,” in
IEEE Globecom Workshops (GC Wkshps), Dec. 2014, pp. 2898–2904.
[14] D. W. Matolak and R. Sun, “Air-ground channel characterization
VI. C ONCLUSION for unmanned aircraft systems-part iii: The suburban and near-urban
environments,” IEEE Transactions on Vehicular Technology, vol. 66,
In this article we have considered a network of SDrs no. 8, pp. 6607–6618, 2017.
aiming to sense the presence of ADrs and localize them [15] M. Mozaffari, W. Saad, M. Bennis, and M. Debbah, “Efficient
deployment of multiple unmanned aerial vehicles for optimal wireless
over a no-fly zone by means of the passive RF sensing. coverage,” IEEE Communications Letters, vol. 20, no. 8, pp. 1647–
An overview of the state-of-the-art RF passive sensing and 1650, 2016.
detection showed that an SNR above 0 dB is required for
accurate detection of ADrs. However, using aerial based
sensors improves the received SNR due to better channels
between ADrs flying at low altitudes and SDrs. Therefore,
the characteristics of the channel between SDrs and ADrs
have been thoroughly studied considering the worst case Mohammad Mahdi Azari
scenario at which the ADrs are 2 m above the ground. Our ([email protected]) has received the
results show that a tenfold increase of the coverage radius B.Sc. and M.Sc. degrees in electrical and
communication engineering from University
and a 25 dB reduction of the minimum detectable power of Tehran, Tehran, Iran. Currently he is a
can be achieved by flying the SDrs at the optimal altitude. Ph.D. candidate at the Department of Electrical
Furthermore, it has been shown that 4 times better localiza- Engineering, KU Leuven, Belgium. His main
research interests include unmanned aerial
tion accuracy is gained by careful optimizing the altitude of vehicle (UAV) communication and networking,
the SDrs for a given no-fly zone. We expect that academic modeling and analysis of cellular networks, and
and industrial research and development activities can use mmWave communication. He has also received
Iran’s National Elites Foundation (INEF) Award.
the proposed framework to address the drone surveillance
challenges introduced in the paper.

R EFERENCES
[1] A. Solodov, A. Williams, S. Al Hanaei, and B. Goddard, “Analyzing
the threat of unmanned aerial vehicles (uav) to nuclear facilities,”
Hazem Sallouha ([email protected])
Security Journal, Springer, pp. 1–20, April 2017.
received the B.Sc. degree in electrical engineering
[2] A. Rozantsev, S. Sinha, D. Dey, and P. Fua, “Flight dynamics-based
from Islamic University of Gaza, Palestine, in
recovery of a uav trajectory using ground cameras,” in Conference on
2011, the M.Sc. degree in electrical engineering
Computer Vision and Pattern Recognition (CVPR), July 2017.
majoring in wireless communications from Jordan
[3] J. Kim, C. Park, J. Ahn, Y. Ko, J. Park, and J. C. Gallagher, “Real-
University of Science and Technology, Jordan
time UAV sound detection and analysis system,” in IEEE Sensors
in 2013. Currently he is a Ph.D. candidate at
Applications Symposium (SAS), March 2017, pp. 1–5.
the Department of Electrical Engineering, KU
[4] F. Hoffmann, M. Ritchie, F. Fioranelli, A. Charlish, and H. Griffiths,
Leuven, Belgium. His main research interests
“Micro-Doppler based detection and tracking of UAVs with multistatic
include localization techniques, communications
radar,” in IEEE Radar Conference (RadarConf), May 2016, pp. 1–6.
with unmanned aerial vehicles, Internet of things
[5] P. Nguyen, M. Ravindranatha, A. Nguyen, R. Han, and T. Vu, “Inves- networks and machine learning algorithms for localization.
tigating Cost-effective RF-based Detection of Drones,” in Proceedings
of the 2nd Workshop on Micro Aerial Vehicle Networks, Systems, and
Applications for Civilian Use. ACM, June 2016, pp. 17–22.
[6] H. Sallouha, A. Chiumento, and S. Pollin, “Localization in Long-range
Ultra Narrow Band IoT Networks using RSSI,” in IEEE International
Conference on Communications (ICC), May 2017, pp. 6076–6081.
[7] M. M. Azari, F. Rosas, K.-C. Chen, and S. Pollin, “Joint sum-rate
and power gain analysis of an aerial base station,” in IEEE Globecom
Workshops (GC Wkshps), Dec. 2016, pp. 1–6. Alessandro Chiumento
[8] ——, “Ultra reliable uav communication using altitude and cooper- ([email protected])
ation diversity,” arXiv preprint arXiv:1705.02877, Accessed on: Apr. received his Ph.D. degree in cellular network
2017. management from Imec, Leuven, Belgium, in
[9] S. Rajendran, R. Calvo-Palomino, M. Fuchs, B. Van den Bergh, 2015. He is currently with the Department of
H. Cordobés, D. Giustiniano, S. Pollin, and V. Lenders, “Electrosense: Electrical Engineering, Katholieke Universiteit
Open and Big Spectrum Data,” arXiv preprint arXiv:1703.09989, Leuven. His research interests include massive
Accessed on: March. 2017. machine-to-machine communication, channel
[10] T. J. O’Shea, J. Corgan, and T. C. Clancy, “Convolutional radio prediction, very dense networks, and the
modulation recognition networks,” in International Conference on application of machine learning to theoretical
Engineering Applications of Neural Networks. Springer, Feb. 2016, problems in telecommunication and information
pp. 213–226. management.
[11] N. West and T. O’Shea, “Deep architectures for modulation recogni-
tion,” in IEEE International Symposium on Dynamic Spectrum Access
Networks (DySPAN), March 2017, pp. 1–6.
 9
Sreeraj Rajendran received his Masters degree Sofie Pollin Sofie Pollin obtained her PhD at
in communication and signal processing from KU Leuven in 2006. She continued her research
the Indian Institute of Technology, Bombay, in on wireless communication at UC Berkeley. In
2013. He is currently pursuing the Ph.D. degree November 2008 she returned to imec to become
in the Department of Electrical Engineering, KU a principal scientist in the green radio team.
Leuven. Before joining KU Leuven, he worked Since 2012, she is tenure track assistant professor
as a senior design engineer in the baseband team at the electrical engineering department at KU
of Cadence and ASIC verifidetecation engineer in Leuven. Her research centers around Networked
Wipro Technologies. His main research interests Systems that require networks that are ever more
include machine learning algorithms for wireless dense, heterogeneous, battery powered and spec-
and low power wireless sensor networks. trum constrained.

Evgenii Vinogradov received the Dipl. Engineer

degree in Radio Engineering and Telecommunica-
tions from Saint-Petersburg Electrotechnical Uni-
versity (Russia), in 2009. After several years of
working in the field of mobile communications, he
joined UCL (Belgium) in 2013, where he obtained
his Ph.D. degree in 2017. His doctoral research
interests focused on radio propagation channel
modeling. In 2017, Evgenii joined the electrical
engineering department at KU Leuven (Belgium)
where he is working on wireless communications
with UAVs.