Survey on UAV Cellular Communications: Practical Aspects, Standardization Advancements, Regulation, and Security Challenges

by   Azade Fotouhi, et al.

The rapid growth of consumer Unmanned Aerial Vehicles (UAVs) is creating promising new business opportunities for cellular operators. On the one hand, UAVs can be connected to cellular networks as new types of user equipment, therefore generating significant revenues for the operators that guarantee their stringent service requirements. On the other hand, UAVs offer the unprecedented opportunity to realize UAV-mounted flying base stations that can dynamically reposition themselves to boost coverage, spectral efficiency and user quality of experience. Indeed, the standards bodies are currently exploring possibilities for serving commercial UAVs with cellular networks. Industries are beginning to trial early prototypes of flying base stations or user equipments, while academia is in full swing researching mathematical and algorithmic solutions to many interesting new problems arising from flying nodes in cellular networks. In this article, we provide a comprehensive survey of all of these developments promoting smooth integration of UAVs in cellular networks. Specifically, we survey the types of consumer UAVs currently available off-the-shelf, the interference issues and potential solutions addressed by standardization bodies for serving aerial users with existing terrestrial base stations, the challenges and opportunities for assisting cellular communications with UAV-based flying relays and base stations, the ongoing prototyping and test bed activities, the new regulations being developed to manage the commercial use of UAVs, and the cyber-physical security of UAV-assisted cellular communications.



There are no comments yet.



Safeguarding Wireless Network with UAVs: A Physical Layer Security Perspective

Integrating unmanned aerial vehicles (UAVs) into future wireless systems...

Enabling 5G on the Ocean: A Hybrid Satellite-UAV-Terrestrial Network Solution

Current fifth generation (5G) cellular networks mainly focus on the terr...

Airborne Urban Microcells with Grasping End Effectors: A Game Changer for 6G Networks?

Airborne (or flying) base stations (ABSs) embedded on drones or unmanned...

A Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements

As an emerging field of aerial robotics, Unmanned Aerial Vehicles (UAVs)...

Handover Experiments with UAVs: Software Radio Tools and Experimental Research Platform

Mobility management is the key feature of cellular networks. When integr...

Potential Data Link Candidates for Civilian Unmanned Aircraft Systems: A Survey

This survey studies the potential data link candidates for unmanned airc...

Mobility for Cellular-Connected UAVs: challenges for the network provider

Unmanned Aerial Vehicle (UAV) technology is becoming more prevalent and ...
This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.

I Introduction

From aerial photography to search-and-rescue to package delivery — the use cases of consumer UAVs (a.k.a. drones) are exploding. According to a report from Federal Aviation Administration, 2.5 million drones were sold in 2016 [1]. It is expected that new use cases will continue to emerge, fuelling further growth in UAVs. As many of these use cases would benefit from connecting the UAVs to the cellular networks for better control and communications, the growth in the UAV market is expected to bring new promising business opportunities for cellular operators.

The Third Generation Partnership Project (3GPP), which oversees the standards activities for cellular networks, has recently concluded a study item [2] to explore the challenges and opportunities for serving the UAVs as a new type of user equipment (UE), referred to as aerial UE. An interesting finding of this study is that the enhanced line-of-sight (LOS) between aerial UE and ground base stations (BSs) would significantly increase interference in the system, which calls for new strategies to seamlessly accommodate both aerial and ground UEs in the same system. A variety of new techniques have already been proposed to address the interference issue, which show promising results.

While 3GPP is mainly concerned with connecting UAVs to cellular networks, industry and academia are advancing to the next level of research and development that promises to harness the full potential of UAVs communications. In particular, they are exploring the unprecedented opportunity to realize UAV-mounted flying relays and BSs that can dynamically reposition themselves to boost coverage, spectral efficiency and user quality of experience. Major vendors have already field trialled their prototypes to demonstrate the proof-of-concept of such UAV-mounted flying BSs [3, 4]. A large number of papers have been published in recent years proposing novel algorithms to optimize positioning and mobility of flying relays and BSs.

Given the significant momentum and recent activities promoting UAV in cellular networks, it is timely to survey the field. Although several survey articles on UAV have been published in recent years, none of them had focused on the practical aspects of cellular UAV communications. For example, the surveys in [5, 6] focussed on forming ad hoc networks between many UAVs in the sky. Hayat et al., [7] surveyed communications demands for various applications of UAV and analysed the suitability of existing wireless technologies, including Bluetooth, Zigbee, Wi-Fi, WiMAX, and cellular, to meet these demands. In a magazine paper, Sekander et al., [8] analyzed the opportunities for drones to assist cellular networks, but specialized on combining drones from different altitudes to form a multi-tier drone network. In another magazine paper, Zeng et al., [9] surveyed the issues and opportunities for using drones to assist wireless networks in general without specific focus on cellular networks. Recently, Mozaffari et al., [10] delivered a comprehensive tutorial on UAV wireless communications. Our work complements their vision by covering a variety of cellular-specific issues such as the relevant 3GPP developments, vendor prototypes of flying BSs, regulations and cyber-security issues affecting cellular UAVs, and the potential impacts of UAV adoption on the cost and business models of cellular networks.

Fig. 1: Taxonomy of the survey.
(a) Parrot Swing (front view).
(b) Kogan Nano Drone.
(c) Parrot Disco.
(d) DJI Spreading Wings S900.
(e) Scout B-330 UAV helicopter.
(f) Predator B.
Fig. 2: Images of different UAV types.

A graphical illustration of the detailed taxonomy of our survey is presented in Fig. 1. Specifically, we survey the types of consumer UAVs currently available off-the-shelf highlighting their potential roles within cellular networks (Section II), the 3GPP developments regarding interference issues and solutions for serving aerial users (Section III), the challenges and opportunities for assisting cellular communications with UAV-mounted flying relays and base stations (Section IV), the most advanced UAV prototyping and field trial examples (Section V), the new regulations being developed to manage the commercial use of UAVs (Section VI), the cyber-physical security issues for UAV-assisted cellular communications (Section VII), and finally an overlook of the most promising future research directions (Section VIII) followed by the conclusions (Section IX).

Micro (weight100g) Very Small (100gweight2kg) Small (2kgweight25kg) Medium (25kgweight150kg) Large (weight150kg)
Model Kogan Nano Drone Parrot Disco DJI Spreading Wings S900 Scout B-330 UAV helicopter Predator B
Ref. [11] [12] [13] [14] [15]
Illustration Fig. 2(b) Fig. 2(c) Fig. 2(d) Fig. 2(e) Fig. 2(f)
Weight 16g 750g 3.3kg 90kg 2223kg
Payload N/A N/A 4.9kg 50kg 1700kg
Flying Mechanism Multi-rotor Fixed-wing Multi-rotor Multi-rotor Fixed-wing
Range 50-80m 2km N/A N/A 1852km
Altitude N/A N/A N/A 3km 15km
Flight Time 6-8min 45min 18 min 180min 1800min
Speed N/A 80km/h 57.6km/h 100km/h (horizontal) 482km/h
Power Supply 160mAh Li-battery 2700mAh 3-cell LiPo Battery 12000mAh LiPo Battery 21kW Gasoline 712kW 950-shaft-horsepower Turboprop Engine
Application Recreation Recreation; suitable to carry cellular UEs Professional aerial photography and cinematography; suitable to carry cellular BSs or UEs Survey (data acquisition), HD video live stream; can carry cellular BSs or UEs Armed reconnaissance, airborne surveillance, and target acquisition
TABLE I: Characteristics of different drone types

Ii UAVs Types and Purposes

UAVs, commonly known as drones, are available in different sizes and specifications, as illustrated in Fig. 2. They possess unique channel characteristics and can be deployed quickly whenever needed, which makes them promising candidates for providing cellular connectivity. In this section, typical characteristics of drones are summarized and explained (see Table I), which in particular focus on their influence on UAV-aided cellular communications.

Ii-a Payload

Payload refers to the maximum weight a drone can carry, which measures its lifting capability. Payloads of drones vary from tens of grams up to hundreds of kilograms [16]. The larger the payload, the more equipment and accessories can be carried at the expense of a larger drone size, higher battery capacity and shorter duration in the air. Typical payloads include video cameras, which could be used for reconnaissance, surveillance and commercial purposes [17]. When assisting cellular communications, drones can carry cellular UEs such as mobile phones or tablets, whose weight is usually less than 1 kilogram [18]. BSs or remote radio heads (RRHs) can also be carried by or mounted on drones to provide cellular services. In this case, payload of drones should be at least a few kilograms.

Ii-B Flying Mechanism

Depending on their flying mechanisms, drones can be classified into three types:

  • Multi-rotor drones (also known as rotary-wings drones) allow vertical take-off and landing, and can hover over a fixed location to provide continuous cellular coverage for certain areas. This high manoeuvrability makes them suitable for assisting cellular communications, since they can deploy BSs at the desired locations with high precision, or fly in a designated trajectory while carrying BSs. However, multi-rotor drones have limited mobility and consume significant power as they have to fight against gravity all the time.

  • Fixed-wing drones can glide over the air, which makes them significantly more energy efficient and able to carry heavy payload. Gliding also helps fixed-wing drones to travel at a faster speed. The downsides of fixed-wing drones are that they require a runway to take off and land as vertical take-off and landing are not possible [19], and they cannot hover over a fixed location. Fixed-wing drones are also more expensive than multi-rotor drones.

  • Hybrid fixed/rotary wing drones have recently reached the market to provide a compromise between the two above-mentioned drone types. An illustrative example of a hybrid fixed/rotary wing drone is the Parrot Swing shown in Fig. 2(a), which can quickly reach its destination by gliding over the air and then hover using four rotors.

Ii-C Range and Altitude

The range (one hop) of a drone refers to the distance from which it can be remotely controlled. The range varies from tens of meters for small drones to hundreds of kilometers for large drones. Altitude here refers to the maximum height a drone can reach regardless of the country-specific regulations. The maximum flying altitude of a given drone is a critical parameter for UAV-aided cellular communications, since a UAV BS needs to vary its altitude to maximize the ground coverage and satisfy different quality of service (QoS) requirements[20]. Overall, aerial platforms can be classified into two types depending on their altitude:

  • Low-altitude platforms (LAPs) are usually employed to assist cellular communications since they are more cost-effective and allow fast deployment. Moreover, LAPs usually provide short-range line-of-sight (LOS) links that can significantly enhance the communication performance [9].

  • High-altitude platforms (HAPs) such as balloons can also provide cellular connectivity [9]. Compared to LAPs, HAPs have wider coverage and can stay much longer in the air. However, HAPs require higher cost.

Ii-D Speed and Flight Time

Small drones typically travel at speeds below 15 m/s [21], while large drones can impressively reach 100 m/s [9]. When a UAV BS/relay flies in a designated trajectory to maximize its energy and spectral efficiency, its speed needs to be carefully considered if the trajectory requires frequent turns. The trade-off between a drone’s speed and its turning agility is revealed in [22]. The maximum time a drone can spend in the air without recharging or refueling is referred to as its flight time or endurance. Small commercial drones usually have a flight time of 20-30 minutes, while some large drones can last for hours [23]. Emerging technologies have prolonged the endurance of small drones. For example, the Skyfront Tailwind drone can achieve an endurance of up to 4.5 hours with hybrid-electric power sources [24]. Nevertheless, the limited endurance of the existing off-the-shelf UAVs is currently one of the major practical factors restricting their full-scale deployment in cellular networks.

Ii-E Power Supply

A drone’s power supply significantly determines its endurance. While rechargeable batteries power most commercial drones, some large drones can be powered by fuels such as gas for longer flight times [25]. Employing solar energy to power drones is also a promising technique [26]. For drone-mounted BSs, power supply needs to support the functionality of both the drone and its on-board equipment such as antenna array, amplifier, circuits, etc.

Ii-F An Example of Drone Classification

Civil aviation authorities usually classify drones based on their gross weights. Table I demonstrates such weight-based classification as adopted by Civil Aviation Safety Authority (CASA) Australia [27] by listing typical drones and their features, which are depicted in Fig. 2(b)(f).

Iii 3GPP Standardization:
Enabling UAV Cellular Communications

The cellular industry has recognized the importance of providing support to low-altitude UAVs for enabling beyond LOS control and establishing a reliable communication [2, 28, 29, 30, 31]. As a result, the third generation partnership group (3GPP) defined a study item (SI) in March 2017 with four fundamental objectives: A) the understanding of the UAV traffic requirements; B) the development of a channel model to characterize air-to-ground propagation characteristics; C) the determination of whether the current LTE infrastructure could be reused to provide cellular service to aerial devices; D) the definition of the enhancements required to effectively serve UAVs building up on LTE Release 14 functionalities [32]. The SI finalized in December 2017 and the main results in each of the above four areas are summarized in the following [2].

Iii-a UAV Traffic Requirements

The 3GPP identified the traffic types that cellular networks should cater for UAVs flying between ground level and 300 meters. These are summarized in Table II and can be classified into three categories: 1) synchronization and radio control, 2) command & control, and 3) application data.

  1. The information contained within the synchronization and radio control messages is essential for a successful association and connectivity to the network. The transmission of these signals must be robust enough to guarantee that they can be decoded by flying UAVs. Examples of synchronization and radio control signalling include primary and secondary synchronization signals (PSS/SSS) and the physical downlink control channel (PDCCH), respectively.

  2. Command & control (C&C) traffic enables beyond line-of-sight UAV piloting and has strict quality of service requirements (QoS) in terms of latency and reliability. Cellular operators have identified an attractive business opportunity in the management of this traffic, since it can be offered as a complementary network service to organizations interested in reliably controlling their UAVs.

  3. While downlink data traffic is predominant in existing cellular communications, UAV application data transmissions are expected to be uplink-dominated. Transferral of live video streaming data and photos captured by camera-equipped UAVs contribute towards this traffic imbalance.

Data Type Data Rate Critical?
DL Synchronization (PSS/SSS) N/A
Radio control (PDCCH)
Command and control (C&C) 60-100 kbps
UL Command and control (C&C) 60-100 kbps
Application data Up to 50 Mbps
TABLE II: UAV communication requirements [2, 30]

Iii-B Channel Modelling

In order to characterize the performance of existing cellular networks when serving both ground and aerial devices, the 3GPP developed a UAV-specific statistical channel model building upon that defined in [33]. This channel model complements those developed by the academic community. These academic models are summarized in Table III and Table IV into three different categories (Air to Ground (A2g), Air to Air (A2A), and Ground to Air (G2A)) for completeness. The 3GPP-proposed models for rural-macro (RMa), urban-macro (UMa), and urban-micro (UMi) BS deployments are the result of a large number of measurement campaigns carried out by the standard-contributing companies and its main UAV-related features can be summarized as follows:

  • UAV spatial placement: The 3GPP defines five different cases depending on the density of UAVs in the network, i.e., it considers the deployment of

    UAVs per cellular sector (out of a total of 15 mobile devices). Airborne devices are uniformly distributed between 0 and 300 meters and travel at a speed of 160 km/h.

  • LOS probability:

    The LOS probability between ground BSs and UAVs grows as the latter increase their height. Remarkably, UAVs flying higher than 100 meters are considered to be in LOS with all the cellular BSs deployed in the network in the UMa scenario.

  • Path loss: The 3GPP model also captures the fact that the path loss exponent of ground-to-aerial links generally decreases as UAVs increase their height. Indeed, UAVs in LOS with their BSs experience a path loss similar to of free-space propagation ().

  • Shadowing:

    The standard deviation of the log-normally distributed shadowing gain diminishes for increasing UAV heights, provided that the considered UAV-BS pair is LOS.

  • Fast-fading model: Three different alternatives with a varying degree of implementation complexity are considered in [2], namely, 1) a variation of the cluster delay-based channel model developed in [33] with UAV-specific channel characteristics such as the existence of a specular reflection on the building roof for the UMa scenario, 2) an approach where the mean and standard deviation of the large scale parameters defined in [33] (delay spread, angular spreads of departure and arrival, and K-factor) are adjusted, and 3) a simpler alternative where, when compared with the channel model of [33], only the K-factor is adjusted.

Type Ref. Frequency/Protocol Altitude Environment Experiment Details Objective
A2G [34] PCS, AWS, and 700MHz below 120m mixed suburban (California) Custom designed quadrotor drone (5m/s) Enhance the understanding of aerial communications
[35] 970Mhz (L-band), and 5060MHz (C-band) 560m Near-urban (Cleveland) S-3B Viking aircraft Characterizing the channel for air to ground communication
[36] 850Mhz, LTE 15,30,60,90, and 120 m suburb in Victoria, Australia one commercial UAV, 4.8m/s, sony Xperia phone for logging Modeling the excessive path loss exponent over the terrestrial model
[37] 800MHz, LTE 15,30,60, and 120 m Denmark One commercial UAV , 15km/h Modeling the path loss exponent and shadowing
A2A [38] 2.4GHz, IEEE 802.11 Below 50m - Two AscTec Firefly Hexacopter UAVs Studying the impact of distance
[39] ZigBee 802.145.4 Below 20m - Two Hexacopters Measuring path loss exponent
G2A [40] 802.11b/g 75m Fix-wing fuselag Measuring the diversity in G2A links
[39] ZigBee 802.145.4 Below 20m - Two Hexacopters Measuring path loss exponent
TABLE III: Experimental studies for channel modeling
Type Ref. Frequency Altitude Environment Simulation Details Description
A2G [41] 700MHz, 2000 MHz and 5800MHz 200m-3000m Suburban, Urban, Dense and Highrise Urban MATLAB, Wireless InSite, One quasi-stationary UAV Finding a generic path loss model based on urban parameters
[42] 5GHz 100m, 200m, 500m, 1000m and 2000m Bristol area with irregular street Ray tracing software Modeling LOS probability based on building geometry
A2A [43] Two UAVs (22m/s) in the simulated area Modeling the packet dropout using Rician channel model
[44] 2.4GHz (80MHz BW) Three UAVs, one as a relay Modeling the bit rate over different channel models
G2A [45] 5GHz MATLAB, one aircraft (300m/s) Studying the characters of channel such as the Doppler, and the type of fading
TABLE IV: Simulation studies for channel modeling

Iii-C UAV Performance Analysis: Interference Challenges

Based on the above traffic and channel characterizations, the companies involved in the SI evaluated the performance of cellular networks serving both airborne and ground users (GUEs). These studies demonstrate that UAVs are more likely to undergo downlink and uplink interference problems than GUEs [2, 30, 31, 46, 47]. This is mainly due to two factors, namely, that flying UAVs are likely to be in line-of-sight with a large number of base stations (BSs), and that the majority of these BSs are downtilted, since their deployment has been optimized for providing coverage to GUEs. This impacts all phases of the cellular communication, i.e., 1) association and handover, 2) downlink transmissions, and 3) uplink transmissions:

Fig. 3: 2D location of 150 meter-high UAVs (red dots) associated to a three-sector BS site located at the origin [30]. Hexagons illustrate the ground sectors covered by each BS.
  1. Association and handover: In contrast to GUEs, flying UAVs do not generally associate to their physically closest BS. This is because cellular BSs generally focus their main antenna beam towards the center of their ground coverage area. Instead, the association of airborne devices is dominated by the sidelobes of their directive BS antennas [30, 2]. This can be observed in Fig. 3, which adopts the perspective of a tri-sector BS site located in the center of the scenario and illustrates the 2D location of its associated 150 meter-high UAVs (represented by red dots). These UAVs are clustered in three angular regions, which are consistent with the orientations of the central BSs (, , and ). Different association ranges highlighted in green can be observed, each corresponding to different antenna sidelobes of the -downtilted BSs. Instead, the white regions delimit the areas where the antenna gain of the considered BSs is smaller than -30 dB, i.e., close to the radiation nulls of the downtilted antennas. In these areas, UAVs associate to BSs other than those located in the origin because they perceive an insufficient signal strength from the latter. As a result of the existence of these non-contiguous association regions and the reception of high-power interfering reference signals, UAVs experience increased outage and handover failure probabilities when compared to ground devices.

  2. Downlink transmissions: As illustrated in Fig. 4, a UAV receives line-of-sight transmissions from a large number of BSs when increasing its altitude. Indeed, measurement campaigns have demonstrated that airborne devices flying at around 100 meters can receive signals from BSs located up to 10 kilometers away [48]. This entails that a given UAV can undergo a substantial amount of interference from a multiplicity of ground BSs that transmit towards other GUEs or UAVs. Consequently, downlink transmissions towards aerial devices generally suffer from poor signal-to-interference-plus-noise ratios (SINRs) more often than their ground counterparts.

    Fig. 4: Illustration of the UAV interference challenge in cellular networks: UAVs flying above the building clutter (e.g., the red-coloured UAV) generate/perceive interference towards/from a multiplicity of line-of-sight BSs [30, 31]. Instead, UAVs flying at low altitudes (e.g., the blue-coloured UAV) only generate/perceive interference towards/from nearby BSs.
  3. Uplink transmissions: UAVs’ good propagation conditions with a multiplicity of BSs also impact the network’s uplink performance. This is because airborne devices transmitting data towards their serving BS can generate strong interference to a variety of ground BSs, an issue that becomes critical in networks with a large number of UAVs. As shown in Fig. 4, this UAV-generated interference has the potential of damaging the uplink communication of existing GUEs, who are more prone to have non-line-of-sight links with their serving BSs. Overall, both aerial and ground devices have been shown to experience diminished SINRs in networks with a substantial number of UAVs.

Iii-D Enhancing UAV Communications

To address the interference challenges described in Sec. III-C, the 3GPP examined a number of complementary interference mitigation techniques:

  1. Association and handover:

    • UAV location and flight plan knowledge can be leveraged to facilitate the handover procedure, e.g., by anticipating BS candidates for a potential handover.

    • Enhancement of existing report mechanisms through the definition of UAV-specific handover triggering conditions and an optimized control of the reporting load.

    A detailed evaluation of these UAV mobility enhancement techniques is part of the work item (WI) study defined in [49], as detailed in Sec. III-E.

  2. Downlink transmissions:

    • The full dimension MIMO (FD-MIMO) multi-antenna BSs defined in LTE Release 13 enhance the performance of UAV communications thanks to a) their beamformed transmissions, which allow reducing the amount of interference generated towards the constrained spatial regions where UAVs lie, and b) their spatial multiplexing capabilities, which in turn enable a better utilization of the precious time/frequency resources.

    • UAVs with directional antennas and beamforming capabilities contribute to reduce the number of downlink interferers perceived by aerial devices. These interference mitigation gains can be further complemented with a boost of the useful signal power in UAVs beamsteering towards their serving BS. Clearly, this solution entails a complexity increase in the design of hardware UAV transceivers.

    • Cooperative multipoint (CoMP) can convert the harmful line-of-sight BS interferers into useful signal contributors. Indeed, both UAV control and data transmissions benefit from intra- and inter-site cooperation. However, the gains attainable with this approach are limited in practical deployments due to the large number of BSs interfering towards a given UAV and the increased inter-BS signalling load.

  3. Uplink transmissions:

    • Uplink power control is essential to harmonize the coexistence among GUEs and aerial devices. In this line, approaches that define different fractional path loss compensation factors and offsets for UAVs and GUEs proved effective in mitigating the interference generated by UAVs.

    • Full dimension MIMO (FD-MIMO) can also benefit uplink transmissions by enabling spatial separation of ground and aerial users, which have clearly distinguishable propagation characteristics.

    • UAVs with directional antennas and beamforming capabilities generate a diminished amount of interference to the GUE-generated uplink transmissions owing to a potential decrease in both the number of interfered BSs and the UAV transmission power.

As the reader might have noticed, updated information about the flying status of a mobile device is required to effectively implement some of the above-mentioned solutions. The 3GPP considered a variety of device- and network-based solutions to acquire this information. Among others, these include the use of explicit UAV identification signalling, the exploitation of mobility history information, or the employment of measurement reports from mobile devices.

Iii-E 3GPP Conclusions and Way Forward

Overall, the 3GPP concluded that LTE cellular operators should be capable of satisfactorily serving UAVs with their existing infrastructure as long as the number of airborne devices is limited and provided that the network is not severely loaded [2]. However, it also acknowledges that there could be challenges in the management of uplink/downlink interference and mobility. Indeed, these observations led to the definition of a WI in December 2017 [49]. This WI targets to provide further enhancements to LTE in the following areas:

  • Uplink power control, with the possibility of adapting the existing fractional power control mechanism and assigning UAV-specific parameters;

  • Interference detection, by defining new UAV reporting events and controlling their frequency;

  • Subscription-based access, through new network signaling (S1/X2) that will enable to identify subscribed UAVs;

  • Mobility, exploiting additional information such as UAV flight plan;

  • Signaling, to identify the status of airborne devices as well as their radio capabilities.

Ultimately, it is expected that the 3GPP continues its work on the UAV space after the conclusion of this WI, possibly through the definition of a new SI focused on the use of 5G-based solutions like massive MIMO [30, 50, 31, 51, 52].

Iv Aerial Base Stations: Challenges and Opportunities

The falling cost and increasing sophistication of consumer UAVs combined with miniaturization of BS electronics have made it technically feasible to deploy BSs on flying UAVs. Because UAV BSs can be quickly deployed at optimum locations in 3D space, they are expected to provide much better performance in terms of coverage, load balancing, spectral efficiency, and user experience compared to existing ground-based solutions.

The deployment of aerial BSs, however, faces several practical issues. In particular, placement and mobility optimization are challenging problems for aerial BSs, which have attracted significant attention from the research community. The optimization of UAV power consumption and the development of practical recharging solutions for UAVs are also important challenges to overcome for sustaining the operation of aerial BSs. In this section, we survey recent research addressing these issues.

Iv-a Placement Optimization for Aerial BSs

Aerial BSs can be placed at many different heights in the sky. However, the coverage as well as the UL and DL channels change with the altitude of the BS. Consequently, the inclusion of the height parameter makes the placement optimization of aerial BSs more challenging than conventional ground BSs. For mathematical tractability, researchers often ignore the mobility of aerial BSs when formulating the placement optimization problem. They define the objective as either to derive the optimum 3D coordinates for a predetermined number of aerial BSs that will maximize the system performance, or obtain the minimum number of required aerial BSs as well as their coordinates to maximize system performance. Different researchers employed different algorithms to solve the placement optimization problem. Table V compares aerial BS placement studies recently reported in the literature.

to |X|X|X| Reference & Objective & Algorithm

Košmerl et. al., [53]

& Minimize number of BSs and find their optimal positions & Evolutionary Computing

Lyu et. al., [54]

& Minimize number of BSs and find their optimal positions & Spiral placement algorithm, a polynomial-time heuristic of the general

geometric disk cover problem, which is NP-hard.

Merwaday et. al., [55] & Optimize coordinates of a finite set of aerial BSs & Brute force

Mozaffari et. al., [56] & Optimize coordinates of a finite set of aerial BSs and their total transmit power & Optimal transport theory (facility location)

Yaliniz et. al., [57] & Optimize 3D placement of an aerial BS & Combination of the interior point method and bisection search for quadratically-constrained mixed integer non-linear optimization problem

Rohde et. al., [58]

& Optimize coordinates of a finite set of aerial BSs & Genetic algorithm

Galkin et. al., [59]

& Optimize coordinates of a finite set of aerial BSs & K-means clustering algorithm

TABLE V: Aerial BS placement optimization algorithms

Iv-B Mobility Optimization for Aerial BSs

Placement optimization studies surveyed in Table V do not include mobility of aerial BSs in problem formulation. Mobility, however, is an intrinsic feature and capability of aerial BSs, which provide additional opportunities to dynamically improve their placements in response to user movements on the ground. To exploit the mobility features of aerial BSs, the practical hardware limitations of the UAVs must be considered. The limitations on speed and accelerations are studied through filed experiments in [22] using a consumer UAV. Based on the transportation methods of aerial BSs, there are two types of mobilities considered in the literature:

  1. UAVs are used only to transport a BS to a particular ground location where the BS autostarts to serve the users. If the BS needs to be relocated, it must shut down first before being transported to the new location. This type of aerial BSs therefore cannot serve while it is in motion, but it can resume its service as soon as it reaches a target location. As will be explained later in Section V-C, prototypes from Nokia Bell Labs [60] fall in this category.

  2. UAVs continue to carry the BSs and the BSs can continuously serve the ground users while they are flying. For example, prototypes from Eurecom [61], later described in Section V-D, fall in this category.

Considering the first type of aerial BSs, Chou et. al., [62] studied a BS placement mechanism where the ground users are not served by the BS when it is moving to a new location. The loss of service time due to BS mobility therefore becomes a critical parameter for the optimization. An aerial BS in this case should consider both the user density of the target location as well as the moving time to the new location when deciding its target location. The authors of [62] have shown that this problem can be modelled as facility location problem where the transport cost represents the loss of service time due to the movement of the BS from previous location to the new location.

The second type of aerial BS mobility opens up new opportunities to employ aerial BSs due to their ability to serve ground users while in motion. In particular, under this scenario, the cost of BS mobility becomes negligible. It is then possible to design more advanced solutions where aerial BSs can continuously cruise the service area to maximise network performance under geospatial variance of demands.

Designing cruising aerial BSs requires autonomous mobility control algorithms that can continuously adjust the movement direction or heading of the BS in a way that maximizes system performance. These algorithms must also insure that multiple aerial BSs cruising in an area can maintain a safe distance from each other to avoid collisions. Fotouhi et. al., [63] proposed distributed algorithms that take the interference signals, mobile users’ locations and the received signal strengths at UEs into consideration to find the best direction for BS movements at any time. Controlling the mobility of a single serving UAV is also discussed in [64]. Game theoretic mobility control algorithms are proposed in [65, 66] for multiple aerial base stations cruising freely over a large service area without being subject to individual geofencing. The game theoretic mobility control not only increased 5th-percentile packet throughput by 4x compared to hovering BSs, it also helped avoid collisions as the BSs were implicitly motivated to move towards different directions to maximize coverage and throughput.

Iv-C Power-Efficient Aerial BSs

A critical problem of Aerial BSs is their short lifetime due to the battery depletion problem. Power-efficient operation therefore is must to extend the battery lifetime. Power is consumed by both communications (electronics) and mobility (mechanical). Researchers therefore worked on both types of energy saving.

Reducing Communication Energy

One alternative to reduce the communication energy of UAV base stations is to minimize the transmission power. For example, minimizing the transmission power of one UAV [67] or multiple UAVs [68] when they are deployed in the optimal location to cover the target area is studied in the literature. The deployment of UAVs was optimized to minimize the total transmit power for UAVs while satisfying the users’ data-rate requirements is discussed in [69, 56]. Reducing the number of transmissions to decrease the energy consumption of a UAV is addressed in [70]. In this work, the minimum number of stopping and transmission points for a UAV to cover all downlink users are derived.

Another solution to improve communication energy efficiency is to develop optimal transmission schedule of UAVs, especially when UAVs are flying in a predetermined trajectory. A game theoretic approach for time slot allocation problem for one UAV and multiple UAVs collecting data from sensors are studied in [71] and [72], respectively. A sub-optimal transmission scheduling is evaluated in [73] for UAVs to improve the energy efficiency while collecting data from sensors and forward them to a remote base station.

Optimal scheduling for beaconing messages in order to maximize the energy efficiency of two competing UAVs is discussed in [74]

. In this work, two UAVs are moving randomly over areas including mobile users, and send periodically beaconing messages to users to announce their presence. A non-cooperative game theory is proposed for finding the best beaconing period for UAVs. Energy efficient uplink transmission between a terrestrial link and single LAP aerial destination is addressed in

[75]. Terrestrial nodes can select either to communicate directly with LAP or to use other terrestrial nodes as relay to reduce energy cost. Context aware network is assumed in this work, where nodes are aware of the necessary transmission and channel parameters (through gossip control).

Although the proposed methods successfully reduce the communication energy consumption for UAVs, one disadvantage of these alternatives is that the ratio of communication energy consumption to the total energy consumption of UAVs is generally negligible [76]. This observation motivates the efforts in reducing the mechanical energy consumed by UAVs, as detailed in the following.

Reducing Mechanical Energy

To reduce mechanical energy of UAVs, first, an energy consumption model is needed. According to [77, 78], the energy consumption of UAV can be modeled by


where is the minimum power needed to hover just over the ground (when altitude is almost zero) and is a motor speed multiplier. Both and depend on the weight and motor/propeller characteristics. is the maximum power of the motor, is the speed, and is the operating time. The term refers to the power consumption needed to lift to height with speed .

Another model for energy consumption is defined in [76] where the energy consumption of drone is related to its altitude. Following this work, energy consumption of drone can be calculated by , where is the mass of the drone, is the gravitational acceleration and is the altitude of the drone.

According to these models, one solution to control the energy consumption of UAVs is to regulate their height. However, changing the height might reduce the performance of UAVs. For example, in target coverage, there is a tradeoff between energy consumption and coverage radius. Higher altitude means higher observation radius but higher energy consumption [77, 78, 76]. Optimizing the flight radius and speed to improve energy efficiency is also addressed in [79]. One of the major advantages of these methods is that they have targeted the mechanical energy consumption of UAVs, which is considered as the main source of energy consumption for UAVs.

Given that mechanical activities consume much more power compared to electrical activities, manoeuvring of UAV BSs must be controlled in a power-efficient manner. Algorithms that consider the battery and energy consumption of UAVs as a constraint have also been studied in the literature. For instance, some works consider a limited availability of energy [80, 81, 82], and a limited flight time [83] in developing path planning algorithms.

Iv-D Recharging of Aerial BSs

Separate from reducing the energy consumption of UAVs, another attractive solution to combat the short lifetime of UAVs is to consider charging locations for them and replace exhausted UAVs with the fully charged ones. Comparing to the methods that focus on reducing the energy consumption, this solution is more costly and complex, as replacing/charging points must be designed in urban areas [84, 77, 85]. Moreover, the battery consumption of UAVs needs to be monitored regularly.

Sharma et al. [84] propose monitoring the battery level of UAVs by Macro base stations (MBSs). When the battery reaches a critical value, they are returned to MBS, and already charged ones replace them. Similarly, replacing UAVs by new ones is addressed in [77]. In this work, several UAVs exist to monitor mobile targets; however, an optimization problem is formulated to minimize the number of required UAVs. An algorithm for automating the replacement of UAVs is presented in [86], that can be used in a multi-UAVs environment. Employing this algorithm provides continuous uninterrupted service for users. Small recharging garages in BS towers, and in power-lines of urban areas are proposed by [87] and [85], respectively.

Iv-E Paradigm-Shifting Cost Model of Aerial BSs

Next generation cellular networks are expected to be 50 times more cost effective than 4G [88]. Therefore, cost saving has become a major challenge for conventional cellular operators. Costs of conventional cellular operators usually include capital expenditure (CAPEX) and operational expenditure (OPEX). CAPEX comprises acquisition, design and construction of site, purchase and implementation of equipment, etc. OPEX is made up of recurring costs such as site maintenance and rental, personnel expenses, electricity, etc. Fig. 5 depicts the CAPEX/OPEX breakdown in developed countries.

(b) OPEX
Fig. 5: CAPEX/OPEX breakdown in developed countries [89].

It can be observed from Fig. 5 that cellular operators spend more than half (52%) of their CAPEX on site acquisition and construction, which is followed by the costs of BSs, network testing, power and backhaul. However, CAPEX breakdown for drone-cells is expected to be quite different. With flying drones carrying cellular BSs, site-related costs could be significantly reduced or even completely removed. While the costs of BSs and network testing for drone-cells should increase compared to employment of terrestrial BSs due to the purchase and operation of drones. Overall, drone-mounted BSs have the potential to slash down the CAPEX of cellular operators by significantly reducing the site-related costs, which currently account for the lion’s share of the CAPEX.

In terms of OPEX, site-related costs also occupy a major proportion. Site rental makes this percentage even larger, if cellular operators do not own their sites [90]. As it can be seen from Fig. 5, land rent constitutes up to 42% of cellular operators’ OPEX in developed countries. Authors in [91] also reveal that site rental dominates the OPEX of cellular operators. Moreover, due to site shortage and stricter environmental regulations, site rental is becoming increasingly expensive [92]. Furthermore, promising 5G techniques such as ultra-dense cell deployment make the costs of site rental even higher since a large number of BSs will be required. When using flying BSs, the most expensive part of OPEX could be greatly reduced, at the cost of potentially higher expenses on electricity and backhaul. There will also be some additional costs such as annual registration fee for drones, flying insurance and their battery replacement.

Compared to traditional cells equipped with terrestrial BSs, cost model of drone-cells is paradigm-shifting. With BSs mounted on flying drones, both CAPEX and OPEX of conventional cellular operators could be significantly reduced by saving site-related costs, which contributes to the improvement of cost effectiveness for conventional cellular operators. In addition, when employing terrestrial BSs or relays to extend the cellular coverage, cell reorganization is required and could be expensive. This cost could be saved by deploying flying BSs/relays instead. Moreover, new business models may also emerge. For example, the deployment of drone BSs could follow the time-variant movement of people/cellular UEs, which could be achieved by easy and cheap shipping means such as public transportation systems (e.g. trains and buses). For another example, such paradigm-shifting cost model also enables some new cellular operators to provide opportunistic communication services. For drone fleet owners such as Google and Amazon, goods and data could be delivered simultaneously by flying drones following designated routes.

V Prototyping and Field Tests

Several drone communication prototypes have been already presented in literature or shown to a wider audience during exhibitions and commercial events. Their main scope is to extend coverage where ground wireless infrastructures are not feasible or to improve end user performance using flexible and dynamic deployment of serving base stations where required. In this section, we provide a description of the most interesting ones targeting both high altitude (Facebook Aquila and Google Loon) and low altitude (Nokia F-Cell and Eurecom Perfume) applications. We also discuss an example of ‘digital sky ecosystem’ (Huawei) designed to promote and study use cases and applications. Finally, additional examples of testbeds involving drones are presented.

V-a Facebook Aquila

One interesting example of high altitude drone BS is the Facebook Aquila project [93] [94], which aims at providing internet coverage in remote areas directly from the sky. The main component of the system architecture is an unmanned autonomous aircraft, named Aquila, which is capable of flying at an altitude of 18-20km over a defined trajectory to create a communication coverage region of about 100km. Aquila is self-powered through solar panels integrated on wings wider than a Boeing 737 and has light weight to increase the flying time. Moreover, it counts with a control system to adjust the GPS-based route and monitoring the most important flying parameters (like heading, altitude, airspeed, etc …), and implements propellers able to operate at both low and high altitudes, thus at different associated air densities. Aquila employs free space optic (FSO) links to connect ground access points, which in turn serve ground users using either Wi-Fi or LTE technology. Facebook Aquila system architecture is illustrated in Fig. 6.

Fig. 6: Illustration of Facebook Aquila system architecture.

V-B Google Loon

Similar to Facebook Aquila, the Google Loon project [95]

aims at bringing internet connectivity in remote areas. This is achieved by adopting stratospheric balloons to relay radio communication links from ground stations to users’ LTE phones out of the coverage of traditional ground cellular communication infrastructures. However, Google’s Loon has a number of features that makes it different to Facebook’s Aquila. The first one is that the communication may be directly relayed to the end user and not to ground access points. The second one is associated to the way the positioning of the balloons in the sky is controlled to generate the required coverage area at the ground. Instead of using propellers to maintain a pre-defined route, they appropriately adjust their altitudes taking advantage of the stratified wind currents in the stratosphere. In fact, each layer of stratosphere is associated to a different wind direction and speed which can be monitored through machine learning algorithms and used to keep the balloon around the ideal location. Fig.

7 shows the system architecture implemented by the project.

Fig. 7: Google Loon system architecture.

V-C Nokia F-Cell

An interesting example of drone base stations prototyping for low altitude applications is the Nokia F-Cell project [60]. The fundamental problem that F-Cell tries to solve is the high cost associated to the deployment and installation of a large number of small cells. F-Cell is an innovative solar-powered, self-configured and auto-connected drone deployed small cell served by a massive MIMO wireless backhaul. The F-Cell architecture is comprised of a closed loop, 64-antenna massive MIMO array placed in a centralized location that is used to spatially multiplex up to eight energy autonomous F-Cells, each of which has been redesigned to require minimum processing power and mount a solar panel no larger than the cell itself. The key innovations proposed by F-Cell can be summarized in the following three aspects:

(a) Nokia F-Cell design, including the carbon fibre external cradle and the internal hardware.
(b) Nokia F-Cell system architecture.
(c) Nokia F-Cell showcase performed at Sunnyvale, CA, USA.
Fig. 8: Nokia F-Cell: (a) Design (b) System Architecture, and (c) Showcase.
  1. Remove the need for a wired power supply through the design of a energy-efficient small cell that reduces to the minimum the baseband signal processing, and the power consumption of the transceivers and their associated radio frequency hardware components to consume no more than 15 Watts.

  2. Remove the need for a wired backhaul through the exploitation of massive MIMO spatial multiplexing capabilities below 6 GHz to provide high throughput wireless links in non-LOS conditions.

  3. Remove the constraint of a fixed deployment through the flexibilities introduced at point 1) and 2), together with an optimized design of the small cell, which weighs less than 6 Kg. Fig. 8(a) illustrates the design of the F-Cell prototype, including the external carbon-fibre cradle with integrated solar panels and the internal box containing both hardware transceivers processing boards. Overall, this allows the transportation and relocation of small cells anywhere and at any time.

In order to remove these constraints, F-Cell promulgates a novel system architecture, which is illustrated in Fig. 8(b). The essential feature introduced by F-Cell was the combination of both 1) a fully digital massive MIMO hub with 64 active transceivers, and 2) analog repeaters. This transforms the wired backhaul into a multi-stage wireless franthaul with sparse remote radio heads (RRHs). Fig. 8

(c) presents a flying test demonstration of the F-cell project, where the drone-transported F-Cell and the massive MIMO hub are highlighted in red. The proposed solution is completely transparent both to the baseband units (BBUs) and to the user equipments (UEs). Importantly, F-Cell supports non-LOS wireless networking in frequency division duplex (FDD) mode, where downlink (DL) and uplink (UL) transmissions are performed at different frequencies. This generally entails the need of performing complex signal processing tasks at the analog repeater during the channel estimation procedure

[96], [97], [98], [99]. However, in order to simplify the repeater and reduce both its power consumption and associated weight, F-cell implements a joint uplink (UL) and downlink (DL) channel estimation procedure only at the massive MIMO hub side [100], [101]. Additional details of the baseband hardware (HW) and software (SW) processing architecture and operations implemented at the massive MIMO hub are presented in [102].

UAV Model Objective Equipments Protocol Controller
AR.Drone 2 Communication between multiple UAVs [6] Raspberry Pi (Model B) 802.11 Software controller (NodeJs)
Autonomous helicopter Accuracy of the navigation system [103, 104] Software controller (C)
Skywalker X8 Investigating the impact of altitude, and motion on communication quality [105] 802.11g/ 802.11b Manual remote control
Senior Telemaster airplane kit Evaluating load-carry-deliver protocol by one UAV [106] 802.11a/ 802.11g/ 802.11b
Dragonflyer X6 Aerial localization [107] Lenovo W500 laptop, Nokia N900 XBee Software controller
NexSTAR Controlling the UAV path to improve link quality between two mobile nodes [108] 802.11b/ 802.11g Software controller
Fixed-wing and rotor based UAVs Validating the impact of UAV as a relay [109] WiFi
IRIS quadrotor Deriving the energy consumption model [81]
Firefly (Ascending Technologies) Enhancing video streaming [110] AscTec’s Ubuntu 12.04 IEEE 802.11 Manual controller
Fixed wing UAV, captive balloon, Comparing the analytical and simulation results of different channel models [111] ARM-7 CPU with Linux OS WLAN Manual controller
Commercial UAV Distinguishing between terrestrial and aerial UEs [112, 113] QualiPoc1 Android smart phone, LTE 800MHz
TABLE VI: Specifications of testbeds and real experiments in the literature

V-D Eurecom Perfume

In the context of a European Research Council (ERC) 5-years funding scheme, Perfume project has studied and developed the concept of "autonomous aerial cellular relay robots", where UAVs act as a relay base stations capable to enhance connectivity and throughput performance for off-the-shelf commercial terminals [61]. The key target of the Perfume project is to design of machine learning algorithms able to find and constantly update the optimal 3D position of flying wireless relays using fine-grained information of their LOS conditions together with other radio measurements [114].

V-E Huawei Digital Sky

Huawei’s Wireless X Lab activated in 2017 the Digital Sky Initiative to boost trials of specific use cases involving connected drones. With this purpose, the city of Shanghai has created an end-to-end ecosystem populated by key stakeholders such as mobile operators, cloud management firms, delivery companies, public transportation institutions, or even air quality monitoring agencies. Two authorized flying zones of 6km diameter and with a maximum height of 200m have been created, with distributed wireless charging zones at the ground. In this ecosystem, cellular networks are involved to ensure C&C between drones and ground control stations. In particular, one interesting use case is a remotely operated passenger carrying drone (a taxi drone), which is controlled through live high quality video streaming transmitted over a 4.5G cellular network directly to the operation room [4].

V-F Other relevant testbeds

There exist a number of smaller testbeds that verify practical problems like the reliability of UAV communication links, the UAV battery consumption, or the impact of weather conditions. These testbeds are summarized in Table VI, and below we expand about some of the most relevant ones:

  1. In [6], a realization of a low-cost test bed based on AR.Drone 2 and Raspberry Pi is presented. The main idea is to validate the possibility of creating a flying ad-hoc network based on 802.11 standard able to establish communication links among UAVs.

  2. An autonomous helicopter is also used in [103, 104] to investigate navigation system, comparing a pre-planned trajectory with the actual flight path by the helicopter. A maximum error of 3 meters in one square kilometre area of field experiment is reported. The navigation system is tested for a single UAV.

  3. An experimental test when a UAV acts as a relay to transmit information from an underwater vehicle to a ground base station is described in [105]. In this setup, a UAV moves in a circular way around an underwater vehicle with the speed of 18m/s. The effect of altitude, trajectory radius, waves, and weather conditions are analyzed in the communication between the UAV and the underwater vehicle.

Vi Regulation

Country Applicability Technical Requirements Operational Limitations Administrative Procedures Human Resources Ethical Constraints
Australia [27] classification: weight/purpose N/A
  • minimum distance to people: 30m

  • height limit: 120m

  • minimum distance to airport: 5.5km

  • daytime only (not after sunset)

  • by visual line of sight (VLOS) only

  • cannot operate over popular areas

insurance strongly recommended 2kg: pilot’s license required respect personal privacy
UK [115] classification: weight/purpose beyond visual line of sight (BVLOS): collision avoidance required
  • minimum distance to people: 50m

  • height limit: 122m

  • minimum distance to congested area: 150m

  • by VLOS only (up to 500m)

approvals vary for different operations pilot competency required protect data integrity and confidentiality
China [116] maximum weight: 7kg N/A
  • minimum horizontal distance to other aircraft: 10km

  • minimum vertical distance to other aircraft: 600m (altitude8400m); 1200m (altitude8400m)

  • maximum speed: 120km/h

  • daytime only

  • by VLOS only

operational safety evaluation required pilot certification required N/A
US [117] classification: weight/purpose operation during civil twilight: anti-collision lights required
  • height limit: 122m

  • minimum vertical distance to clouds: 152m

  • minimum horizontal distance to clouds: 610m

  • minimum distance to airport: 8km

  • maximum speed: 161km/h

  • by VLOS only

aircraft registration required; 25kg: operational certificate required remote pilot certification required respect privacy-related laws
South Africa [118] maximum weight: 7kg; classification: purpose N/A
  • height limit: 46m

  • minimum distance to people and property (unless permitted): 50m

  • minimum distance to airport: 10km

  • by VLOS only

  • in daylight and clear weather conditions

air services license and operational certificate required remote pilot’s license required respect privacy-by-laws
Chile [119] classification: weight emergency parachute required
  • minimum distance to people: 20m (vertical); 30m (horiontal)

  • height limit: 130m

  • maximum take-off weight: 9kg

  • by VLOS only

  • daytime only

flight authorization required remote pilot’s license required respect the privacy of others
TABLE VII: Summary of the most representative current UAV regulations

The evolution of UAV regulations should keep pace with the rapid emergence of UAVs, which significantly contributes to the integration of UAV into national and international aviation systems. In this section, socio-technical concerns of drone operations are outlined first. Then we explain the main criteria that constitute the current UAV regulation frameworks. Finally, the past and current status of UAV regulations are reviewed.

Vi-a Socio-Technical Concerns of Drones

Emerging technologies have enabled the widespread use of drones and their strong operational capabilities. As a result, there are increasing concerns regarding privacy, data protection and public safety from national and international aviation authorities. To understand the motivation of the development of UAV regulations, socio-technical concerns of drones need to be analyzed.

  • Privacy: The operation of drones can be a serious threat to the privacy of both individuals and businesses. For example, in the case that drones are employed for deliberate surveillance, they could intentionally violate individuals’ and businesses’ privacy. For missions such as aerial photography and traffic monitoring, privacy breaches can instead be unintentional. Moreover, high maneuverability and sensitive on-board instruments have made drones even more capable of privacy breaches. For example, small drones with a low noise level can easily enter a private property without being noticed.111The DJI Mavic Pro Platinum has achieved 4 dB (60%) noise reduction compared to the DJI Mavic Pro [120]. Indeed, images and videos taken by high definition (HD) camera could be streamed live. Although every country has legislation to protect the privacy of the public’s and citizens’, such as the Commonwealth Privacy Act 1988 in Australia and the US Privacy Act of 1974 [121][122], these rules might be out-of-date due to the rapid development of emerging technologies. Therefore, the operation of drones needs to be regulated to further protect the privacy.

  • Data Protection: During their operation, drones are usually equipped with sensors that collect personal data such as images, videos and location data. How these personal data will be processed, used, stored and disclosed should concern public institutions. According to data protection laws, citizens’ personal information should be protected from abuses [123]. Invisible and indiscriminate data collection capabilities of drones contribute to infringe data protection rules. “Invisible” refers to the fact that drones can secretly collect data due to their aerial capabilities and sensitive equipment on-board such as high resolution and night vision camera. Then collected data could be immediately uploaded online or transferred to a location that is distant from the data subject. Therefore, it is difficult for data subjects to be aware of the leakage of their personal information. Moreover, as a result of high mobility, drones indiscriminately collect and store a mass of data, which is against data protection principles [123]. Consequently, operation of drones should also be governed to protect personal information.

  • Public Safety: Public safety is another major concern for drone operations. Compared to traditional manned aircraft, drones are usually insufficiently maintained and more likely to encounter pilot errors. As a result, drone operations are faced with higher safety risks. According to [124], the accident rates for UAVs are significantly higher than those of manned aircraft. UAV accidents include collisions with manned aircraft or terrain:

    • UAV collisions with manned aircraft might lead to engine shutdowns or damaged surfaces, risking the loss of control. Therefore, in many countries there are constraints such as maximum allowed flight heights and minimum distances to airports for drones’ operation [125].

    • UAV collisions with terrain usually cause loss of control, which might hurt civilians on the ground. Hence some countries forbid drones to fly over certain areas such as specific urban areas with high population density [125].

Vi-B Criteria

Based on the socio-technical concerns specified above, UAV regulations are framed and developed. Current UAV regulations are mainly based on six criteria, as explained below [126].

  • Applicability: Applicability describes the scope that UAV regulations apply to. For example, drones are usually classified into groups based on weight or purpose, which might be treated differently by UAV regulations.

  • Technical requirements: Technical requirements specify the mandatory instruments or techniques for drones. For example, collision avoidance mechanism could be a typical one.

  • Operational limitations: UAV’s operation is usually restricted by many factors. Typical operational limitations include maximum flight height, minimum distance to airport and individuals, prohibited areas, etc.

  • Administrative procedures: Certain procedures and documents might be required before a UAV is allowed to operate, which include registration, operational certificate and insurance.

  • Human resource requirements: For certain categories of UAV and purposes of operation, the pilot needs to be qualified.

  • Implementation of ethical constraints: This criterion appoints the demands for data and privacy protection when operating drones.

Vi-C UAV Regulation: Past and Present

  • Past: The first UAV regulation was proposed in 1944, right after the World War II. The first internationally recognized aviation regulation, the Chicago Convention, pointed out that the operation of UAVs should be authorized to ensure the safety of manned civil aircraft [127]. Since 2000, due to the rapid development of UAV and its increasing popularity, UAV regulations have evolved both nationally and internationally. In 2002, the United Kingdom and Australia first published their UAV regulations. In 2006, the International Civil Aviation Organization (ICAO) announced that it was necessary to issue an internationally acknowledged legislation for civil operations of UAVs. Since 2012, an increasing number of countries have established their own UAV regulations.

  • Present: Table VII examines the current UAV regulation frameworks of six countries based on the criteria detailed in Sec. VI-B. Notice that UAV operations are currently prohibited in some countries such as Egypt and Cuba [128]. Besides international and national aviation authorities, large enterprise groups are also making effort to help develop safe operations for drones. Recently, waivers for regulations have been given to Apple, Microsoft and Uber for their drone-testing projects, which will help the Federal Aviation Authority (FAA) shape the future development of UAV regulations in the US [129].

Vii Security

Security is a very important issue for any digital system. For a UAV-aided wireless communication system, due to its unmanned nature and required remote wireless communication, security is an even more serious problem. For example, compared to terrestrial BSs, if a flying cellular BS is compromised by attackers, then its serving UEs are more likely to lose cellular connections since the UAV may directly crash. Moreover, cellular UEs served by terrestrial BSs might suffer from strong interference due to LOS links, if a UAV is manipulated by attackers. Therefore, it is significant to ensure the security of UAV systems when drones are used for cellular communications. In this section, cyber-physical security of UAV-assisted cellular communications is discussed.

Vii-a Cyber Security

Since 2007, an increasing number of cyber-attacks to the UAV systems have been reported due to popularity of drones [130]. When launching cyber-attacks, adversaries target the radio links of the UAV systems, which carry information such as data requested by cellular UEs, control signals and global positioning system (GPS) signals for UAVs’ navigation. For example, with interception of these information, adversaries are able to steal data transmitted and requested by drones or even directly manipulate operating drones by taking advantage of their control signals. Since both data and control signals are transmitted through the radio links, ensuring the security of these wireless communication channels has become an important aspect of the whole UAV system’s security. In this subsection, we will first present the scenarios where drones are used for cellular communications. Then, we evaluate the overall risk levels of different UAV connection types, namely, satellite, cellular and Wi-Fi links. After that, we analyze and list potential attack paths and corresponding defense strategies.

Use Cases

When drones are employed for cellular communications, they can serve as cellular UEs or flying cellular BSs/relays, as shown in Fig. 9. In case of UAV UEs, drones can be directly controlled either by terrestrial BSs or by ground control stations (GCSs) through non-cellular connections, which are mainly Wi-Fi connections [131]. In the first case, both data and C&C signals are transmitted via cellular connections. In the second case, data and C&C signals use two separate radio links. Moreover, some drones are remotely controlled by GCSs via satellite connections, such as the well-known Predator [132]. When drones serve as flying BSs/relays, similar conditions are present with respect to the previous UAV UEs case. However, in the presence of a third party GCS, flight path related data needs to be communicated by the cellular network to the GCS. Navigation information such as position, timing and velocity can be acquired from GPS satellites through satellite connections. As could be seen from above, there are three categories of radio links in these cases, which are satellite connection, cellular connection and Wi-Fi connection respectively. It is known that compared to cellular networks and GPS networks, Wi-Fi networks are more insecure due to the unreliable and vulnerable security techniques. Since GPS signals are broadcast and the signal format is specified to the public, it is easier to attack satellite connections than cellular connections where encryption keys and scrambling code are exchanged end-to-end. Therefore, risk levels of cellular connections, satellite connections and Wi-Fi connections are evaluated as low, medium and high, respectively. Table VIII summarizes the radio links and associated risk levels in the three cases above.

(a) UAV UEs.
(b) UAV BSs/relays controlled by terrestrial BSs.
(c) UAV BSs/relays controlled by third party GCSs.
Fig. 9: Use cases for cellular UAVs.
Role of Drones Radio Link (Risk Level)
Cellular UEs Terrestrial BSs to drones: cellular (Low)
Drones to GPS satellites: satellite (Medium)
Drones to GCSs: Wi-Fi (High)
Flying BSs/Relays
(Terrestrial BSs
Terrestrial BSs to drones: cellular (Low)
Drones to GPS satellites:
satellite (Medium)
Flying BSs/Relays
(Third Party GCSs
Terrestrial BSs to drones: cellular (Low)
Terrestrial BSs to third
party GCSs: cellular (low)
Drones to GPS satellites: satellite (Medium)
Third party GCSs to drones: Wi-Fi (High)
TABLE VIII: Risk analysis of radio links

Threat Identification and Countermeasure

After assessing the overall risk levels of the radio links in the use cases above, threats for these radio links are identified and listed. Corresponding countermeasures are also presented. Finally, likelihood and impact of identified threats are evaluated in Table IX.

  • Jamming: Adversaries generate interference signals in the same frequency band to disrupt the reception process, which is a common way of integrity attacks. For example, GPS jamming has become a critical threat for drones. It was reported that in 2012 a small drone crashed and led to casualties, which was suspected to be caused by GPS jamming for the legitimate receiver [133]. For jamming attacks, increasing the signal to noise ratio (SNR) could be a typical defense solution. However, this is always limited by how much power the transmitter can provide and how to lower the noise at the receiver by effective receiver algorithms.

  • Eavesdropping: Since both cellular and Wi-Fi connections employ wireless channels, adversaries might be able to obtain the transmitted information directly from the open environment. Eavesdropping breaches the confidentiality aspect of security. Encryption and physical layer security techniques could be used as a protective mechanism.

  • Hijacking: Hijacking here refers to attacks that adversaries take over a radio link. For example, radio links between drones and GCSs in our scenarios are all Wi-Fi connections. To launch hijacking attacks, adversaries could first use deauthentication management frames to disconnect the association between a drone and the corresponding GCS. Then the drone can be remotely controlled by adversaries via 802.11 protocols. There are several security techniques against deauthentication attacks. Effective detection algorithms could be applied and transmitted frames could be encrypted. For example, WPA2 (802.11i-2004) encryption mechanism with proper key length is recommended as a countermeasure [134]. Alternatively, a Wi-Fi access point (AP) could be hidden by disabling service set identifier (SSID) broadcasting and Wi-Fi UEs could be restricted to those with certain media access control (MAC) addresses.

  • Spoofing: Adversaries can pretend to be some entity using false information. A typical spoofing attack to drones is GPS spoofing. By transmitting false GPS signals with higher power than the authentic ones, drones could be taken over by adversaries. To prevent GPS spoofing attacks, defense solutions such as jamming-to-noise sense and multi-antenna defense could be employed [135][136].

Vii-B Physical Security

Besides cyber attacks, adversaries could also launch physical attacks to drones, which is another aspect of security concerns for UAV systems. To launch physical attacks, adversaries first need to obtain access to drones, which is achievable under two circumstances. First, adversaries can access a drone on the ground (damaged or ran out of battery) or capture a flying drone. Second, adversaries can control drones by successfully launching cyber attacks as introduced before. Here we summarize the attack paths and corresponding countermeasures according to attackers’ capability levels [137].

  • Low: Attackers aim to disassemble the captured drone to access its internal data, e.g. telemetry data via common interfaces such as USB. To defend such attacks, self-destruction mechanisms could be applied on drones, which will be enabled under pre-defined circumstances. However, self-destruction mechanism should only be triggered when necessary due to its strong side effects, e.g. potential threat to public safety, loss of both data and drones.

  • Medium: Attackers could access data through higher standard interfaces such as Joint Test Action Group (JTAG) and also access the embedded system. In this case, information stored on the drone needs to be encrypted. However, encryption may only delay the time taken by adversaries to obtain their desired data.

  • High: Adversaries are capable of launching advanced attacks such as side-channel attacks, fault injection attacks and software attacks to retrieve desired information from a drone. To deal with such attacks, superior cryptographic mechanisms and secure key management should be equipped by drones.

Threat Likelihood Impact
Jamming High Low
Eavesdropping High Medium
Hijacking Medium High
Spoofing Medium High
TABLE IX: Threat analysis

Viii Future Research Directions

In this section, we discuss the future research directions of UAV cellular networks.

Viii-a UAV Simulator

Real experiments with UAVs are inherently difficult due to tough regulations and need for large open space. Consequently, majority of researchers resort to simulations to evaluate the performance of their proposed systems and algorithms. These simulations often assume that UAVs can move at any direction at any time without any specific constraints of obstacles or any hardware restrictions. As a result, such simulation results may be far from realistic for many specific scenarios. Indeed, recent experiments with DJI quadcoptor drones have revealed that there are specific hardware limitations in terms of lateral acceleration, which prevents the drone from making turns at arbitrary angles [22]. It would be useful to develop publicly available simulators for UAVs, which would allow researchers to accurately simulate different types of drones according to their hardware specifications and subject to location-specific obstacles.

There exists sophisticated open license simulators for ground vehicles, such as SUMO [138], which allows researchers to precisely simulate the microscopic movements of each vehicle on the road subject to road restrictions available from open maps, such as OpenStreetMap [139]. Currently, these maps only show the ground-based structures, such as roads and traffic lights. A future direction could be to add extensions to these maps which include details of other obstacles, such as high-rise buildings, roof-top cranes, etc., which could affect UAV mobility in urban environment. Simulators such as SUMO could be extended to simulate microscopic mobility of UAVs of different makes and models. These extensions would allow UAV communications researchers to enjoy similar level of simulation support currently availed by researchers working on vehicular communications. It is worth noting that some drone vendors are offering their own simulators, such as DroneKit from 3D Robotics [140], Sphinx from Parrot [141], and DJI Assistant from DJI [142], which allow researchers to connect a propelar-less drone to the laptop and then collect drone telemetric data without actually having to go out and fly the drone. However, these simulators are still tied to the real drone, which still need to be purchased and connected to the laptop. As such, they do not provide the full benefit of a simulator, which can be used widely by anyone without having to purchase drones.

Viii-B Advanced UAV Mobility Control Based on Image Processing and Deep Learning

Most consumer drones are equipped with high-fidelity cameras. With image processing and deep learning, UAVs can be programmed to identify the optimal hovering location or the optimal flying direction which would provide the best signal propagation between the UAV and the target ground BS (when UAV is acting as an aerial UE), or between the UAV and a ground UE (when the UAV is acting as an aerial BS), while considering real-life obstacles, such as buildings, rooftop cranes etc. Furthermore, energy status estimation and potential battery charging operations should be considered in advanced UAV mobility control as well.

Viii-C UAVs Antennas

Because of the a drones’ ability to move in any direction with different speed, a new antenna design for airborne communication is required to achieve high data rate. One alternative to have high data rate transmission between UAVs and ground base stations is to have a tracking antenna installed on UAVs. The gyro, accelerometer and GPS information are utilized in order to track the ground station and tilt the antenna accordingly [143]. Moreover, limited space is another concern for installing antennas on UAVs [144], specially for small UAVs. A tilted beam circularly polarized antenna is proposed to install on the bottom of UAV [145] to save space. Simulation results showed that high performance in terms of return losses, axial ratio and radiation pattern can be achieved using such antenna.

Viii-D Aerial UE Identification

One of the main challenges in introducing aerial UEs in LTE and 5G is identifying that an aerial UE has the proper certification to connect to the cellular networks. To this end, 3GPP proposed both UE-based and network-based solutions to indicate that a UE is airborne. In the UE-based solutions, the UE can report informations such as altitude, flight mode and etc. With network-based techniques, different characteristics of the UE such as the mobility history, handover, etc. can help in UE detection. As stated before, since having a LOS is more probable for aerial UEs, they experience different radio conditions and interference than ground users. In this regards, three different machine learning techniques are employed in [113, 112] to detect the presence of aerial UEs using standard LTE radio measurements, such as RSRP (Reference Signal Received Power) and RSSI (Received Signal Strength Indicator). The aerial UEs can be detected by up to 99% accuracy through these methods. Developing more advance and intelligent algorithms, and utilizing various characteristics in future will lead in more precise aerial UE detection.

Ix Conclusion

We have surveyed complementary activities from academia, industry, and standards on the important issue of integrating UAV into cellular systems. Our survey reveals that 3GPP has recently launched new study items to explore opportunities and challenges for serving UAVs using existing 4G cellular networks. The outcomes of these preliminary studies point to some major interference issues arising from the height of the UAVs, but most of them can be addressed by deploying additional mechanisms in existing 4G systems. It is expected that 5G systems will be better equipped to deal with UAV related challenges. We have found that a number of vendors have already built and demonstrated UAV-mounted flying base stations as a sign of their readiness to embrace UAV into the cellular systems. Our survey has identified a rapidly growing interest on this topic in the academic community, which has resulted in a growing number of publications and workshops. There is an increasing activity within the regulation bodies to design and implement new regulations for UAVs to promote safety and privacy. Finally, we have identified new cyber-physical security threats, business and cost models, and future research directions for UAV-assisted cellular communication. We believe that UAV cellular communication is at a very early stage of development and we can expect to see continued interest and progress in this exciting new direction of cellular networking research in the coming years.


  • [1] Jonathan Vanian. Drone Sales Are About To Go Crazy., 2016. [Online; accessed 1-May-2018].
  • [2] 3GPP Technical Report 36.777. Technical specification group radio access network; Study on enhanced LTE support for aerial vehicles (Release 15). Dec. 2017.
  • [3] International Business Times. Nokia and EE trial mobile base stations floating on drones to revolutionise rural 4G coverage. url:, 2016.
  • [4] Huawei. Connected Aerial Vehicle Live. [Online; accessed 20-May-2018].
  • [5] L. Gupta, R. Jain, and G. Vaszkun. Survey of Important Issues in UAV Communication Networks. IEEE Communications Surveys Tutorials, 18(2):1123–1152, Secondquarter 2016.
  • [6] I. Bekmezci, I. Sen, and E. Erkalkan. Flying ad hoc networks (FANET) test bed implementation. In 2015 7th International Conference on Recent Advances in Space Technologies (RAST), pages 665–668, June 2015.
  • [7] S. Hayat, E. Yanmaz, and R. Muzaffar. Survey on Unmanned Aerial Vehicle Networks for Civil Applications: A Communications Viewpoint. IEEE Communications Surveys Tutorials, 18(4):2624–2661, Fourthquarter 2016.
  • [8] Silvia Sekander, Hina Tabassum, and Ekram Hossain. Multi-tier Drone Architecture for 5G/B5G Cellular Networks: Challenges, Trends, and Prospects. CoRR, abs/1711.08407, 2017.
  • [9] Yong Zeng, Rui Zhang, and Teng Joon Lim. Wireless communications with unmanned aerial vehicles: opportunities and challenges. IEEE Communications Magazine, 54(5):36–42, 2016.
  • [10] M. Mozaffari, W. Saad, M. Bennis, Y.-H. Nam, and M. Debbah. A Tutorial on UAVs for Wireless Networks: Applications, Challenges, and Open Problems. ArXiv e-prints, Mar 2018.
  • [11] Kogan. Kogan Nano Drone. Accessed: 2018-05-10.
  • [12] Parrot. PARROT DISCO. Accessed: 2018-05-10.
  • [13] DJI. SPREADING WINGS S9000. Accessed: 2018-05-10.
  • [14] Aeroscout. Scout B-330 UAV Helicopter. Accessed: 2018-05-10.
  • [15] WIKIPEDIA. General Atomics MQ-9 Reaper, 2018 (accessed May 11, 2018).
  • [16] Lina Tang and Guofan Shao. Drone remote sensing for forestry research and practices. Journal of Forestry Research, 26(4):791–797, 2015.
  • [17] Paul Fahlstrom and Thomas Gleason. Introduction to UAV systems. John Wiley & Sons, 2012.
  • [18] Akram Al-Hourani and Karina Gomez. Modeling cellular-to-uav path-loss for suburban environments. IEEE Wireless Communications Letters, 2017.
  • [19] Kenzo Nonami, Farid Kendoul, Satoshi Suzuki, Wei Wang, and Daisuke Nakazawa. Introduction. In Autonomous Flying Robots, pages 1–29. Springer, 2010.
  • [20] A. Al-Hourani, S. Kandeepan, and S. Lardner. Optimal LAP Altitude for Maximum Coverage. IEEE Wireless Communications Letters, 3(6):569–572, Dec 2014.
  • [21] Steven AP Quintero, Francesco Papi, Daniel J Klein, Luigi Chisci, and Joao P Hespanha. Optimal UAV coordination for target tracking using dynamic programming. In Decision and Control (CDC), 2010 49th IEEE Conference on, pages 4541–4546. IEEE, 2010.
  • [22] A. Fotouhi, M. Ding, and M. Hassan. Understanding autonomous drone maneuverability for Internet of Things applications. In 2017 WoWMoM Workshop on the Internet of Things, pages 1–6, June 2017.
  • [23] Kelley Sayler. A World of Proliferated Drones: A Technology Primer., 2015. [Online; accessed 20-May-2018].
  • [24] Skyfront. Accessed: 2018-05-10.
  • [25] Magdalena Dudek, Piotr Tomczyk, Piotr Wygonik, Mariusz Korkosz, Piotr Bogusz, and Bartłomiej Lis. Hybrid Fuel Cell – Battery System as a Main Power Unit for. Small Unmanned Aerial Vehicles (UAV). Int. J. Electrochem. Sci, 8:8442–8463, 2013.
  • [26] M Hasan Shaheed, Aly Abidali, Jibran Ahmed, Shakir Ahmed, Irmantas Burba, Pourshid Jan Fani, George Kwofie, Kazimierz Wojewoda, and Antonio Munjiza. Flying by the Sun only: The Solarcopter prototype. Aerospace Science and Technology, 45:209–214, 2015.
  • [27] CASA Advisory Cirular. AC 101-10 v1.2 - Remotely-piloted aircraft systems – Operation of excluded RPA (other than model aircraft), 2016. Available online: (accessed on 25 May 2018).
  • [28] Xingqin Lin, Vijaya Yajnanarayana, Siva D Muruganathan, Shiwei Gao, Henrik Asplund, Helka-Liina Maattanen, Sebastian Euler, Y-P Eric Wang, et al. The sky is not the limit: LTE for unmanned aerial vehicles. Available as arXiv:1707.07534, July 2017.
  • [29] B. van der Bergh, A. Chiumento, and S. Pollin. LTE in the sky: Trading off propagation benefits with interference costs for aerial nodes. IEEE Communications Magazine, 54(5):44–50, May 2016.
  • [30] G. Geraci, A. Garcia-Rodriguez, D. López-Pérez, L. Galati Giordano, and E. Björnson. Supporting UAV cellular communications through massive MIMO. In Proc. IEEE ICC, June 2018, to appear. Available as arXiv:1802.01527.
  • [31] A. Garcia-Rodriguez, G. Geraci, D. López-Pérez, L. Galati Giordano, Ming ding, and E. Björnson. The Essential Guide to Realizing 5G-Connected UAVs with Massive MIMO. Available as arXiv:1805.05654, May 2018.
  • [32] NTT DOCOMO Inc., Ericsson. New SID on enhanced support for aerial vehicles. 3GPP RP-170779 RAN75, Mar. 2017.
  • [33] 3GPP Technical Report 38.901. Study on channel model for frequencies from 0.5 to 100 GHz (Release 14). May 2017.
  • [34] Qualcomm. LTE Unmanned Aircraft Systems, Trial Report. In Qualcomm Technologies, Inc., 2017.
  • [35] D. W. Matolak and R. Sun. Air-ground channel characterization for unmanned aircraft systems: The near-urban environment. In MILCOM 2015 - 2015 IEEE Military Communications Conference, pages 1656–1660, Oct 2015.
  • [36] A. Al-Hourani and K. Gomez. Modeling Cellular-to-UAV Path-Loss for Suburban Environments. IEEE Wireless Communications Letters, PP(99):1–1, 2017.
  • [37] R. Amorim, H. Nguyen, P. Mogensen, I. Z. Kovács, J. Wigard, and T. B. Sørensen. Radio Channel Modeling for UAV Communication Over Cellular Networks. IEEE Wireless Communications Letters, 6(4):514–517, Aug 2017.
  • [38] N. Goddemeier and C. Wietfeld. Investigation of Air-to-Air Channel Characteristics and a UAV Specific Extension to the Rice Model. In 2015 IEEE Globecom Workshops (GC Wkshps), pages 1–5, Dec 2015.
  • [39] N. Ahmed, S. S. Kanhere, and S. Jha. On the importance of link characterization for aerial wireless sensor networks. IEEE Communications Magazine, 54(5):52–57, May 2016.
  • [40] H. T. Kung, C. K. Lin, T. H. Lin, S. J. Tarsa, and D. Vlah. Measuring diversity on a low-altitude UAV in a ground-to-air wireless 802.11 mesh network. In 2010 IEEE Globecom Workshops, pages 1799–1804, Dec 2010.
  • [41] A. Al-Hourani, S. Kandeepan, and A. Jamalipour. Modeling Air-to-Ground Path Loss for Low Altitude Platforms in Urban Environments. In 2014 IEEE GLOBECOM, pages 2898–2904, Dec 2014.
  • [42] Q. Feng, E. K. Tameh, A. R. Nix, and J. McGeehan. Modelling the Likelihood of Line-of-Sight for Air-to-Ground Radio Propagation in Urban Environments. In IEEE Globecom 2006, pages 1–5, Nov 2006.
  • [43] Y. Zhou, J. Li, L. Lamont, and C. A. Rabbath. Modeling of packet dropout for UAV wireless communications. In 2012 International Conference on Computing, Networking and Communications (ICNC), pages 677–682, Jan 2012.
  • [44] I. Y. Abualhaol and M. M. Matalgah. Performance analysis of multi-carrier relay-based UAV network over fading channels. In 2010 IEEE Globecom Workshops, pages 1811–1815, Dec 2010.
  • [45] L. Ting-ting and L. Shun-lan. The simulation model of ground-to-air channel and its application in SC-FDE. In 2009 ISECS International Colloquium on Computing, Communication, Control, and Management, volume 3, pages 186–189, Aug 2009.
  • [46] M. M. Azari, F. Rosas, A. Chiumento, and S. Pollin. Coexistence of terrestrial and aerial users in cellular networks. Available as arXiv:1710.03103, Oct. 2017.
  • [47] M. M. Azari, F. Rosas, and S. Pollin. Reshaping cellular networks for the sky: The major factors and feasibility. Available as arXiv:1710.11404, Oct. 2017.
  • [48] R. Amorim, H. Nguyen, P. Mogensen, I. Z. Kovács, J. Wigard, and T. B. Sørensen. Radio channel modeling for UAV communication over cellular networks. IEEE Wireless Communications Letters, 6(4):514–517, Aug. 2017.
  • [49] Ericsson. New WID on enhanced support for aerial vehicles. 3GPP RP-172826 RAN78, Dec. 2017.
  • [50] G. Geraci, A. Garcia-Rodriguez, L. Galati Giordano, D. López-Pérez, and E. Björnson. Understanding UAV cellular communications: From existing networks to massive MIMO. available as arXiv:1804.08489, Apr. 2018.
  • [51] P. Chandhar and E. G. Larsson. Massive MIMO for drone communications: Applications, case studies and future directions. Available as arXiv:1711.07668, Nov. 2017.
  • [52] P. Chandhar, D. Danev, and E. G. Larsson. Massive MIMO for communications with drone swarms. Available as arXiv:1707.01039, July 2017.
  • [53] J. Košmerl and A. Vilhar. Base stations placement optimization in wireless networks for emergency communications. In 2014 IEEE International Conference on Communications Workshops (ICC), pages 200–205, June 2014.
  • [54] J. Lyu, Y. Zeng, R. Zhang, and T. J. Lim. Placement Optimization of UAV-Mounted Mobile Base Stations. IEEE Communications Letters, 21(3):604–607, March 2017.
  • [55] A. Merwaday and I. Guvenc. UAV assisted heterogeneous networks for public safety communications. In Wireless Communications and Networking Conference Workshops (WCNCW), 2015 IEEE, pages 329–334, March 2015.
  • [56] M. Mozaffari, W. Saad, M. Bennis, and M. Debbah. Optimal transport theory for power-efficient deployment of unmanned aerial vehicles. In 2016 IEEE International Conference on Communications (ICC), pages 1–6, May 2016.
  • [57] R Irem Bor-Yaliniz, Amr El-Keyi, and Halim Yanikomeroglu. Efficient 3-D placement of an aerial base station in next generation cellular networks. In Communications (ICC), 2016 IEEE International Conference on, pages 1–5. IEEE, 2016.
  • [58] Sebastian Rohde, Markus Putzke, and Christian Wietfeld. Ad hoc self-healing of OFDMA networks using UAV-based relays. Ad Hoc Networks, 11(7):1893–1906, 2013.
  • [59] B. Galkin, J. Kibilda, and L. A. DaSilva. Deployment of UAV-mounted access points according to spatial user locations in two-tier cellular networks. In 2016 Wireless Days (WD), pages 1–6, March 2016.
  • [60] NOKIA. F-Cell technology from Nokia Bell Labs revolutionizes small cell deployment by cutting wires, costs and time. url:, Oct 2016.
  • [61] Eurecom. ERC Perfume Project.
  • [62] S. F. Chou, T. C. Chiu, Y. J. Yu, and A. C. Pang. Mobile small cell deployment for next generation cellular networks. In 2014 IEEE Global Communications Conference, pages 4852–4857, Dec 2014.
  • [63] A. Fotouhi, M. Ding, and M. Hassan. Dynamic base station repositioning to improve spectral efficiency of drone small cells. In 2017 IEEE 18th International Symposium on A World of Wireless, Mobile and Multimedia Networks (WoWMoM), pages 1–9, June 2017.
  • [64] A. Fotouhi, M. Ding, and M. Hassan. Dynamic Base Station Repositioning to Improve Performance of Drone Small Cells. In 2016 IEEE Globecom Workshops (GC Wkshps), pages 1–6, Dec 2016.
  • [65] A. Fotouhi, M. Ding, and M. Hassan. Flying drone base stations for macro hotspots. IEEE Access, 6:19530–19539, 2018.
  • [66] A. Fotouhi, M. Ding, and M. Hassan. Service on Demand: Drone Base Stations Cruising in the Cellular Network. In 2017 IEEE Globecom Workshops (GC Wkshps), pages 1–6, Dec 2017.
  • [67] M. Mozaffari, W. Saad, M. Bennis, and M. Debbah. Drone Small Cells in the Clouds: Design, Deployment and Performance Analysis. In 2015 IEEE GLOBECOM, pages 1–6, Dec 2015.
  • [68] M. Mozaffari, W. Saad, M. Bennis, and M. Debbah. Efficient Deployment of Multiple Unmanned Aerial Vehicles for Optimal Wireless Coverage. IEEE Communications Letters, 20(8):1647–1650, Aug 2016.
  • [69] H. Claussen. Autonomous self-deployment of wireless access networks. volume 14, pages 55–71, Spring 2009.
  • [70] M. Mozaffari, W. Saad, M. Bennis, and M. Debbah. Unmanned Aerial Vehicle With Underlaid Device-to-Device Communications: Performance and Tradeoffs. IEEE Transactions on Wireless Communications, 15(6):3949–3963, June 2016.
  • [71] A. E. A. A. Abdulla, Z. M. Fadlullah, H. Nishiyama, N. Kato, F. Ono, and R. Miura. An optimal data collection technique for improved utility in UAS-aided networks. In IEEE INFOCOM 2014 - IEEE Conference on Computer Communications, pages 736–744, April 2014.
  • [72] A. Abdulla, Z. M. Fadlullah, H. Nishiyama, N. Kato, F. Ono, and R. Miura. Toward Fair Maximization of Energy Efficiency in Multiple UAS-Aided Networks: A Game-Theoretic Methodology. IEEE Transactions on Wireless Communications, 14(1):305–316, Jan 2015.
  • [73] K. Li, W. Ni, X. Wang, R. P. Liu, S. S. Kanhere, and S. Jha. Energy-Efficient Cooperative Relaying for Unmanned Aerial Vehicles. IEEE Transactions on Mobile Computing, 15(6):1377–1386, June 2016.
  • [74] S. Koulali, E. Sabir, T. Taleb, and M. Azizi. A green strategic activity scheduling for UAV networks: A sub-modular game perspective. IEEE Communications Magazine, 54(5):58–64, May 2016.
  • [75] S. Kandeepan, K. Gomez, L. Reynaud, and T. Rasheed. Aerial-terrestrial communications: terrestrial cooperation and energy-efficient transmissions to aerial base stations. IEEE Transactions on Aerospace and Electronic Systems, 50(4):2715–2735, October 2014.
  • [76] Elhadi M. Shakshuki, Dimitrios Zorbas, Tahiry Razafindralambo, Di Puglia Pugliese Luigi, and Francesca Guerriero. Energy Efficient Mobile Target Tracking Using Flying Drones. Procedia Computer Science, 19:80 – 87, 2013.
  • [77] Luigi Di Puglia Pugliese, Francesca Guerriero, Dimitrios Zorbas, and Tahiry Razafindralambo. Modelling the mobile target covering problem using flying Drones. Optimization Letters, 10(5):1021–1052, 2016.
  • [78] Dimitrios Zorbas, Luigi Di Puglia Pugliese, Tahiry Razafindralambo, and Francesca Guerriero. Optimal drone placement and cost-efficient target coverage. Journal of Network and Computer Applications, 75:16 – 31, 2016.
  • [79] Y. Zeng and R. Zhang. Energy-Efficient UAV Communication With Trajectory Optimization. IEEE Transactions on Wireless Communications, 16(6):3747–3760, June 2017.
  • [80] Sang-Jo Yoo, Jae hyun Park, Su hee Kim, and Anish Shrestha. Flying path optimization in UAV-assisted IoT sensor networks. ICT Express, 2(3):140 – 144, 2016. Special Issue on ICT Convergence in the Internet of Things (IoT).
  • [81] C. D. Franco and G. Buttazzo. Energy-Aware Coverage Path Planning of UAVs. In Autonomous Robot Systems and Competitions (ICARSC), 2015 IEEE International Conference on, pages 111–117, April 2015.
  • [82] Marina Torres, David A. Pelta, José L. Verdegay, and Juan C. Torres. Coverage path planning with unmanned aerial vehicles for 3D terrain reconstruction. Expert Systems with Applications, 55:441 – 451, 2016.
  • [83] E. Yanmaz, R. Kuschnig, M. Quaritsch, C. Bettstetter, and B. Rinner. On path planning strategies for networked unmanned aerial vehicles. In Computer Communications Workshops (INFOCOM WKSHPS), 2011 IEEE Conference on, pages 212–216, April 2011.
  • [84] Vishal Sharma, Kathiravan Srinivasan, Han-Chieh Chao, Kai-Lung Hua, and Wen-Huang Cheng. Intelligent deployment of UAVs in 5G heterogeneous communication environment for improved coverage. Journal of Network and Computer Applications, 2016.
  • [85] Max Blenkin. Powerlines to charge drones:Lockheed Martin’s Melbourne research lab takes flight., 2017. Accessed: 2017-08-23.
  • [86] Milan Erdelj, Osamah Saif, Enrico Natalizio, and Isabelle Fantoni. Uavs that fly forever: Uninterrupted structural inspection through automatic uav replacement. Ad Hoc Networks, 2017.
  • [87] Y. Li and L. Cai. UAV-Assisted Dynamic Coverage in a Heterogeneous Cellular System. IEEE Network, 31(4):56–61, July 2017.
  • [88] Wonil Roh. 5G mobile communications for 2020 and beyond-vision and key enabling technologies. Key note: at IEEE WCNC, 2014.
  • [89] Market Research Report: Emerging Market Opportunites. Analysys Mason, 2010.
  • [90] Djamal-Eddine Meddour, Tinku Rasheed, and Yvon Gourhant. On the role of infrastructure sharing for mobile network operators in emerging markets. Computer Networks, 55(7):1576–1591, 2011.
  • [91] Ekkehard Lang, Simone Redana, and Bernhard Raaf. Business impact of relay deployment for coverage extension in 3GPP LTE-Advanced. In Communications Workshops, 2009. ICC Workshops 2009. IEEE International Conference on, pages 1–5. IEEE, 2009.
  • [92] Harish Vadada. Radio Network Sharing - new paradigm for LTE. url:, March 2011.
  • [93] Facebook. Building Communications Networks In The Stratosphere.
  • [94] T. Perry. Facebook Aims to Remake Telecom With Millimeter Waves and Tether-tennas., Apr. 2017.
  • [95] Google X. Balloon-powered Internet for everyone.
  • [96] E. Björnson, E. G. Larsson, and T. L. Marzetta. Massive MIMO: ten myths and one critical question. IEEE Communications Magazine, 54(2):114–123, Feb. 2016.
  • [97] T. L. Marzetta and B. M. Hochwald. Fast transfer of channel state information in wireless systems. IEEE Transactions on Signal Processing, 54(4):1268–1278, Apr. 2006.
  • [98] L. P. Withers, R. M. Taylor, and D. M. Warme. Echo-MIMO: A two-way channel training method for matched cooperative beamforming. IEEE Transactions on Signal Processing, 56(9):4419–4432, Sep. 2008.
  • [99] J. Choi, D. J. Love, and P. Bidigare. Downlink training techniques for FDD massive MIMO systems: Open-loop and closed-loop training with memory. IEEE Journal of Selected Topics in Signal Processing, 8(5):802–814, Oct. 2014.
  • [100] T. L. Marzetta, O. Blume, P. Rulikowski, S. Maier, A. Pascht, and T. Klein. Frequency division duplex (FDD) massive MIMO backhaul for repeater small cells, May 2016.
  • [101] S. Wesemann and T. L. Marzetta. Channel training for analog FDD repeaters: Optimal estimators and cramer-rao bounds. IEEE Transactions on Signal Processing, 65(23):6158–6170, Dec. 2017.
  • [102] L. Guan, L. Galati Giordano, and A. Bonfante. A flexible HW and SW co-operated baseband research platform for massive MIMO system. In 2017 IEEE ICC, pages 1–7, May 2017.
  • [103] Gurkan Tuna, Bilel Nefzi, and Gianpaolo Conte. Unmanned aerial vehicle-aided communications system for disaster recovery. Journal of Network and Computer Applications, 41:27 – 36, 2014.
  • [104] G. Tuna, T. V. Mumcu, and K. Gulez. Design strategies of unmanned aerial vehicle-aided communication for disaster recovery. In High Capacity Optical Networks and Emerging/Enabling Technologies, pages 115–119, Dec 2012.
  • [105] T. A. Johansen, A. Zolich, T. Hansen, and A. J. Sorensen. Unmanned aerial vehicle as communication relay for autonomous underwater vehicle - Field tests. In 2014 IEEE Globecom Workshops (GC Wkshps), pages 1469–1474, Dec 2014.
  • [106] C. M. Cheng, P. H. Hsiao, H. T. Kung, and D. Vlah. Maximizing Throughput of UAV-Relaying Networks with the Load-Carry-and-Deliver Paradigm. In 2007 IEEE Wireless Communications and Networking Conference, pages 4417–4424, March 2007.
  • [107] Z. Liu, Y. Chen, B. Liu, C. Cao, and X. Fu. HAWK: An Unmanned Mini-Helicopter-Based Aerial Wireless Kit for Localization. IEEE Transactions on Mobile Computing, 13(2):287–298, Feb 2014.
  • [108] C. Dixon and E. W. Frew. Optimizing Cascaded Chains of Unmanned Aircraft Acting as Communication Relays. IEEE Journal on Selected Areas in Communications, 30(5):883–898, June 2012.
  • [109] Z. M. Fadlullah, D. Takaishi, H. Nishiyama, N. Kato, and R. Miura. A dynamic trajectory control algorithm for improving the communication throughput and delay in UAV-aided networks. IEEE Network, 30(1):100–105, January 2016.
  • [110] Severin Kacianka and Hermann Hellwagner. Adaptive Video Streaming for UAV Networks. In Proceedings of the 7th ACM International Workshop on Mobile Video, MoVid ’15, pages 25–30, New York, NY, USA, 2015. ACM.
  • [111] N. Goddemeier, K. Daniel, and C. Wietfeld. Coverage evaluation of wireless networks for Unmanned Aerial Systems. In 2010 IEEE Globecom Workshops, pages 1760–1765, Dec 2010.
  • [112] J. Wigard, R. Amorim, H. C. Nguyen, I. Z. Kovacs, and P. Mogensen. Method for detection of airborne UEs based on LTE radio measurements. In 2017 IEEE 28th Annual International Symposium on Personal, Indoor, and Mobile Radio Communications (PIMRC), pages 1–6, Oct 2017.
  • [113] R. Amorim, J. Wigard, H. Nguyen, I. Z. Kovacs, and P. Mogensen. Machine-learning identification of airborne uav-ues based on lte radio measurements. In 2017 IEEE Globecom Workshops (GC Wkshps), pages 1–6, Dec 2017.
  • [114] J. Chen and D. Gesbert. Optimal positioning of flying relays for wireless networks: A LOS map approach. In 2017 IEEE International Conference on Communications (ICC), pages 1–6, May 2017.
  • [115] Civil Aviation Authority UK. Unmanned Aircraft System Operations in UK Airspace – Guidance), 2015. Available online: (accessed on 25 May 2018).
  • [116] Civil Aviation Administration of China. Civil Aviation Administration of China Measures for the Administration of Air Traffic in Civil Unmanned Aircraft Systems, 2016. Available online: (accessed on 26 May 2018).
  • [117] U.S. Department of Transportation Federal Aviation Administration. Advisory Circular Small Unmanned Aircraft Systems (sUAS), 2016. Available online: (accessed on 26 May 2018).
  • [118] South African Civil Aviation Authority. RPAS regulations (Part 101), 2015. Available online: (accessed on 26 May 2018).
  • [119] Direccion General De Aeronautica Civil Chile. Direccion General De Aeronautica Chile Dan 151—Operaciones de Aeronaves Pilotadas a Distancia (RPAS), 2015. Available online: http:// (accessed on 26 May 2018).
  • [120] DJI. MAVIC PRO PLATINUM, 2018 (accessed May 11, 2018).
  • [121] Beverly Sibthorpe, Erich Kliewer, and Len Smith. Record linkage in australian epidemiological research: health benefits, privacy safeguards and future potential. Australian and New Zealand Journal of Public Health, 19(3):250–256, 1995.
  • [122] Marc Langheinrich. Privacy by design—principles of privacy-aware ubiquitous systems. In International conference on Ubiquitous Computing, pages 273–291. Springer, 2001.
  • [123] Rachel L Finn, David Wright, L Jacques, and P De Hert. Study on privacy, data protection and ethical risks in civil remotely piloted aircraft systems operations: Final report. Retrieved February, 27:2015, 2014.
  • [124] Christopher Bolkcom and Elizabeth Bone. Unmanned aerial vehicles: Background and issues for congress, report for congress, congressional research service. In Library of Congress, 2003.
  • [125] Bill Canis. Unmanned aircraft systems (UAS): Commercial outlook for a new industry. Congressional Research Service Washington, 2015.
  • [126] Claudia Stöcker, Rohan Bennett, Francesco Nex, Markus Gerke, and Jaap Zevenbergen. Review of the current state of uav regulations. Remote sensing, 9(5):459, 2017.
  • [127] WM Sheehan. Air cabotage and the chicago convention. Harvard Law Review, 63(7):1157–1167, 1950.
  • [128] UAV System International. Banned Countries For Drones. url:, May 2017.
  • [129] The Guardian. Apple, Microsoft and Uber test drones approved but Amazon left out in cold. url:, May 2018.
  • [130] Ahmad Y Javaid, Weiqing Sun, Vijay K Devabhaktuni, and Mansoor Alam. Cyber security threat analysis and modeling of an unmanned aerial vehicle system. In Homeland Security (HST), 2012 IEEE Conference on Technologies for, pages 585–590. IEEE, 2012.
  • [131] Imad Jawhar, Nader Mohamed, Jameela Al-Jaroodi, Dharma P Agrawal, and Sheng Zhang. Communication and networking of uav-based systems: Classification and associated architectures. Journal of Network and Computer Applications, 84:93–108, 2017.
  • [132] Joseph A Ouma, Wayne L Chappelle, and Amber Salinas. Facets of occupational burnout among us air force active duty and national guard/reserve mq-1 predator and mq-9 reaper operators. Technical report, School of Aerospace Medicine Wright Patterson AFB OH, 2011.
  • [133] CG Leela Krishna and Robin R Murphy. A review on cybersecurity vulnerabilities for unmanned aerial vehicles. In Safety, Security and Rescue Robotics (SSRR), 2017 IEEE International Symposium on, pages 194–199. IEEE, 2017.
  • [134] Daojing He, Sammy Chan, and Mohsen Guizani. Communication security of unmanned aerial vehicles. IEEE Wireless Communications, 24(4):134–139, 2017.
  • [135] Daniele Borio and Ciro Gioia. Real-time jamming detection using the sum-of-squares paradigm. In Localization and GNSS (ICL-GNSS), 2015 International Conference on, pages 1–6. IEEE, 2015.
  • [136] Jaroslaw Magiera and Ryszard Katulski. Detection and mitigation of gps spoofing based on antenna array processing. Journal of applied research and technology, 13(1):45–57, 2015.
  • [137] Kamesh Namuduri, Serge Chaumette, Jae H Kim, and James PG Sterbenz. UAV Networks and Communications. Cambridge University Press, 2017.
  • [138] Michael Behrisch, Laura Bieker, Jakob Erdmann, and Daniel Krajzewicz. Sumo–simulation of urban mobility: an overview. In Proceedings of SIMUL 2011, The Third International Conference on Advances in System Simulation. ThinkMind, 2011.
  • [139] OpenStreetMap contributors. OpenStreetMap., 2004. [Online; accessed 1-May-2018].
  • [140] 3DR. DroneKit, Developer tools for drones., 2017.
  • [141] Parrot. What is Sphinx., 2017.
  • [142] DJI Developer. Testing, Profiling and Debugging ., 2017.
  • [143] S. R. Ganti and Y. Kim. Design of Low-Cost On-board Auto-tracking Antenna for Small UAS. In 2015 12th International Conference on Information Technology - New Generations, pages 273–279, April 2015.
  • [144] Randall L Musselman and Steve Eugene Watkins. Antenna Design for Small UAV Locator Applications. American Society for Engineering Education (ASEE), 2017.
  • [145] Yohandri, Asrizal, and J. T. S. Sumantyo. Design of tilted beam circularly polarized antenna for CP-SAR sensor onboard UAV. In 2016 International Symposium on Antennas and Propagation (ISAP), pages 658–659, Oct 2016.