A Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements
AA Survey on Cellular-connected UAVs: Design Challenges, Enabling5G/B5G Innovations, and Experimental Advancements
Debashisha Mishra ∗ , Enrico Natalizio Université de Lorraine, CNRS, LORIA, France
A R T I C L E I N F O
Keywords :Cellular-connected UAV5G/B5GUAV communicationsUAV integration
A B S T R A C T
As an emerging field of aerial robotics, Unmanned Aerial Vehicles (UAVs) have gained significantresearch interest within the wireless networking research community. As soon as national legislationsallow UAVs to fly autonomously, we will see swarms of UAV populating the sky of our smart cities toaccomplish different missions: parcel delivery, infrastructure monitoring, event filming, surveillance,tracking, etc. The UAV ecosystem can benefit from existing 5G/B5G cellular networks, which can beexploited in different ways to enhance UAV communications. Because of the inherent characteristicsof UAV pertaining to flexible mobility in 3D space, autonomous operation and intelligent placement,these smart devices cater to wide range of wireless applications and use cases. This work aims atpresenting an in-depth exploration of integration synergies between 5G/B5G cellular systems andUAV technology, where the UAV is integrated as a new aerial User Equipment (UE) to existing cellularnetworks. In this integration, the UAVs perform the role of flying users within cellular coverage, thusthey are termed as cellular-connected UAVs (a.k.a. UAV-UE, drone-UE, 5G-connected drone, oraerial user). The main focus of this work is to present an extensive study of integration challengesalong with key 5G/B5G technological innovations and ongoing efforts in design prototyping and fieldtrials corroborating cellular-connected UAVs. This study highlights recent progress updates withrespect to 3GPP standardization and emphasizes socio-economic concerns that must be accountedbefore successful adoption of this promising technology. Various open problems paving the path tofuture research opportunities are also discussed.
1. Introduction
Unmanned Aerial Vehicles, abbreviated as UAVs, areaircrafts without any human pilot onboard, mainly con-trolled and managed remotely or via embedded autonomouscomputer programs. UAVs are also popularly known asdrones. It is a new paradigm emerged from aerial roboticswith enormous potential for enabling new applications in di-verse areas and business opportunities [1, 2, 3]. The globalUAV market was valued at US$ . billion in and isexpected to reach US$ . billion by , at a compoundannual growth rate (CAGR) of . % during a forecast pe-riod [4].Unique features of UAVs pertaining to high mobility inthree-dimensional space, autonomous operation, flexible de-ployment tend to find appealing solutions for wide rangeof applications including civil, public safety, Industrial IoTplatforms (IIoT), security and defence sectors, cyber phys-ical systems, atmospheric and environmental observationetc [5, 6, 7]. By leveraging other emerging technologies likeArtificial Intelligence(AI), Internet of Things (IoT), Aug-mented Reality/Virtual Reality(AR/VR), UAVs have beenable to showcase substantial value proposition to a widerange of civil and industrial applications across diverse ar-eas. The UAVs are flying platforms with adaptive altitudesupport and hence, the emerging use cases for each of thementioned applications demand a secure, reliable wirelesscommunication infrastructure for command and control, aswell as an efficient information dissemination towards theground control station [8]. E-mails: [email protected], [email protected] ∗ Corresponding author.
On this advent, there are two main research directionsto be investigated. First, how to integrate a suitable wire-less communication platform into UAVs for ubiquitous con-nectivity and seamless service for the identified use cases.Second, what are the scientific and technological challengesthat arise from such integration. We aim to focus on both thedirections in this paper and highlight several distinctive char-acteristics, challenges with state-of-the-art solutions fromthe viewpoint of aerial networking.UAVs are inherently mobile in nature and hence, re-quire wireless support for communication needs [9]. Thewireless communication infrastructure can be provided overlicensed or unlicensed spectrum. Unlicensed spectrum isshared by multiple parties and are more prone to interfer-ence/contention scenarios. On the other hand, licensed spec-
Figure 1:
Integration opportunities of UAV with Cellular Net-workD. Mishra et al.:
Preprint submitted to Elsevier
Page 1 of 30 a r X i v : . [ c s . N I] M a y Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements
UAV-BSTerrestrial BS Terrestrial BSTerrestrial BSGround-UE (a) UAV-Assisted Cellular Communication
UAV-UETerrestrial BS Terrestrial BSTerrestrial BS UAV-UEUAV-UEUAV-UEGround-UE (b) Cellular-Assisted UAV Communication
Terrestrial BS Terrestrial BSTerrestrial BSGround-UEUAV UAVUAVU2U U2U (c) UAV-UAV Communication
Figure 2:
Integration Opportunities of UAV to Cellular Network trum provides reliable channel allocation for UAV communi-cations and also requires regulatory decisions. The licensedspectrum operations for UAV can be realized via severalways, such as satellite technology, separate licensed spec-trum allocated for UAV, or by existing cellular bands. Satel-lite spectrum is well suited for wide area coverage, but oftenlimited by higher costs, higher latency and lower through-put. Laying out a dedicated spectrum for UAV operations iscostly and requires substantial effort to build a system sup-porting drone operations. To this end, the UAV ecosystemcan benefit from existing cellular networks [10] for commu-nication and networking purpose.Recently, the ambitious requirements of Fifth Genera-tion and Beyond Fifth Generation (5G/B5G) wireless net-works envision to cater to a wider variety of goals in termsof higher coverage and connectivity, ultra-reliable low la-tency communication (URLLC), support for massive num-ber of devices via machine type communication (mMTC),greater bandwidth and throughput (extreme mobile broad-band or eMBB) [11, 12]. The new specifications in ThirdGeneration Partnership Project (3GPP) Rel-15 and improve-ments for 5G radio interface (termed as 5G New Radio orNR) is designed to offer the above mentioned features [13].The UAVs are envisioned to be an essential part of 5G/B5Gnetworks with potentials of supporting high data transmis-sion ( ∼
10 Gbits/s), stringent latency (1 ms round trip de-lay) and enhancements to radio access technologies (RATs).Moreover, the licensed mobile spectrum provides wide ac-cessibility beyond visual line of sight (BVLoS), secure andreliable connectivity enabling cost-effective UAV operationfor a multitude of use cases [14, 15, 16].
From the communication viewpoint, the requirements ofUAV can be classified into two broad categories [17]:•
Control and Non-Payload Communication(CNPC) -
It refers to the time critical control andsafety commands to maintain the flight operations.CNPC includes the navigation, waypoint updates,telemetry report and air traffic control (ATC) updatesto ensure secure and reliable UAV operation. CNPC
Table 1
UAV Cellular Communication Requirement [18]Type CNPC Uplink CNPC Downlink PayloadRate ∼
100 Kbps ∼
100 Kbps ∼
50 MbpsLatency - ∼
50 ms Same as Ground UE usually demands highly secure and reliable com-munication with low data rate (few hundred Kb/s)requirements. The reliability requirement for CNPCis less than −3 packet error rate (PER).• Payload Communication -
It refers to all the in-formation dissemination activities between UAV andground station pertaining to a UAV mission. Forinstance, in a surveillance operation, UAV needs totransmit real time video to the ground station/remotepilot via payload communication. Payload communi-cation demands the underlying transmission mediumto be capable of supporting high data rates (oftenhigher in full HD video transmission or wireless back-hauling).Table 1 summarizes the rate and latency requirementsfor UAV cellular communication.
The integration of UAVs to cellular network falls underthree broad paradigms [19, 20], as shown in Fig. 1:•
UAV-Assisted Cellular Communication -
In thisparadigm, UAVs are realized as flying base stations,relays or localization anchors, that can intelligentlyreposition themselves to assist the existing terrestrialwireless communication system to improve the userperceivable Quality of Experience (QoE), spectral ef-ficiency and coverage gains [21, 22]. This architec-ture is shown in Fig. 2a. Due to dynamic mobilityand repositioning, the integration of UAV brings manyadvantages to existing terrestrial communication sys-tem [9]. The base station mounted on the UAV (fly-
D. Mishra et al.:
Preprint submitted to Elsevier
Page 2 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements ing base station or relays) could be provisioned on de-mand, which is an absolute appealing solution for dis-aster management, search and rescue or emergency re-sponse. The coverage and data rate of existing cellularnetworks can be improved by optimal 3D placementand coordination of flying base stations to cater theusers need in hotspot areas. These benefits definitelycope well with diverse, dynamic and increasing datademands in 5G/B5G cellular systems.•
Cellular-Assisted UAV Communication -
This isalso known as
Cellular-connected UAVs . As shownin Fig. 2b, flying UAVs are realized as new aerial UserEquipments (UEs) coexisting with terrestrial UEs thataccess the cellular network infrastructure from the sky.This paradigm has gained significant interest in recenttimes, because of the effective solution for establish-ing reliable wireless connectivity with ground cellularstations [23].•
UAV-UAV Communication -
In this paradigm, agroup of UAVs reliably communicate directly witheach other sharing the cellular spectrum with groundusers in order to facilitate autonomous flight be-haviours, cooperation in a UAV fleets, and collisionavoidance. This architecture is shown in Fig. 2c.In [24], the authors investigated reliable and directUAV-to-UAV communications that leverage same fre-quency spectrum with uplink of cellular ground users.In this work, we prioritize the focus on the promisingfeatures of cellular-connected UAVs. In the next section, wesurvey the existing classification works, in order to highlightthe key contributions of this work. L IST OF A BBREVIATIONS
2. Related surveys and tutorials
There are growing research efforts to investigate theinterplay of UAVs with cellular networks. During the lastfew years, novel solutions have been proposed to solvescientific, technical, socio-economical and security chal-lenges. Several surveys, demonstrations and tutorials arealso presented in the literature to provide the unified viewof this research domain. These works not only helps the
D. Mishra et al.:
Preprint submitted to Elsevier
Page 3 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements research community to track ongoing research efforts, butalso consolidate the necessary knowledge for the interestedpractitioners and researchers in the community.A majority of the surveys and tutorials mostly focuson the (i) integration opportunity of UAV with 5G/B5Gcellular networks from the perspective of UAV-assistedcellular communication or (ii) highlight recent advances,future trends, challenges for UAV cellular communication,or (iii) present detailed analysis and performance studywith respect to a specific communication challenge, suchas channel modelling, physical layer techniques, securityetc. However, our work aims at focussing on the paradigmof cellular-assisted UAV communications. In order toemphasize the relevance and uniqueness of our currentsurvey work compared to existing surveys, first, we planto summarize the existing surveys and tutorials along withworks pertaining to cellular-connected UAVs in Table 2.As summarized in Table 2, majority of existing surveysare based on UAV-assisted cellular communication anddiscusses them in-depth. There are few surveys that focuseson cellular-connected UAV paradigm, but these existingworks are largely fragmented and do not provide a holisticview of this paradigm. In other words, only few selected as-pects of cellular-connected UAV like UAV-ground channelmodelling or trajectory optimization or MIMO are studiedin depth so far. These works do not present an extensivestudy including all kinds of research highlights dedicatedto cellular-connected UAVs, rather present a singulartopic in depth. Thus, a unified work providing the broadpicture of all kinds of research developments is still missing.With this survey, we aim at addressing this gap andfocus solely on cellular-connected UAVs. The researchhighlights pertaining to state-of-the-art advancements, syn-ergistic integration challenges of UAVs as aerial users in5G/B5G cellular networks, underlying network architec-tures, physical layer enhancements of 5G, field trials, sim-ulations and testbed developments are some of unique con-tributions made in this survey. The cloudification and soft-warization of network resources portrayed as the interplayof Network Functions Virtualization (NFV) and cloud com-puting technologies for the cellular-connected UAVs is alsopresented from the architectural context of enabling 5G/B5Ginnovations supporting them.
The key contributions of this work are the following. Thefinal column of Table 2, bearing the heading “This Work",also summarizes the contributions made in this work:• To present an overview of emerging applications andtaxonomy of use cases for cellular-connected UAVs;• To highlight the state-of-the-art trends of communica-tion requirements of UAVs and detailed discussion ofdesign challenges, which must be accounted for suc- cessful integration of this technology within 5G/B5Gcellular systems;• To showcase the emerging 5G technology innova-tions in network architectures such as virtualization &softwarization of network resources, slicing & phys-ical layer improvements in the interest of cellular-connected UAVs;• To present the detailed efforts for design and develop-ment of experimental testbeds, trials and prototypingcarried out by academia, industries and standardiza-tion bodies to understand the gap of theoretical analy-sis and realistic deployments;• To identify a fairly exhaustive outline of featuresfor realization of an ideal experimental prototype forcellular-connected UAV & existing works to achievethem;• To discuss about the ongoing standardization ac-tivities, regulatory frameworks, market and socio-economic issues that must be thoroughly investigatedbefore successful and widespread adoption of cellular-connected UAVs;• To present insights to future research opportunities.The high level organization of this work is summarizedin Fig. 3.
Related Surveys andTutorialsTaxonomy of UAVApplications & UsecasesIntegration Challengesof UAVs over 5GSynergies of 5G/B5Ginnovations for Cellular-connected UAVsDesign Trials andPrototypingStandardization &Socio-economicConcernsConclusions Existing surveys & tutorials, key contributions3GPP standardization, regulatory framework, market &social challenges Experimental test-beds, field trials 3D Coverage, channel model,system ops & mobility, trajectoryoptimization, security concernsExploration of various emergingcellular connected UAVapplications & diverse use casesKey 5G technology innovations innetwork architectures,virtualization & softwarization &PHY layer improvementsConcluding remarksIntroduction Overview of UAV Cellular Communication FundamentalsFuture Outlooks Future research opportunities
Figure 3:
High level organization of this workD. Mishra et al.:
Preprint submitted to Elsevier
Page 4 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements T a b l e E x i s t i n g s u r v e y s a nd t u t o r i a l s f o r U AV ce ll u l a r c o mm un i c a t i o n B r o a d C a t e go r y R e f e r e n c e s →→ C o n t r i bu t i o n s ↓↓ [ ][ ][ ][ ][ ][ ][ ][ ][ ][ ][ ][ ][ ][ ][ ][ ][ ][ ][ ] T h i s W o r k N a t u r e o f I n t e g r a t i o n U AV - A ss i s t e d C e ll u l a r C o mm ✓✓✓✓✓✓✓✓✓✓✓✓✓✓ C e ll u l a r - A ss i s t e d U AV C o mm ✓✓✓✓✓✓✓✓✓✓✓✓✓✓ A pp li c a t i o n s & U s ec a s e s A pp li c a t i o n s & U s ec a s e s ✓✓✓✓✓ D e s i g n & C h a ll e n g e s T ec hn i c a l C h a ll e n g e s ✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓ P r o p a g a t i o n C h a nn e l M o d e l s ✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓ M o b ili t y & H a nd o v e r s ✓✓ T r a j ec t o r y O p t i m i z a t i o n ✓✓✓ T ec hn o l og y & E x p e r i m e n t N e t w o r k A r c h i t ec t u r e ✓✓✓✓✓✓ G / B G I nn o v a t i o n s ✓✓✓✓✓✓✓✓✓✓✓ E x p e r i m e n t a l P r o t o t y p i n g ✓✓ I d e a l F e a t u r e s o f P r o t o t y p e ✓ H a r m o n i z a t i o n & C o m p li a n ce S t a nd a r d i z a t i o n ✓✓✓ R e g u l a t i o n s ✓✓✓ C o mm un i c a t i o n R e qu i r e m e n t ✓✓✓✓✓✓ S o c i o - ec o n o m i c C o n ce r n s S ec u r i t y A s p ec t s ✓✓✓ S o c i a l C o n ce r n s ✓✓ M a r k e t C o n ce r n s ✓ D. Mishra et al.:
Preprint submitted to Elsevier
Page 5 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements
First, we begin with the detailed taxonomy of appli-cation domains and corresponding use cases for cellular-connected UAVs, and then highlight the key integrationchallenges of UAVs being supported from cellular 5G/B5Gsystems. For seamless integration, the recent technical inno-vations of 5G/B5G mobile network architectures and phys-ical layer improvements are also presented. Then, we high-light the testbeds, field trials and measurement campaignsthat showcase some early efforts to develop working proto-types of cellular-connected UAV. Furthermore, the ongoingstandardization works, regulatory and socio-economic con-cerns are also discussed that must be accounted before suc-cessful adoption of this new technology.
3. Taxonomy of UAV Applications and Usecases
Cellular-connected UAVs find their applicability in awide range of emerging applications with varying demandsand goals. In this work, we showcase some of the attrac-tive researched domains as a starting basis for the followingdiscussion. A bird’s-eye view of this section is presented inFig 4.
As an innovative and efficient platform for gatheringdata, UAVs have become a preferred choice over traditionalgeomatics mechanisms of data acquisition. UAVs could au-tonomously fly in a defined trajectory and could preciselycapture real-time measurements of the ongoing geophysicalprocesses for abnormal hazards, such as volcanoes, land-slides, sea dynamics, earthquakes, etc. Furthermore, theUAVs are equipped with various sensors to capture atmo-spheric temperature, pollutant levels in the air, carbon emis-sions, terrestrial biomass characterization, precipitation dis-tribution in industrial zones, etc. As an efficient mechanism,the deployment of a fleet of UAVs, equipped with onboardsensors can perform the sensing for the presence of pollutantlevels or any hazardous chemicals in the target areas [36, 37].In a disaster situation, first 48 to 72 hours are very cru-cial to perform any kind of mitigation to the damage or out-age and to restore the normal state of the environment. Theresponse time is the key in saving lives in the affected re-gions. The major problems in these initial hours are: lack ofproper communication infrastructure, massive or often un-predictable losses of lives and property. Thus, the situationforces the first responder teams to implement and improvisethe search and rescue (SAR) mission to be conducted quicklyand efficiently. Latest advancements of UAVs and sensornetworks are capable to meet this need in terms of disasterprediction, assessment and fast recovery. UAVs can gatherthe information (e.g., situational awareness, early warnings,persons movement) during disaster phase and these infor-mation are helpful for first responder teams to react effi-ciently. UAVs can re-establish the communication infras-tructure ( i.e.,
UAV-assisted paradigm) destroyed at the timeof disaster.
Government constructions and public infrastructuressuch as highways and railways are greatly benefited by theseflying platforms for efficient surveillance, land surveying,tracking workers and employees, on-site construction anddemolition [25, 38, 39]. Furthermore, UAV-based deliv-ery systems are gaining wide popularity in logistics do-main to achieve faster and cost effective good delivery ser-vice [40]. Such a system handles consumer orders, managesautonomous flight and status tracking using real time con-trol. Google’s Wing project and Amazon Prime Air are thesome of the efforts to realize such a use case of UAVs. In pre-cision agriculture, UAVs are capable of observing the agri-cultural fields for health monitoring, spraying pesticides andperform hyper spectral imagery. Such activities by humansare time consuming and prone to risks. Unmanned aircraftsare well suited for such use cases enhancing productivityand cost efficacy. The cellular operators have started envi-sioning UAVs as backup wireless infrastructure (flying basestations or relays) in the absence of terrestrial communica-tion infrastructure to boost network capacity [43]. Google’sLoon project aims to provide ubiquitous Internet services &wireless connectivity to both remote and rural areas by em-ploying high altitude platform (HAP) UAVs as balloons.
UAVs are an effective means of surveillance and mon-itoring of areas stricken by a natural disaster. For instance,autonomous UAVs are sent to landslide, fire, earthquake andflooding areas to help with assessing the risks, the damagesand support first responders teams as well as providing con-nectivity to isolated people [41, 42, 44]. Similarly, low costUAVs revolutionize the conservation and management offorest and wildlife ecosystem by assisting in counting ani-mal populations, tracking illegal activities, etc. UAVs arealso an effective means of surveillance and control for thehomeland security and public safety [45, 46, 47]. In caseof anti-terroristic operations, UAVs are used to develop andprepare for situational awareness of threat, carrying out pre-emptive strikes or reconnaissance mission. UAVs assist inspeeding up the rescue and recovery (search and rescue) mis-sions in certain disastrous and crime control situations in atarget area.
Industry 4.0 is an emerging paradigm embracing nextgeneration industrial developments with the ideas of us-ing Internet of Things (IoT) to industrial automation, cy-ber physical systems, smart production and service systems.This industrial revolution is a gateway to boost economy andoperational excellence under the umbrella term of “SmartFactory”.UAVs have already begun to become a vital componentof Industrial IoT platforms [48, 49]. Practical usage of UAVsin industrial settings include monitoring terrains of manu-facturing sites or regions that are impenetrable for humansdue to hazardous exposures. The manual on-site inspec-tion carried out by humans are time-consuming and often
D. Mishra et al.:
Preprint submitted to Elsevier
Page 6 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements
Figure 4:
Taxonomy of UAV Cellular applications & use cases include very challenging terrains with inaccessible/unsafezones. Such human-driven inspections pose threats to hu-man lives. On the bright side, not only industrial UAVscan penetrate complex and inaccessible areas, but also areequipped with a multitude of sensors with cognitive com-puting to facilitate on-demand real-time bidirectional com-munication with industrial control stations. UAVs used inindustrial settings can measure many parameters for the re-gion under study via onboard sensors, such as electric andmagnetic field strength, humidity, temperature, pressure inthe atmosphere, methane or toxic pollutants. The commu-nication could occur the same way as an IoT sensor send-ing signals to the Supervisory Control and Data Acquisition(SCADA) system.
Some emerging technologies such as augmented reality(AR) and virtual reality (VR) combined with capabilities ofUAV open up novel possibilities [50]. Real life videos fromhigh altitude or high quality aerial photographs bring a greatlook and feel experience for users. Also, in the enterprisemarkets, the VR technology clients can accelerate buyer’sdecisions by presenting them best scenery and viewing ofthe real estate. AR- and VR-enabled UAVs are also used forvirtual tour of the real environments, 3D models of build-ings, graphical overlays of maps, streets, gaming, etc.
The important lessons learnt in the previous Sections canbe summarized in the following two main items:• The popularity of UAV is growing day-by-day and it isconsidered as a preferred technology to cater to a widevariety of emerging real-world use cases. UAVs canbe autonomous, intelligent, adaptive and highly mo-bile. From communication and networking perspec-tive, UAVs play an important role in cellular domain.The cellular ecosystem can benefit from UAV technol-ogy. UAVs can be efficiently integrated to existing cel-lular networks as a flying base station or a relay or anaerial UE. These different types of integration show-case several promising applications and use cases.• Owing to the implicit benefits of cellular networksin terms of ubiquitous accessibility, large coverage,scheduled and safe information exchange protocols,cellular-connected UAVs are well suited and find theirapplicability in many real-world applications such asearth and environmental observation, civil infrastruc-ture and surveillance, defence and security, industrialIoT platforms, etc. Integrating UAVs to 5G/B5G cel-lular systems proves to be a win-win situation for boththe parties.
D. Mishra et al.:
Preprint submitted to Elsevier
Page 7 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements
This work aims at presenting an extensive study ofcellular-connected UAVs, where the UAVs are integratedinto the existing cellular networks as new aerial UEs. Inorder to carry out a mission specific task, UAVs requiresupport from ground infrastructure (base station/control sta-tion), with which they exchange CNPC commands in down-link direction, and both CNPC & payload data in uplink.Cellular-connected UAVs bring several open challenges andoperational complications that need to be thoroughly inves-tigated and motivate our work to offer researchers and prac-titioners a handful guide to approach this field.
4. Integration Challenges of UAVs over 5G
The aerial communications and networking of cellular-connected UAVs pose several challenges to thoroughly in-vestigate. For instance, a reliable and low latency commu-nication for efficient control of the UAV is of utmost im-portance. Existing cellular infrastructures are primarily de-signed and developed to offer enhanced communication ser-vices for the terrestrial users. Also, the geographical terrainswith limited coverage from terrestrial Base Station (BS) maynot provide the required connectivity services to the cellular-connected UAVs, thereby demand promising solutions forsuccessful adoption of this technology.Various studies and research efforts have shown tremen-dous potential for the support and operation of low altitudeUAVs using cellular networks [51]. The benefits of cost-effective cellular spectrum in terms of low latency and highthroughput connectivity services, make it a suitable candi-date for integration of UAVs. Moreover, this technologyis scheduled, robust, secure and offers reliable services. Interms of the security aspects of data communication, exist-ing mobile networks already encompass the needful securityand authentication features in their protocol layers. A workitem to study and evaluate LTE as a potential candidate forUAV operation is carried out in 3GPP Rel-15, and the resultsare summarized in TR 36.777 [13]. In addition to existingcellular spectrum bands (600 MHz - 6 GHz), 5G ecosystemis also considering the use of spectrum in millimeter wave(mmWave) bands (24-86 GHz). As a foundation of cellularoperations, the licensed spectrum provide scheduled, reli-able and wide area connectivity that can potentially be lever-aged for UAV operations in BVLoS range.There are a lot of challenges to be tackled in order tomake the cellular-connected UAVs as an attractive solutionfor a plethora of emerging use cases. In the following sub-sections, we highlight the primary design challenges andperspectives to be considered for cellular-connected UAVs,as well as the studies and solutions already available.
The existing radio access technologies are not primar-ily suited for supporting flying radio devices as their de-ployments are mainly focused to optimally serve the groundUEs (or terrestrial UEs). The base stations (eNodeBs) are typically designed and developed to provide optimal perfor-mance to the ground users. The current eNodeBs are down-tilted to serve above purpose. Down-tilting the antennas pro-duces radiation patterns that are not useful to serve aerialUEs, which are expected to be positioned at different alti-tudes with respect to the ground surface. The inherent as-sumptions made for the ground UEs are quite different fromaerial UEs. The aerial users typically fly higher than the BSantenna height and therefore, need 3D coverage suitable forvarying UAV altitude [53]. The BS antennas of LTE net-works may provide weak channel gain by using their antennaside lobes. In the 3D space, the coverage criterion of UAVsare functions of BS antenna height, UAV altitude, antennapattern, and association rules. Hence, the network modelfor aerial users coexisting with ground users necessitates a3D coverage model [54].
In [52], the authors present 3D coverage and channelmodelling of cellular-connected UAVs in the downlink anduplink directions. The BS antenna pattern tremendously im-pacts the coverage distribution that affects the UAV oper-ation and mobility. As the down-tilt angle increases, theground BS offers smaller gains to UAV above the BS height,thereby impacting the uplink and downlink coverage proba-bilities. In Fig. 5 (a), the power gain pattern is illustrated fora synchronized uniform linear antenna array (ULA) with 10co-polarized dipole antenna elements and BS down-tilt an-gle 𝜃 𝑡𝑖𝑙𝑡 = −10 °. (b) and (c) are the 2D elevation pattern forBS with down-tilt angle for 𝜃 𝑡𝑖𝑙𝑡 = −10 ° and 𝜃 𝑡𝑖𝑙𝑡 = −20 °,respectively. One of the primary design challenges in realizingthe cellular-connected UAVs is to ensure harmonious co-existence mechanisms between ground users and aerialusers [55]. Proper UAV-ground interference managementis central to realize this coexistence. Also, the interfer-ence patterns in ground BS to UAVs communication linkexperiences remarkable difference than that of link betweenground BS to ground UE [26]. The higher altitude of UAVsthan base stations results in LoS links, which are more re-liable than the link with ground users. Additionally, theyexploit large macro diversity gains being served from sev-
Figure 5:
Power gain and elevation pattern of ground BS [52]D. Mishra et al.:
Preprint submitted to Elsevier
Page 8 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements eral BSs. On the other hand, the dominant LoS links createmore uplink/downlink interference as compared to groundusers, thereby making the interference management (ICIC)highly difficult. Other relevant effects to take into accountare fading, shadowing and path-loss. Existing ICIC mech-anisms may be well suited for current cellular designs, butfail to handle UAV interference management, which involvesmany BSs and impose limitations due to high complexity.Therefore, there is a need for efficient interference man-agement techniques for harmonious coexistence of groundusers and UAVs. There are several works in literature [56,57, 54] that investigate this problem considering downlinkand uplink interference.The communication channel mainly involves two typesof links, namely Ground-to-UAV (G2U) link and UAV-to-Ground (U2G) link. In cellular-connected UAV, the G2Ulink serves the downlink purpose of control and commandfor proper UAV operations, whereas U2G link serves the up-link purpose of payload communication. Rayleigh fading isthe commonly used small-scale fading model for terrestrialchannel model, whereas due to the presence of LoS propa-gation characteristics, Nakagami-m and Rician small-scalefading are usually preferred for U2G channels. The large-scale fading is affected because of the 3D coverage regionand varying altitude of UAV. The large-scale fading mod-els used can be based on a free-space channel model or al-titude/angle dependent channel model or probabilistic LoSmodels:• Free-space model - In free-space channel model, thereis no effect of fading and shadowing with very lim-ited obstruction. This model is typically suited for ru-ral regions where the LoS assumption holds valid be-tween high altitude UAVs and ground station. How-ever, in urban environment, the low altitude UAVsmay encounter non-LoS links, therefore need other ap-proaches to properly map with the propagation envi-ronment.• Altitude/Angle dependent model - In this case, thechannel parameters such as shadowing and path lossexponents are functions of UAV altitude or elevationangle. These models find their applicability in urbanor sub-urban regions depending upon the deployment.However, if the altitude does not change or UAVs flyhorizontally, altitude dependent models may not befound suitable. The elevation angle based models aremostly used for theoretical study purpose and existingliteratures are also limited in this regard.• Probabilistic LoS model - The models based on thisapproach are typically suited for urban environmentwhere the LoS and NLoS link between UAV andground are considered, due to buildings, obstacles orblockages. Moreover, the LoS and NLoS compo-nents are separately modelled based on their occur-rence probability in urban environment. The natureof urban environment with respect to building heights and density are key factors that statistically determinethe LoS and NLoS propagation characteristics.
The study item of 3GPP TSG on the enhanced LTE sup-port for aerial vehicles [18] highlights the channel modellingbetween ground base station and UAV flying at different al-titudes. The study includes the modelling of small scalefading, path loss, shadowing and LOS probability ( 𝑃 𝑙𝑜𝑠 )for three 3GPP deployment scenarios, namely Urban-Micro(UMi), Urban-Macro (UMa) and Rural-Macro (RMa). TheLoS probability is specified by:• 2D distance between UAV and ground station ( 𝑑 )• Altitude of UAV ( ℎ 𝑢 )The existing terrestrial communication channel modelcan be directly used for low UAV altitude (height below cer-tain threshold 𝐻 𝑙𝑜𝑤 ) to model the LoS probability. For alti-tude greater than a certain threshold 𝐻 ℎ𝑖𝑔ℎ , 3GPP suggeststo use 100% LoS probability. For height in between 𝐻 𝑙𝑜𝑤 and 𝐻 ℎ𝑖𝑔ℎ , the LoS probability is a function of 𝑑 and ℎ 𝑢 .Hence, for the three deployment scenarios, 𝑃 𝑙𝑜𝑠 is given by, 𝑃 𝑙𝑜𝑠 = ⎧⎪⎨⎪⎩ 𝑈 𝐸 _ 𝑃 𝑙𝑜𝑠 , if . meter ≤ ℎ 𝑢 ≤ 𝐻 𝑙𝑜𝑤 𝑓 ( ℎ 𝑢 , 𝑑 ) , if 𝐻 𝑙𝑜𝑤 ≤ ℎ 𝑢 ≤ 𝐻 ℎ𝑖𝑔ℎ , if ℎ 𝑢 ≥ 𝐻 ℎ𝑖𝑔ℎ and ℎ 𝑢 ≤ meter 𝑈 𝐸 _ 𝑃 𝑙𝑜𝑠 is the LoS probability for ground mobile termi-nal in conventional terrestrial communication in Table 7.4.2of [58]. 𝑓 ( ℎ 𝑢 , 𝑑 ) is given by, 𝑓 ( ℎ 𝑢 , 𝑑 ) = { , if ℎ 𝑢 ≤ 𝑙 𝑙 ℎ 𝑢 + 𝑒𝑥𝑝 ( − ℎ 𝑢 𝑝 )(1 − 𝑙 ℎ 𝑢 ) , if ℎ 𝑢 > 𝑙 The variables 𝑙 and 𝑝 are given as the logarithmic in-creasing function of UAV height ℎ 𝑢 as specified in [18]. Thevalues of 𝐻 𝑙𝑜𝑤 , 𝐻 ℎ𝑖𝑔ℎ , 𝑝 and 𝑙 are also defined with re-spect to different 3GPP deployment scenarios. Table B-2and B-3 in [18] provides detailed path-loss and shadowingstandard deviation, respectively. UAVs are inherently mobile in nature and section 3 high-lights many use cases of cellular-connected UAVs that im-plicitly demand BVLoS [59]. The mobility and handovercharacteristics of terrestrial cellular users are quite differ-ent from the 3D aerial mobility of cellular-connected UAVs.With increase in height, the radio environment changes andmobile UAVs face connectivity challenges. In this case, theperformance of the system depends on the handover rates,including failed and successful handovers and radio link fail-ures. Radio link failures occurs when the UAV is unableto maintain a successful connection with the serving cell.This could be because of the problematic RACH or expiry
D. Mishra et al.:
Preprint submitted to Elsevier
Page 9 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements
Figure 6:
UAVs being served from side lobes [62] of timers or after a certain maximum number of retransmis-sions is reached [60].In cellular-connected UAV, the protocol operations andregulatory needs of UAVs as aerial users are quite differentfrom the ground user. Hence, the network must first detectif the user device is aerial or not [61]. This detection can bedriven by the ground BS by estimating:• the elevation angle of the reference signal;• vertical location (altitude) or velocity of user device;• path loss/delay spread measurement of user devices.
The handover characteristics vary significantly betweenground UEs and aerial UE due to the nature of cell selec-tion, as shown in Fig. 6. In [62], the authors demonstratedthe impact of UAV flight path on handovers. The resultsshow that UAVs are prone to frequent handovers, and ping-pong handovers, due to varying altitude and speed. Evensmaller flight distances can have a large impact on handoverrate. Also, the handover frequency increases when flight al-titude increases. Table 3 summarizes the number of han-dovers occurring per minute for UAV, as compared to ter-restrial users. Scenario1 is equivalent to a ground user hav-ing one handover per minute. However, in scenario4, UAVs,at an altitude of 150m, experiences 5 handovers per minute.Many of the handovers are unnecessary and generate highsignalling overhead. Handover decisions are mainly madedepending upon received RSRP (Referenced Signal Refer-enced Power) values from different BS antennas. Groundusers are benefited by this approach, because the radio trans-mission power are directed to ground from the main lobesof the antenna, thereby improved radio power and every re-ceived RSRP is well separated from others. However, theaerial users are served primarily by the antenna side lobes,whose RSRP tends to be very similar to the radio power fromother surrounding BS. Hence, the UAV connects with morecells (distant cells), as there is a small difference in the RSRPvalues resulted from BS antenna side lobes.Hence, integration of cellular-connected UAVs with fu-ture 5G/B5G networks necessitates enhanced solutions forcell selection and handovers that seamlessly cover changingaltitudes of UAVs and support their 3D mobility patterns.
Table 3
Rate of handovers with varying UAV altitude [62]Scenario Height (Meters)
UAV trajectory or flight path refers to the path throughwhich UAV completes its mission for a specified use case.It involves a pair of locations that need to be covered, con-sidering communication requirements of payload and CNPClinks. The flying direction of UAV is usually optimized tomeet the application requirements, based on some cost func-tion involving BS locations, association sequence and mis-sion type [63, 64, 65]. A UAV trajectory is optimized tominimize the UAV flight time by ensuring that the UAV isalways connected to at least one BS, often with some discon-tinuity tolerance limit [66]. An optimization of flight pathwith above assumption is known as communication-awaretrajectory design.The rural and unpopulated areas with poor or no cellularconnectivity impact UAV trajectory, as the persistent con-nection controlling the UAV might be interrupted. Addi-tionally, UAVs operations in mmWave bands of 5G sufferfrom greater path loss and blockages leading to interruptedconnections during mission path. Fig. 7 demonstrates theneed for communication-aware trajectory design in cellular-connected UAVs consisting of many ground BSs and a singleUAV. Assume that the UAV has to cover a path from start po-sition S to final position F. As shown in the figure, the cov-erage from all the ground stations does not fully meet theconnection requirement and may suffer from discontinuity.Following are two main observations that complicate thismission path and must be accounted in the communication-aware trajectory design:• The flight path may not be a linear or straight pathfrom S to F, although it is distance-optimal. The UAV
S F
Figure 7:
UAV trajectory with cellular discontinuityD. Mishra et al.:
Preprint submitted to Elsevier
Page 10 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements (a) UAV trajectory with first cell layout (b) UAV trajectory with second cell layout
Figure 8:
UAV trajectory for two different cellular layouts with respect to a discontinuity threshold [66]
Table 4
Reference works on trajectory optimization for cellular-connected UAVsKey Considerations Approach Taken Goals of OptimizationDisconnectivity constraints [66] Dynamic Programming based approximatesolution with low complexity Minimize the UAV trajectory distance with-out staying out of coverage for certainthresholdConnectivity constraints [64] Graph connectivity-based approach Minimize the UAV’s mission completion timeby optimizing the trajectoryInterference-aware [63] Deep reinforcement learning algorithm basedon echo state network (ESN) cells Maximize the energy efficiency, minimize thewireless transmission latency and interfer-ence on ground network, minimize the timeneeded to reach destination must exhibit persistent connection with cellular net-works during flight path, thereby making it non-linearor curved.• The optimal path may pass beyond cellular coverageand hence, proper tolerance limits have to be appliedbefore the UAV connection is interrupted. The cases,where the discontinuity duration exceeds beyond theacceptable tolerance limit, the UAV fails to accom-plish the given mission being unable to maintain a suc-cessful connection to cellular network.
Table 4 highlights the existing literature for UAV tra-jectory optimization. The authors in [66] formulate an ap-proximate optimum trajectory finding problem for cellular-connected UAVs without exceeding a given discontinuitytolerance limit between a pair of locations. The problemis solved by a dynamic programming approach having lowcomputational complexity and is shown to achieve close tooptimal results. Fig. 8 demonstrates the UAV trajectory fortwo different cellular layouts with respect to a discontinu-ity threshold. It is clear that, the UAV respects this thresh-old limit to generate the flying coordinates for trajectory. Threshold value of zero ( i.e., continuous connection) gener-ates a trajectory that must pass through the cellular coverage,as shown by a dark black line in Fig. 8a. When the thresh-old value is 15 time units (shown by a red line in Fig. 8a),then the trajectory tries to minimize the distance coveredand nearly follows a straight path for distance optimization.Similar justifications are also valid for second cellular layoutshown in Fig. 8b.
Cellular-connected UAVs are usually equipped with amultitude of sensors that collect and disseminate data. Thisprovides numerous opportunities to expose them to vulner-abilities. These flying platforms are prone to cyber phys-ical attacks, with an intention to steal, control and misusethe UAV payload information by reprogramming it for un-desired behaviour. For instance, in business use case suchas goods delivery, the attacker can gain physical access tothe customer package as well as to the UAV device. Ex-isting information security measures are not well suited forcellular-connected UAVs, because these measures do nottake into account possible threats imposed on numerous on-board sensors and actuator measurements of UAVs [67, 68].
D. Mishra et al.:
Preprint submitted to Elsevier
Page 11 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements
An attacker can manipulate the UAV’s communication andcontrol system, thereby making it very difficult to bring itback online. Thus, it is crucial to develop new protectionmethodologies to avoid aforementioned intrusions and hack-ing procedures [69].
Inspired by the efficacy of the AI and ML-empoweredapproaches, in [69], the authors presented various securitychallenges focusing from the viewpoint of three differentcellular-connected UAV applications. They are - (i) UAV-based delivery systems (UAV-DS), (ii) UAV-based real-timemultimedia streaming (UAV-RMS) and (iii) UAV-enabledintelligent transportation systems (UAV-ITS). In order tosolve this challenge, the authors proposed an artificial neuralnetwork (ANN) based solution approach which adaptivelyoptimizes the network changes to safeguard the resource andUAV operation.• UAV-DS: These systems are vulnerable to cyber-physical attacks where the delivery of goods is com-promised. The malicious intruder takes control of theUAV with an intention to destroy, steal or delay thetransported goods. Even the UAVs can be physicallyattacked to acquire the goods being transported alongwith physical UAV assets.• UAV-RMS: UAV-enabled VR, online video transmis-sion and online tracking are some of the use cases inthis type of application. An attacker can manipulatethe identity of the UAV and transmit disrupted infor-mation to the control station using their identities. Ina large-scale deployment of UAVs, the control stationmust process the multi-media files incurring a largedelay and burdening high utilization of computationalresources.• UAV-ITS: This application ensures road safety, trafficanalysis to monitor accidents, track compromised ve-hicles, etc. Such benefits are achieved by a swarm ofcellular-connected UAVs cooperating to capture need-ful data during mission. An attacker can choose tosend an unidentified UAV to join the swarm of UAVsto steal the information or initial self-collision to dis-rupt the UAV-UAV communication. Such attacks canbring serious consequences to the entire mission.In [68], the authors have presented a brief survey ofstate-of-the-art intrusion detection system (IDS) mecha-nisms for networked UAVs. It highlights existing UAV-IDSapproaches and areas that need attention for building a se-cure UAV-IDS system.
In this Section, we have seen that, despite of the benefitsand wide popularity of cellular-connected UAVs, there areseveral challenges and operational complications that needsto be investigated to realize their true potential. The impor-tant lessons learnt from this Section are listed as follows: • The varying altitude of UAV necessitates a 3D wire-less coverage model for base stations, because the cur-rent design of terrestrial base station is highly op-timized for ground users. Typically, the UAVs flyhigher than base station creating LoS links that areprone to be interfered from other neighbouring basestations. Proper interference management becomeschallenging and critical in terms of harmonious co-existence between aerial UEs and ground UEs simul-taneously.• UAVs are highly mobile and mainly served by the sidelobes of existing base stations. This produces a pe-culiar cell association and increased handover rates,completely different than that of ground users. Themobility of UAV in 3D space necessitates enhancedcell selection and seamless handover patterns to opti-mize its operation.• The battery life of a UAV is limited. During the mis-sion, UAV must intelligently plans its trajectory frominitial to destination location considering applicationand use case. The key performance metrics, such asmaximum allowed time to complete the mission, per-sistent cellular connectivity, QoS guarantees, energyconsumption, etc. are some of the factors the UAVmust respect during its mission. Hence, trajectory op-timization is an essential aspect of UAV mission.• While carrying out sensitive and real-time criticaltasks, UAVs are prone to security threats and cyberphysical attacks. Any malicious attempt to steal, mis-use or control the UAV, can trigger undesirable situa-tions and cause loss of confidential and private assets.Such security threats require stringent protection mea-sures, guidelines and regulations by the operator.• Before 5G/B5G cellular systems can really benefitfrom the UAV technology, above mentioned technicalintegration challenges demand a thorough investiga-tion and practical solutions.
5. Synergies of 5G/B5G innovations forCellular-connected UAVs
By design, cellular-connected UAVs are expected to becontrolled and managed remotely by a Ground Control Sta-tion (GCS). Depending upon the UAV application and usecase, the UAVs carry out different missions, which requireunique networking characteristics. In general, these net-working requirements are very tightly coupled with the usecase and hardware infrastructure support. Especially, multi-UAV systems comprising many functional and coordinatedUAVs, establishing the reliable and secure communicationpath as well as the design and development of efficient re-configurable network architectures is a challenging issue.The key innovations of 5G/B5G systems are the cloudifi-cation and virtualization of network resources through Soft-ware Defined Networking (SDN), Network Function Virtu-
D. Mishra et al.:
Preprint submitted to Elsevier
Page 12 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements
Table 5
Envisioned Network architectures of Cellular-connected UAVTechnology Principle BenefitsNFV-Oriented[70], [71], [72] Decouples the hardware and softwarethat exists in traditional vendor net-work setting • Greater flexibility for NF deployment• Dynamic service provisioning• Easily deployable and well scalable• Efficient allocation to general purpose hardwareMEC-Oriented[73], [74], [75] The cloud computing capabilities areplaced close to edge of mobile net-work • Significant reduction in data exchange cost• Computational offloading to local servers• Improvement of QoE for end usersIoT-Oriented[31], [76], [45], [49] Connecting massive number of di-verse, smart devices to 5G/B5G cel-lular network • Assists efficient decision making on huge data• Extracting meaningful information for end users• Control automation without human intervention• Information sharing and communicationService-Oriented[77], [78], [79] Network services are provided over acommunication protocol that is inde-pendent of vendors or product • Improves the modularity of application• Transforms monolithic networking application into aset of microservices• Each microservice is a basic unit of functionality alization (NFV), Service Function Chaining (SFC), networkslicing, and physical layer improvements. SDN segregatesthe control functions and forwarding functions of a device. Itallows softwarization of the control functions, thereby mak-ing the network programmable. NFV transforms the tradi-tional network services into software based solutions (Vir-tual Network Functions i.e., VNFs) that can be dynamicallydeployed on a general purpose hardware platforms. SFC is achain of simple and smaller network functions that must fol-low an execution sequence to realize a complex and largenetwork function. Future 5G-centric networking applica-tions and services are driven by programmable network ar-chitectures, where softwarization and cloudification of net-work functions are the key enablers. Therefore, the above-mentioned 5G innovations are envisioned as a part of thecellular-connected UAV applications and will be detailed inthis Section. Specifically, Section 5.1 focuses on the envi-sioned network architectures for cellular-connected UAVs.The hardware and physical layer improvements are discussedin Section 5.2.
Based on the key enablers of future 5G-centric network-ing applications, the cellular-connected UAV network archi-tectures can be summarized in the following groups. Theyare (i) NFV Oriented, (ii) MEC Oriented, (iii) IoT Oriented,(iv) Service Oriented (SOA). In next subsection, we high-light the respective architectures and existing works. Table 5shows the glossary of related works for each of these networkarchitectures.
In [70], the authors present the feasibility of an agile,automated and cost-effective UAV deployment architecturecarrying out heterogeneous missions with the help of NFVtechnology. This work proposes an adaptable way to achievea reconfigurable UAV management system, which is capableof carrying out missions with varying objectives. For exam-ple, some UAVs could incorporate a VNF that provides ac-cess point connectivity services, another VNF for networklayer routing functionalities, a third VNF for flight controlsystem that can be easily upgraded as per the changing needsof the mission. The work is validated by a prototype builtupon open-source technologies. The high-level architectureof such a system is shown in Fig. 9. As shown in the figure,the communication infrastructure formed by a set of UAVs,where the mission planner used a MANO NFV framework(defined by ETSI), installed at ground station to flexibly de-ploy a set of VNFs over the set of UAVs. Overall design ofsuch a system consists of the following components:• Management and Orchestration (MANO):- It is lo-cated with the GCS and realized by Open SourceMANO (OSM) Release TWO. It contains all thenecessary functionalities of service orchestration andVNF manager as per ETSI NFV reference architec-ture [80]. OpenStack Ocata is used for VIM. BothOSM and VIM were deployed in mini-ITX computerhaving 4 Gb Ethernet ports, 8 GB RAM, Intel Core i72.3 GHz, 128 GB SSD with DPDK support.• UAV hardware and software:- It provides the infras-tructure support for execution and deployment on
D. Mishra et al.:
Preprint submitted to Elsevier
Page 13 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements
Figure 9:
NFV based achitecture for UAVs [70]
Ground Control Station for Management & Orchestration UAV UAVUAV UAV NFs
Flight Control
NFs
Flight ControlRouterData Collection
NFs
Flight ControlRouter
NFs
Flight ControlAccess PointIP Telephony Service
Figure 10:
UAVs with diverse network functions
Figure 11:
Deployment on multi-domain UAV services [72] light-weight VNFs. It is realized by Parrot AR.Drone2.0 carrying single board Raspberry Pi 3 Model B.• Mission Planner:- It is located at the GCS and definesthe nature and characteristics of different network ser-vices or network functions (NFs) to be deployed alongwith their placement policies. It also interfaces withMANO component to call for the light-weight VNFdeployment on set of UAVs.In order to carry out routing of VNFs to different targetUAVs, LXC Linux containers on Ubuntu OS are used. Eachrouting VNF requires resources of 1 vCPU, 128 MB RAMand 4 GB storage.The authors in [71] have presented a practical NFV basedapproach to support UAV multi-purpose deployment, whichcan be rapidly configured according to the need of the civil-ian mission. They have considered the UAVs to provideinfrastructure and hardware that enable agile integration ofnetwork functions at deployment time by a network operator.As shown in Fig. 10, a set of UAVs could be used for pro-viding communication infrastructure (virtual access points)in case of disaster or can be used in SAR operation in a re-mote area. The mission specific UAV behaviours are soft- warized as network functions and installed to UAV infras-tructure (hardware) at the time of deployment. Some net-work functions pertaining to mandatory features of any UAVsuch as flight control and telemetry are installed on all UAVhardware, irrespective of the mission. The implementationof the system prototype and the light-weight VNF is doneusing open-source software technologies. The orchestrationand life-cycle management of light-weight VNF is done byOSM Release FOUR. OpenStack Ocata version is used forrealizing the virtual infrastructure layer (VIM). The virtualmachine environment runs mini-ITX computer which con-sists of Intel Core i7 2.3 GHz processor, 4Gb Ethernet ports,16 GB RAM, 128 GB SSD. The UAV hardware platformconsists of DJI Phantom 3 carrying a Raspberry Pi 3 ModelB computing board and serves the platform for execution oflight-weight VNFs needed for specific mission.A software based service architecture running on a dis-tributed cloud environment is demonstrated in [72]. In thisdemonstration, an Industry 4.0 application controlling theindoor drones is considered for study. The application isimplemented using SFC orchestrated by a multi-domain or-chestrator known as ESCAPE. The orchestrator is able tosetup and configure VNFs onto the physical UAV boards
D. Mishra et al.:
Preprint submitted to Elsevier
Page 14 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements
UAV 2 UAV 1 UAV 3MECServerBaseStation (a) MEC based Architecture [73] (b) UAV Computational Offloading [74]
Core Network CloudStorage & DataProcessing Management &Orchestrator User3User2User14G/5G WiMax Wi-Fi3G (c) UAV-based IoT Platform
Figure 12:
Network Architectures of Cellular-connected UAV according to mission’s policies and requirements. The pro-posed implementation is shown in Fig. 11. The deploymentoccurs when the service requests are triggered to ESCAPEas per requirement. OpenStack is used for running the cloudenvironment and few laptop hosts are used as edge execu-tion machines by Docker platform. High level commandssuch as take-off, land, fly are used for controlling the UAVbehaviour from factory controller.
In general, UAVs possess physical constraints in termsof computational capability, storage and battery capacity.MEC has been identified as one of the promising techniquesto deal with the limitations of low computational capabilityand restricted battery capacity of flying UAV. Some exam-ples of resource-intensive tasks are trajectory optimization,object recognition, AI processing in crowd-sensing. Dueto the limited onboard resources of the UAVs, computationof above resource intensive tasks are not very efficient.Hence, in such case, edge-cloud based network architec-tures provide substantial improvements for operations ofcellular-connected UAVs.In [73], the authors presented a UAV-enabled MEC ar-chitecture applicable for cellular-connected UAVs. Fig. 12aillustrates this architecture, where the UAV has some com-putational task to be executed. This task can be offloadedto the MEC server located with the ground station and, afterthe computation, obtained results can be sent back to UAVfor their exploitation. Depending upon the volume of the of-fload, there can be two modes of operation: (i) partial mode,and (ii) binary mode. In partial offload mode, the whole taskis split into two parts. One part is executed locally and theother part is executed by the MEC server (e.g., face recog-nition use case). In binary offload mode, each task is exe-cuted as one unit, irrespective of whether it is done locallyor at the MEC server (e.g., channel state information (CSI)estimation). Both of these offload modes have advantagesand drawbacks. The selection of the suitable mode dependson the nature of computational task being performed, UAVstructure and characteristics. Considering the use case of trajectory optimization andcomputational offloading in cellular-connected UAV, thework in [74] presents a novel MEC setup, where the UAVneeds to offload some of its processing task to the groundstation. The UAV flies from an initial location to a desti-nation location and offload the task to selected ground basestations during the trajectory. The goal of the MEC setupis to minimize the total time for UAV mission consideringthe maximum speed and ground station capacity constraints.This setup is shown in Fig. 12b.In reference work [75], the authors proposed a 5G net-work slicing concept extend to video monitoring with UAVshaving MEC facilities. The surveillance area is divided intomultiple zones and a set of UAVs are assigned the task tomonitor a specific zone. The MEC enabled UAVs could of-fload the captured data and video streams with acceptablequality and performance.
In [31], the authors envision a heterogeneous UAV net-work architecture, where UAVs are used to deliver value-added IoT services from the sky. The UAVs are consideredas key enabler of IoT framework that are deployed by follow-ing a specific vision. Each UAV is equipped with variousIoT sensors or camera to gather data. The deployment spansacross a large area, where UAVs are grouped to form UAVclusters (because of close geographical proximity or missiontype or altitude). A fixed UAV is designated as cluster head(CH), and is mainly responsible for disseminating collecteddata to the other UAVs or orchestrator via core network. Thecore network performs the intelligent decisions and employsalgorithms for efficient processing the data gathered fromUAV sensors. The high level architecture schematic of thisproposal is shown in Fig. 12c.In [76], the authors presented the network architecture ofa UAV-enabled IoT framework developed for disaster miti-gation. In this case, the UAV not only acts as a flying basestation in emergency situation, but also behaves as a cellular-connected UAV for information dissemination in scenariossuch as wildfire or environmental losses. The frameworkconsists of three main components, (i) ground-IoT network,
D. Mishra et al.:
Preprint submitted to Elsevier
Page 15 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements (ii) connectivity of UAV and ground-IoT network and (iii)data analytics.The authors in [45] demonstrated a UAV-based IoTframework for crowd surveillance application which collectsthe data and performs facial recognition to track and identifysuspicious activities in a crowd. The fleet of UAVs are man-aged by a centralized orchestrator component.
The work in [77] demonstrates the design and de-velopment of a UAS Service Abstraction Layer (USAL)for UAV which implements different types of missionswith minimal re-configuration time. USAL contains aset of predefined useful services that can be configuredquickly according to the requirements of civil mission.The architecture is service oriented, and the service ab-straction layer provides the re-usability of the system.The mission functionalities are split into smaller partsand are implemented as independent services. USALreplies on a middleware that manages the services andtheir communication needs. USAL may contain a largenumber of services, however, all of them need not bepresent. Depending upon the mission, suitable servicescan be loaded and activated to meet the objective of mission.In [78], the authors presented Dronemap Planner, aservice-oriented cloud based UAV management system,which performs overall management of UAVs over Internetand control their communication and mission. It virtualizesthe access mechanism of UAVs via REST API or SOAP. Ituses two communication protocols: (i) MAVLink and (ii)ROSLink. The objective of designing such a system is toprovide seamless control to monitor UAVs, offload computeintensive tasks to cloud platform, and dynamically schedulethe mission on demand. The cloud computing model cre-ates an elastic model that scales well with the numbers ofUAVs as well as with the offered services. Fig 13 shows theschematic of system architecture developed in this study.
Figure 13:
Service Oriented System Architecture of UAV [78]
Table 6
Candidate waveforms for 5GScheme Short DescriptionGeneralized Fre-quency DivisionMultiplexing(GFDM) It is a block-based modulation approachwhere the available bandwidth is eitherdivided into several narrow bandwidthsubcarrier or few subcarriers with highbandwidth for each.Universal FilterBank Multi-carrier(UFMC) Multicarrier signal format to handle lossof orthogonality at receiver end. It usessub-band short duration filters.Filter Bank Multi-carrier (FBMC) It uses a preamble burst based approachto ensure flexible resource allocation.BiorthogonalFrequency Divi-sion Multiplexing(BFDM) It uses a relaxed form of orthogonalitywhere transmitter and receiver are bi-orthogonal. In other words, the trans-mitted and received pulses have to bepairwise orthogonal. BFDM is more ro-bust than OFDM.
The performance of cellular-connected UAVs in 5G net-works significantly depends on the underlying physical layersignal processing. In this section, we highlight the candidatephysical layer techniques that influence the UAV communi-cation. The key techniques are massive MIMO (MultipleInput Multiple Output) antenna, mmWave communication(3-300 GHz), beamforming and beam division multiple ac-cess (BDMA), as well as some new modulation schemes.In 4G LTE, Orthogonal frequency division multiplexing(OFDM) and Orthogonal frequency division multiple ac-cess (OFDMA) are predominantly used for multiplexing andmultiple access method. 5G/B5G networks is consideringnew waveforms to support efficient air interface [81]. Thesenew waveforms are superior than OFDM and no longer re-quire strict orthogonality and synchronization. Table 6 pro-vides a brief categorization of different waveforms for 5Gfrom implementation perspective.
5G new radio (NR) is a new radio interface and radio ac-cess network which is designed and developed for advancedcellular connectivity. It utilizes novel modulation schemesand access technologies that help the underlying system tocater to high data rate services and low latency requirements.The first version of 5G NR started in 3GPP Rel-15. 5G NRsupports the frequency ranges in sub 6 GHz or in mmWaverange (24.25 to 52.6 GHz). It has greater coverage and en-hanced efficiency because of beamformed controls, MIMOand access mechanisms. 5G NR is expected to cater to threebroad categories of services i.e., (i) extreme mobile broad-band (eMBB), (ii) ultra-reliable low latency communication(URLLC) and (iii) massive machine type communication(mMTC). Specifically, the expectations for each of the men-tioned scenarios are:
D. Mishra et al.:
Preprint submitted to Elsevier
Page 16 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements • For eMBB use case scenario, the data rate is promisedas 100 Mbps and three time more spectral efficiencythan 4G systems. It will be able to support a devicethat moved with a maximum speed of 500 km/h.• For URLLC use case scenario, the goal is to achieve1 ms latency with reliability 99.999%. It means, thereliability of the wireless link will not be met, if morethan one data unit out of data units does not getdelivered within 1 ms.• For mMTC use case scenario, the density of de-vices that 5G will be able to handle will reach nearly1000000 per square kilometer.URLLC ensures strict latency and reliability require-ments for the application. 5G NR focuses on framing,packetization, channel coding and diversity enhancementsfor achieving URLLC. One of the most vital scenarios ofURLLC is the remote piloting of cellular-connected UAVsin BVLoS range. Package delivery, remote surveillance andborder patrolling are some of the use cases that demand UAVoperations in BVLoS range. Due to the changing altitude,velocity and distance between remote UAV and ground sta-tion, the URLLC requirements may vary. URLLC is the keyuse case scenario to enable BVLoS UAV operations, whichassist in safe UAV piloting to avoid crashes, obstacles etc.Cellular-connected UAVs can benefit from 5G NR design asit offers dominant uplink data transmissions from UAV toground BS, especially for many demanding use cases per-taining to streaming, surveillance, imaging etc. The down-link data transmission requirement is much smaller in con-trast to uplink. Moreover, the sub 6 GHz and millimetrewave spectrum could potentially be used for the downlinkand uplink respectively, considering the asymmetric trafficrequirements. Massive MIMO is a promising technology that consistsof a large number of controllable antenna arrays. It issupported by 3GPP in Rel-15 for 5G NR. 5G will exploitfull benefits of MIMO by leveraging the uncorrelated anddistributed spatial location of cellular-connected UAVs, aswell as ground users. Massive MIMO enhances the signalstrength, where multiple data streams can include uniquephase and weights to the waveforms to be constructivelygenerated at the UAV receiver [82, 83]. It minimizes theinterference to other cellular-connected UAV receivers.The work [82] presents an evaluation of a massiveMIMO system for cellular-connected UAVs. It demonstratesthat, massive MIMO assists in harmonious coexistence ofcellular-connected UAVs with ground users, supports largeuplink data rates and results in consistent CNPC link be-haviour. The test uses 20 MHz bandwidth in sub-6GHz li-censed spectrum operating in TDD mode. Massive MIMO-enabled systems are useful to restrict the impact of interfer-ence to the existing terrestrial users. Such system requiresfrequent and accurate CSI updates.
Millimeter-wave (mmWave) spectrum has been exten-sively investigated in UAV cellular communication thatoffers high bandwidth services using frequency spectrumabove 28 GHz. The channel between cellular-connectedUAVs and ground BS is typical LoS dominant and mmWavehaving high bandwidth are favourable for communications.However, the mmWave signals are affected by any kind ofblockage, which poses several implementation challenges.Therefore, efficient beamforming and tracking are neededfor cellular-connected UAV operation.The work [84] presents a simulated study to showcasethe feasibility of using 28 GHz 5G link for public safetyuse case. The results claim that, it is feasible to achieve 1Gbps throughput with sub ms latency using mmWave linkswhen the grounds base station is situated close to the missionarea. In [85], the authors conducted an analysis on the air-to-ground channel propagation for two different mmWavebands at 28 GHz and 60 GHz using ray tracing simulations.During experiment, the UAV speed was kept at 15 m/s andlimited to a flight distance of 2 km. A total of four scenariosare validated such as urban, sub-urban, rural and over-sea. Itis observed that, received signal strength (RSS) follows thetwo ray propagation model as per UAV flight path at higheraltitudes. This two-ray propagation model is impacted in ur-ban scenario due to high rise scattering obstacles.
Beamforming is a technique by which a beam (signalelement directed to the users) is transmitted from the groundbase station and directed to a specific user to minimizeinterference to other neighbouring users and maximizes theuseful signal for the given user. In 5G NR, the antennas cancreate and exploit beam patterns for the specific cellular-connected UAV. This is of great importance, because ofaerial mobility of UAVs and high LoS channel conditionsfrom ground BSs. Beam division multiple access (BDMA)is capable to handle large number of users and to enhancethe communication system capacity. In this case, a separatebeam is allocated to each user. This access technology isdependent on the user positioning, location and speed ofuser movement.Beam forming requires the base station to have morethan one transceiver RF chain and the user (both aerial andground) to have single RF transceiver. In order to support agreater number of users, the beam should be split. The keychallenge is to find a way to group users that are served bya single beam without causing interference to other users atthe same time. Strategies like angle of departure (AoD) orangle of arrival (AoA) help to measure the steering anglefrom BS to mitigate interference to some extent.In [86], the authors presented a study on using steer-able directional transmitters on UAVs to evaluate the co-existence of cellular-connected UAVs with ground user.
D. Mishra et al.:
Preprint submitted to Elsevier
Page 17 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements
Figure 14:
Coexistence performance of aerial UE and groundUE, 700 MHz in rural setting [86]
This work jointly optimized the flight path and antenna steer-ing angle to improvise uplink throughput while minimizinginterference to other neighbour base stations. The proposalis validated by a testbed setup involving
MIMO wherewide beam transmitters are employed half-power beams at ° on azimuth and elevation planes and dBi forward gain.Fig. 14 shows the performance variation with respect to thethroughput in presence of both ground users and aerial users.The results show that such techniques are of utmost impor-tance when the ground and aerial users coexist at such scale. Non-orthogonal multiple access (NOMA) is a promisingcandidate technology for 5G wireless communication, as itleads to higher spectrum utilization than orthogonal multipleaccess methods (OMA). NOMA has been widely exploredin UAV-assisted wireless communications, where UAV isdeployed as a flying BS to serve the ground users [87].Few studies [88, 89] also investigated the applicability ofNOMA in cellular-connected UAV network.OFDMA and single-carrier (SC)-FDMA are conven-tional orthogonal multiple access methods (OMA) adoptedas a natural choice of 4G LTE/LTE-Advanced wireless sys-tems. The basic principles of OFDMA is to transmit thedifferent user signals over different frequency resources,not to produce mutual interference among users. Cellular-connected UAVs coexisting with ground users benefit fromsuch orthogonal multiple access methods, because the UAVswithin a given coverage can avoid any interference to groundusers by transmitting in those resource blocks that are notassigned to any ground users. Thus, the resource blockscan be allocated within the coverage area in a non overlap-ping manner. However, increased user density and the fre-quency reuse result in poor spectrum performance from suchOMA methods, due to resource block scarcity. On this ad-vent, NOMA methods allows the cellular-connected UAVsto reuse the resource blocks. In other words, NOMA is ca-pable to serve many users at the same time/frequency re-sources.NOMA employs two techniques for multiple access:• Power domain: Multiple access is based on differentpower levels.• Code domain: Multiple access is based on differentcodes. NOMA with interference cancellation (IC) is an appeal-ing solution to the cellular-connected UAVs because theUAVs can reuse the resource blocks that are allocated toground users. Moreover, at high altitude, UAVs experiencestronger LoS channel condition than ground users, so that BScan use IC to decode strong signal from UAVs, then subtractit to decode ground user signal [65].
The important lessons learnt from this section are listedas follows:• Classical cellular network infrastructures are not wellscalable for the diverse and growing use cases ofcellular-connected UAVs. The improvements of5G/B5G cellular networks present many candidate in-novative technologies and PHY layer improvementsthat complements to efficient UAV operation in 5Gspectrum.• Based on the principles of softwarization and cloudifi-cation of networking resources, the network architec-tures involving cellular-connected UAV solve severalpractical limitations with respect to performance andscalability issue.• The notable 5G advancements and trends in deployingNFV, MEC, SOA, IoT driven network architectureshelps UAV technology to establish a reliable and safecommunication link between ground-UAV or UAV-UAV during mission.• 5G-and-beyond hardware (NR) and software upgradesby cellular network operators along with the technicaladvancements by UAV manufacturers suitably catersto application specific latency, rate and reliability de-mands arising from the use cases, thereby improv-ing overall performance of applications using cellular-connected UAVs.• The physical layer enhancements further supplementto the effectiveness of applications encompassingcellular-connected UAVs.
6. Design Trials and Prototyping
Experimental assessment and prototyping are time con-suming and relatively complex, because they must take intoconsideration the deep technical aspects of any realistic de-ployment. There are several ongoing efforts from industryand academia that focus on experimental frameworks forcellular-connected UAVs. These efforts provide more prac-tical insights about the underlying behaviour and complexi-ties involved in integration of UAVs into cellular networks.Field trials and measurement campaigns are a cost-effectiveand powerful step towards the prototyping, as they help in-vestigating the solutions to potential research problems. Inthe following subsections, we shed some light on these ef-forts and classify them into two broad groups such as (i) ex-perimental testbeds and (ii) field trials.
D. Mishra et al.:
Preprint submitted to Elsevier
Page 18 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements
There is hardly any complete real-world testbed thatfully characterizes the challenges and benefits of cellular-connected UAVs. The literature in this regard is scarce.However, several ongoing efforts are being actively pursuedby researchers from both industry and academia to advancethe working prototypes. It is worth mentioning that, re-alization of working prototypes of cellular-connected UAVmainly differ with respect to (i) the main objective for whichthey are built, and (ii) the features being implemented, whichare also dependent on the main objective. For example, oneprototype may completely focus its prototype developmenton investigating 5G/B5G network support to efficient UAVoperations. Another prototype may prioritize its develop-ment on achieving a fail-safe, reliable communication withdesired QoS guarantees. Furthermore, each prototype mayutilize different hardware and software flight stacks to real-ize the goal. The chosen hardware and software platformsmay be open-source or proprietary in nature. Hence, the ex-isting efforts tend to be very specific to the goal being pur-sued, thereby providing unique characteristics or behavioursto the prototype being developed. There are no formal de-velopment guidelines available so far in order to harmonizeavailable features for these prototyping efforts.An ideal view of cellular-connected UAV prototype isstill missing. This ideal prototype can be thought of pos-sessing a non-trivial list of mandatory features and shouldbe adaptable to varying needs of the mission. Our currentwork attempts to foresight such an ideal prototype and enu-merates the list of encompassing features. Table 8 illustratesa feature-oriented comparison of existing testbed works inliterature with the desirable set of features from an ideal pro-totype point of view. Note that, this list of features is not ex-haustive, rather provides a use-case driven analogy to con-solidate the basic set of mandatory features. New featuresmay arise in future with evolution of emerging use cases forcellular-connected UAVs.In this subsection, we aim at investigating the existing ef-forts to design and develop working prototypes for realizingsome UAV operations over LTE/4G/5G/B5G cellular net-work infrastructure along with their implemented features.They are presented as follows.An open-source 4G connected and controlled self-flyingUAV is demonstrated in [90], defining a new, light-weight,
Table 7
List of avionics components used in [90]Component ModelFlight Controller Omnibus F4 ProGPS BN-220Radio Rx TBS NanoCamera & Video Tx TX05Computer Raspberry Pi Zero W4G Modem Verizon USB730L4G Antenna TS9 secure and open-source class of cellular-connected UAV.This work utilizes open-source hardware and software stackto design and develop fully autonomous and fail-safe flightbehaviour. This work provides a comprehensive and de-tailed discussion on the possible hardware and software op-tions for flight controllers, radio receivers, sensors, micro-controllers and 4G cellular modems. Fig. 15 summarizes thehardware and software components used in the prototype de-velopment. The detailed hardware avionics schematics andequipment models are highlighted in Fig. 16 and Table 7,respectively. The performance of the prototype is tested forendurance, terrain alignment, autonomous flying behaviour,wind speed and real-time video quality. The important ac-complishments of this work are summarized as follows.• The entire prototype setup is done by open-sourcehardware and software components with Commercialoff-the-shelf (COTS) components.• The UAV shows longest demonstrated flight time i.e.,over one hour.• This work provides clear, concise and step-to-stepguidelines for entire prototype design and devel-opment along with the programming of individualpieces. This also includes an online manual (wiki) andsupplementary information.
Figure 15:
Prototype design and configurations in [90]
Figure 16:
Schematic of the avionics components in [90]D. Mishra et al.:
Preprint submitted to Elsevier
Page 19 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements
Table 8
A feature-oriented comparison of prototypes of cellular-connected UAVs from viewpoint of idealistic baselineReferences →→ Features ↓↓ Short Description [90] [91] [92] [93]CellularNetwork Cellular network generation type to which the prototype isconnected and tested 4GLTE 4GLTE GSM/GPRS 3G/4GLTEOpen-source Constituent hardware and software components of the pro-totype being developed (cid:51) (cid:51) (cid:51) (cid:51)
Autonomous Whether the UAV can fly autonomously without human in-tervention (self-flying nature) (cid:51) (cid:55) (cid:55) (cid:55)
Fail-safe Ability to be resistant against lost link and returning tohome location after UAV control is interrupted (cid:51) (cid:55) (cid:55) (cid:55)
EncryptedCommunication Use of encryption mechanism to secure the message ex-changes from potential attackers (cid:51) (cid:55) (cid:55) (cid:55)
BVLoSCapable Being able to command and control the UAV, even not inthe direct view of the remote pilot (cid:51) (cid:51) (cid:55) (cid:55)
QoS-Aware UAV successfully fulfils the application demands with re-spect to quality metric such as packet loss, latency, rateand jitter (cid:55) (cid:51) (cid:55) (cid:55)
InternetConnectivity Being able to control and steer UAV from persistent Internetconnection (cid:51) (cid:55) (cid:51) (cid:51)
GroundControl UAV being remotely controlled by ground control station forcommand and control, or payload communication (cid:51) (cid:51) (cid:51) (cid:51)
Light-weight Light-weight of UAV to enhance the prototype performance (cid:51) (cid:55) (cid:55) (cid:55)
TerrainFollowing UAV maintains a fixed altitude and follows the terrain thatis useful in unknown terrains like mountains (cid:51) (cid:55) (cid:55) (cid:55)
FlightLongevity( ∼ (cid:51) (cid:55) (cid:55) (cid:55) Endurance Robustness and integrity of UAV in extreme environment (cid:51) (cid:55) (cid:55) (cid:55)
EnergyEfficient Consumption of very less power to maintain persistent flightoperation to accomplish the mission (cid:51) (cid:55) (cid:55) (cid:55)
NetworkVirtualization Ability of the prototype to be hardware platform indepen-dent and softwarization of UAV network functions (cid:55) (cid:55) (cid:55) (cid:55)
Adaptable Being responsive to current situation and the ability to re-configure the UAV as per changing requirement and missionin minimum time (cid:55) (cid:55) (cid:55) (cid:55)
AI/ML-Powered Ability to leverage efficient AI or ML based approaches toself-learn and apply the learnt knowledge to improvise themission performance over time (cid:55) (cid:55) (cid:55) (cid:55)
SwarmCooperation Ability to properly coordinate information with othercellular-connected UAVs in a multi-UAV deployment sce-nario (cid:55) (cid:55) (cid:55) (cid:55) • The prototype shows self-healing internet architectureand utilizes the fail-safe protocols for the lost links incommunication.• The UAV cellular to ground control is secure by en- cryption and can pass through several firewalls.• As compared to other UAV industry verticals, thesignificant advantages are light weight (UAV weighsnearly 300 grams) and longer flight time (> 1 hour).
D. Mishra et al.:
Preprint submitted to Elsevier
Page 20 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements
A working prototype of LTE controlled drone wasdemonstrated in [91] proving the control of UAV via LTEconnection and then tested as a 3D measurement platform.The goal of this prototype development was to investigateand evaluate LTE as a potential candidate of communica-tion infrastructure for controlling a UAV. The experimentalgoals are to provide answer to below mentioned questions.• whether existing LTE network infrastructure is an ef-ficient means of controlling UAVs?• whether the LTE connection is good enough in termsof providing low latency, jitter and bit error rate?• whether the bit rate is sufficient to perform the use caseof live video streaming in BVLoS range?The prototype is tested with respect to above mentionedgoals and found that LTE is an efficient technology for UAVoperations in BVLoS range satisfying the required the bitrate, latency and jitter. However, this prototype has severalshortcomings and may not be considered as a full-fledgedcellular-connected UAV testbed. Many features are eithermissing or not considered to keep the prototype simple inthis development, thereby leaving enough scope for furtherenhancements. Some of the important features worth high-lighting which are lacking in the prototype are listed below.• The design did not consider cellular network cover-age holes and discontinuity problem which may leadto failure of UAV operation. Flight fail safe mecha-nism is lacking.• The UAV mission specific investigation with respectto trajectory, interference from neighbouring base sta-tions, handover criteria are missing from the design.• The QoS delivered by the UAV application must takeinto account diverse real-world use cases in presenceof obstacles and variation of signal strength. Suchstudy is missing.• It did not consider the factors and performance penal-ties when UAV coexist with other ground UEs.The work presented in [92] proposes an arduino-basedlow-cost, flexible control subsystem for controlling UAVsand ubiquitous UAV mission management by GSM/GPRScellular networks. The ground control station transmits con-trol signals to UAV present in LoS or beyond LoS over GSMor GPRS cellular network, through which, it is shown that ispossible to connect to Internet, send/receive text messagesor voice calls utilizing a GSM antenna and a SIM card. Theexperimental setup includes the following components: (i)UAV is an IRIS+ quadcoptor by 3DRobotics, (ii) Pixhawkautopilot, (iii) Arduino Mega ADK Rev. 3 microcontrollerboard, (iv) GSM/GPRS module by Arduino GSM shieldwith Quectel M10 modem, (v) Mission Planner, an opensource software for ground control station software. Thefield tests are conducted by sending basic control commands
Figure 17:
High level shematics of the prototype setup in [92] from smartphone or laptop to UAV and they are success-fully executed by the UAV. The subsystem initialization timeis high, but occurs only once when the subsystem is pow-ered ON. Fig. 17 shows the high level system schematics ofthe working prototype. Following are the key observationsdrawn from above experiment.• Communication via GPRS using a Mission Plannersoftware has faster response time.• The Internet connectivity of GRPS is very fragilewhich make the GSM text message mode to be anefficient way for command and control message ex-change.A flexible open-source long-range communicationsolution for UAV telemetry based on cellular data transferservice is presented in [93] which is implemented onRaspberry Pi 3 model B (also known as rpi3) and GentooLinux control. The UAV is equipped with a Huawei 3372hdongle to get the cellular data services.
In this subsection, various efforts on field trials and mea-surement campaigns of cellular-connected UAVs are dis-cussed. Authors in [94] conducted a field measurement ina commercial LTE network for cellular-connected UAV op-eration. An LTE smartphone mounted on a consumer gradeDJI Phantom 4Pro radio controlled quadcopter is used togather the UAV flight results. The smartphone has TEMSPocket 16.3 installed for wireless measurement and analysis.The field trial results include distribution and measurementof signal quality metric such as Reference Signal ReceivedPower (RSRP), Reference Signal Received Quality (RSRQ),Signal to Interference and Noise Ratio (SINR) in the serv-ing cell and neighbouring cells with respect to UAV move-ment. The results show the feasibility of UAV operations incommercial LTE network and also highlight the implemen-tation challenges for dynamic radio environment. The simu-lations are also conducted to supplement the field trial results
D. Mishra et al.:
Preprint submitted to Elsevier
Page 21 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements in terms of network performance involving a higher num-ber of cellular-connected UAVs. Following observations aredrawn from the experiments.• The aerial propagation conditions are close to freespace propagation and hence, the aerial UEs experi-ence stronger RSRP than the ground UEs.• The RSRQ and SINR at higher altitude is lower thanthe corresponding ground UE because of the strongdownlink interference from neighbour non-servingcells to the aerial UE.• The uplink throughput for aerial UE is observed to bebetter than ground UE due to free space propagationcondition. Note that, this uplink performance also de-pends upon many other factors like scheduling mech-anism and network load.• In the downlink command and control traffic, higheraltitude results in lower spectral efficiency due toincreased interference and higher physical resourceblock (PRB) utilization.• Considering the mobility of aerial UEs at higher alti-tudes and LoS propagation conditions, a greater num-ber of radio link failures (RLF) occur due to poorSINR and large interference. Also, they may connectto far-away cells instead of the closest cell.In [95], the authors demonstrated an experimental plat-form where UAVs are connected to a commercial 5G NRbase station for radio link measurements. The 5G BS is de-veloped by Magenta Telekom in Austria and operates be-tween 3.7 and 3.8 GHz frequency, using 100 MHz band. Ithas
64 × 64 massive MIMO setup with beam forming ca-pabilities. An Asctec Pelican quadcopter is flown near thisBS and the test measurements are performed by a CellularDrone Measurement Tool (CDMT). This UAV carries a non-standalone Wistron NeWeb mobile test platform based onQualcomm Snapdragon X50 5G modem. It supports sub-6GHz 5G NR using
MIMO and 256-QAM. The goalof the study is to investigate the communication behaviourand performance characterization of flying UAV when con-nected to a commercially operated 5G base station. Thecommunication aspects for 5G-connected UAV measured inthis test are 5G connectivity, RSRP, SNR, throughput andnumber of handovers. The UAV flight includes both verti-cal lift-off and horizontal trajectory. Following observationsare drawn from above testbed driven study of 5G connectedUAV.• The UAV connectivity to 5G cannot be always guar-anteed and fall back to 4G network. This situation iseven worse at higher altitudes with more handoverstowards 4G network.• The UAV is able to receive enough data rate (severalhundred Mbps) from 5G NR based deployment, whichis adequate for many applications and use cases. • The handovers to 4G network could be reduced by de-ploying a larger number of 5G NR base stations, anddownlink rate would be improved. However, the ex-periment did not yield much benefit in the uplink ascompared to 4G. The authors assume that uplink rateanalysis needs further investigation.Qualcomm also tested the UAV operation in commer-cial LTE networks in September 2016 and produced a trialreport in May 2017 on LTE unmanned aircraft system [96].The focus of this test was to understand the operation of lowaltitude UAV platforms being supported by terrestrial cel-lular networks. The overall test encompasses both field tri-als and simulations. The field trials aim to capture datasetsby performing hundreds of flights and then complementedby extensive system level simulations to understand theperformance of UAV operation. The flights and measure-ments were performed by custom designed 390QC quadro-tor drone. Note that, these results are collected in a subur-ban/residential zone which was having good cellular cover-age, hence, cannot be generalized for other zones like urbanor rural areas. Moreover, the performance results are ap-proximate in nature rather than accurate. The key resultsobtained from the trail report are summarized as follows.• The aerial UEs experience higher received signalstrength than ground UEs despite of the down-tiltedBS antennas. This is because of the better free spacepropagation condition at higher altitude.• The SINR in the downlink for aerial UEs is lower ascompared to ground UEs due to the interference expe-rience from neighbour cell.• The UE transmit power is more for ground UEs thanaerial UEs in the uplink, because good free spacepropagation condition at higher altitude enhances theinterference energy from neighbour cell. The field re-sults depicted that aerial UEs experience nearly threetimes more interference than ground UEs in 700 MHzband.• Handover performance in terms of lower handoverfrequency and success rate of handovers is superiorfor aerial UE than ground UE due to signal stability athigh altitude.• The optimization in the power control scheme are ap-plied by simulation and was shown to eliminate theexcess uplink interference.The work in [97] presents the field trial done at a smallairport in vicinity of Odense, Denmark and results were col-lected in an LTE network operating in 800 MHz and the UAValtitude is maintained between 20 to 100 meters. The cel-lular network data was collected by a Samsung Galaxy S5smartphone which was placed inside the flying UAV cav-ity. It was equipped with Qualipoc software for reportingthe radio measurements. The UE was programmed to use a
D. Mishra et al.:
Preprint submitted to Elsevier
Page 22 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements
MIMO capability. The down-link and uplink bitrates are kept as 150 Mbps and 50 Mbps,respectively. The UAV hovers at a height of 50 meters. Toanalyse the performance, both ground UE and aerial UE gen-erate uplink traffic at full buffer capacity for one minute ineach experiment run. It is observed that the ground UE suf-fers a throughput degradation up to 21.75 Mbps because ofthe inter-cell uplink interference by aerial users. The aver-age reduction in throughput is nearly 52% i.e., equivalent to11 Mbps.The work in [98] presents the experimental evaluation ofcellular-connected UAVs communication performance con-nected to an LTE-Advanced network running 3GPP Release13. An Asctec Pelican quadrocoptor carrying a smartphone(Sony Xperia XZ2 H8216) flies in the coverage of an LTE-A network within the premises of University of Klagenfurtcampus. The experiment is performed in open-field andobstacle-free areas ensuring LoS link with at least one BS.The UE supports LTE carrier aggregation and a
MIMOantenna setup is used. The base station has a transmit powerof 20 Watt and 256 QAM and 64 QAM in downlink and up-link, respectively. The UE was able to report various LTEparameters such as RSRP, RSRQ, SINR, serving PCI, TCPuplink and downlink throughput, EARFCN etc. The UAVfollowed a straight path spanning 300 meters with a speedof 3 meters/second. The broad goal of this experimentalstudy was to understand the impact of varying UAV altitudeon achievable throughput and performance measurement ofhandovers by aerial user without any specific change in thenetwork. The keys findings are as follows:• The achievable throughput of UAV is sufficientenough to cater to many applications and use cases. Atan altitude of 150 metres, the UAV’s average through-put is 20 Mbps and 40 Mbps in the downlink and up-link, respectively. • The number of handovers increase with increasingheight of UAV. The reason for high handover fre-quency is the high RSRP and high interference valuesfrom neighbour base stations.In [99], the authors used machine learning algorithms inorder to identify the cellular-connected UAVs in the networkbased on LTE radio measurements. The measurement wasconducted in a rural location of Northern Denmark wherethe airborne aerial UAV users are realized by mounting aQualiPoc android smartphone on commercial UAV attachedto an 800 MHz LTE carrier. The height is maintained at 4different levels and UAV is flown in 4 rectangular routes.This work claims the use of supervised learning algorithmsfor efficient detection of aerial users in the network solelybased on LTE radio measurements with small number oftraining samples.Authors in [100] focuses on aerial communication fieldtrial, where a radio scanner is attached to construction liftand radio signal was measured with heights up to 40 metresin urban scenario. The measurement was carried out in threeLTE carriers such as 800, 1800 and 2600 MHz in northernDenmark. The experimental study aims at providing propa-gation models of UAVs connected to cellular networks. Thekey findings from the trials are as follows:• Increase in the received power from neighbour sourceseven in height of 40 meters that contributes to heavyinterference for the aerial user.• The observed path loss approximated to free spacepath loss after a UAV height of 25 meters.The authors in [101] have demonstrated the feasibil-ity of UAV operation via commercial cellular network forhigh data connectivity in low altitude and BVLoS operationsthroughout different times of the day.Table 9 presents a comparative analysis of different ex-isting works in literature with respect to field trials and mea-surement campaigns. Existing field trials vary greatly in sev-eral aspects, such as type of environment, deployment sce-nario, modelling platform, goal of experiment, performancemetric, etc.
The important lessons learnt from this section are listedas follows:• The solutions proposed to the technical synergisticchallenges of 5G/B5G systems with UAV technologymust be validated and tested for correctness. Hence,the key design considerations of cellular-connectedUAV necessitate sound measurement campaigns, fieldtrials, simulations and working prototypes to study thebehaviour with real-world scenarios. The experimen-tal testbeds and field measurements have significantpractical relevance, because these are very conducivefor realistic evaluation of the system under study.
D. Mishra et al.:
Preprint submitted to Elsevier
Page 23 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements
Table 9
Comparison of existing works on field trials and measurement campaignsReference Work Cellular Network Trial Environment Performance Metric[94] 4G LTE Suburban RSRP, RSRQ, SINR, downlink latency, resource utilization[95] 5G New Radio Rural RSRP, SNR, Throughput, 5G connectivity[96] 4G LTE Mixed Suburban Cellular connectivity for low altitude UAVs[97] 4G LTE Rural RSRP, RSRQ, SINR, Effect of altitude on UAV[86] 4G LTE Rural, Suburban, and Urban Throughput degradation, Interference, Uplink signal power[98] 4G LTE-A Unknown RSRP, RSRQ, SINR, PCI, UL/DL throughput, EARFCN[99] 4G LTE Rural Cellular-connected UAV identification[100] 4G LTE Urban Channel propagation models[101] 4G LTE Rural RSSI, RSRP, RSRQ, uplink/downlink throughput • Most of the existing literature relies on software sim-ulations to evaluate the technical aspects of proposedsolutions. Few works have shown to conduct fieldmeasurement campaigns to observe and study the be-haviour of cellular-connected UAV such as handovers,cell association, signal strength reduction, radio linkstatus, interference mitigation etc. Very few works fo-cus on design and development of real-world workingprototypes of cellular-connected UAV, because suchefforts are time-taking and highly complex. Prototyp-ing works are still in its infancy stage and hence, callfor more contributions in this regard.
7. Standardization & Socio-economicConcerns
Cellular-connected UAVs can pose serious risks in termsof socio-economic operational capabilities. Therefore, ut-most care must be taken by the policy makers and legislationin order to integrate UAVs into national and internationalaviation systems. To this end, in this section, we outline theperspectives of standardization, regulatory activities, marketand social challenges, which the UAV service providers andcellular operators must take into consideration before suc-cessful roll-out of use cases pertaining to cellular-connectedUAV applications.
Third Generation Partnership Project (3GPP) is astandardization body that governs the specifications forthe technical platforms used by the cellular networks.The global partnership 3GPP develops standards to whichalmost all commercial cellular network providers andoperators adhere to. In order to cater to the present andfuture needs of UAV communication, 3GPP aims to layouta unified platform for design and development of wirelessinnovations by gaining wider consensus from variouscontributors from industry and academia. The evolutionsin the standards are published in the name of “3GPPrelease". From the perspective of UAV operations overcellular networks, we are interested in 3GPP Release-15, Release-16 and Release-17. In Release-15, the study mainlyconcerns with the radio level aspects for supporting UAVs.In Release-16 and 17, the study is in the perspective ofSystem and Application layer aspects.
D. Mishra et al.:
Preprint submitted to Elsevier
Page 24 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements
UAV TrafficManagment(UTM) RAN CoreRAN CoreRAN Core UAV + UAVControllerUAV + UAVController UAS 1UAS 2
Figure 18:
High level architecture of 3GPP Release 16 workon remote identification of UAS
This release plan started in September, 2016 andthe approval for stage-3 development was conducted onDecember, 2019. This release work is mainly on Systemand Application layer aspects. A study item on “RemoteIdentification of Unmanned Aerial Systems” is pursued onthis release and it led to the approved report 22.829. Thisstudy aims at the identification of UAV over the commandand control data via a 3GPP network exchanged betweenUAS and centralized UAV Traffic Management (UTM)component. A UAS comprises of UAV and UAV controller.Fig. 18 depicts above model. 3GPP standards must makeprovisions for UAS to send the application data trafficto UTM along with various radio network information,identification and tracking details for UAS. After this study,a WI was agreed by 3GPP to advance the work on servicerequirement for identification of UAV.
In this release, 3GPP proposes a number of study items.The idea is to come up with diverse scenarios and metricsto cater to wide variety of UAV applications and use cases.The study items are as follows:• 5G Enhancement for UAVs :- It includes several KeyPerformance Indicators (KPIs) relevant to UAV ser-vices. The KPIs are provided for command and con-trol, and payload communication.• Study on application layer support for UnmannedAerial System :- This includes the UAS service re-quirements that may have impact on the UAS appli-cation layer. These requirements are in terms of gen-eral requirements, UE capability identification, loca-tion, security etc.• Study on supporting Unmanned Aerial Systems Con-nectivity, Identification, and Tracking :- This studyitem deals with a mechanism that enables the UAStracking and identification within 3GPP systems andUTM.
Ubiquitous accessibility and rapid emergence of UAVtechnology mandate development of regulatory frameworksfor harmonious operation of cellular-connected UAVs in the national and international airspace. Although, each countryhas a specific set of internal rules for UAV operation, fewglobal bodies tend to harmonize their operation across inter-national airspace. The regulatory framework mainly targetaround three key aspects [102]:• To regulate and control the use of unmanned aircraftin the airspace to prevent danger to manned aircraft;• To ensure proper operational limitations to the flight;• To manage and control the administrative privilegessuch as pilot licenses, flight authorizations and datahandling techniques.
The UAV ecosystem leveraging the emerging technolo-gies such as IoT, AI, AR/VR are not much explored bythe manufacturers and their usages are also researched bya handful of organizations. Real-time surveillance is oneof the major use case that has been widely explored by theUAV industry for relaying live information to target audi-ence. The ecosystem is still in its infancy to showcase di-verse capabilities of cellular-connected UAVs. Additionally,the skills necessary for UAV industry to roll out interestinguse case demand sufficient domain training to the equipmentproviders and technical users. It is key to eradicate the bot-tleneck in setting up the ecosystem. Extracting the right setof specifications and requirements from the users is neededto maximize the benefit of the use case and to generate largescale development of cellular-connected UAV applications.
The UAV operation must be properly regulated to protectthe privacy of business organizations as well as individuals.Advancement of drone technology with aerial surveillanceand photography with high definition images and streamingcan easily violate the privacy, even when it is unintentional.The existing regulations to protect privacy may not be suf-ficient due to rapid evolution of UAV technology and its in-creasing capability, thereby further legislation is needed tobe formulated to protect privacy.Most of the UAV use cases deal with gathering a lot of vi-tal data depending upon the application and processing themto extract useful information for taking decisions. Duringa mission, the onboard sensors collect personal or businessdata can be transmitted to a remote location or made livefrom the present location. The data collection capabilitiesmay infringe data protection rules and abuse personal infor-mation without the knowledge of data subject. Additionally,if someone obtains the control of the UAV, the sensors ordata processing circuitry could be tempered for data misuse.Hence, strict guidelines must be governed to protect the per-sonal and business data.In the due course of flight or mission control, any sortof discontinuity in proper command and control poses se-rious safety risks. This may lead to collisions and causesharm to civilians and other UAVs in the vicinity. Colli-sions with manned aircraft can pose serious risks in terms of
D. Mishra et al.:
Preprint submitted to Elsevier
Page 25 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements catastrophic consequence and loss of assets. In case of highdensity urban regions, collisions of UAVs with the groundterrain pose threats to human lives and assets. Hence, thechallenges with respect to public safety must be taken intoaccount and researched thoroughly.
The important lessons learnt from this section are listedas follows:• Cellular-connected UAVs not only impose technicalchallenges, but also require solutions pertaining to pri-vacy, security, licenses, public safety, administrativeprocedures governing them. Standardization bodiessuch as 3GPP have put together study items and work-ing groups in order to harmonize the developmentefforts from industry, academia and independent re-search bodies.• Operation of UAVs over cellular spectrum requiresstrict regulations to operate in national and inter-national airspace without causing trouble to othermanned or unmanned aerial vehicles. There are rulesapplied to control UAV operation that varies withcountries. However, a unified set of rules governingUAV operation in cellular spectrum is still far away.• The commercial production of cellular-connectedUAVs must consider the true requirements and speci-fications to maximize the benefit of a use case.• Care must be taken to safeguard the data collected bythe sensors of UAV and ensuring that it does not in-fringe the privacy of unwanted individuals and orga-nizations. It must be guarded against hackers and ma-licious intruders, whose intent is to control the UAVsfor unauthorized activity, e.g., during aerial surveil-lance and photography.
8. Future Outlooks
In earlier sections, we have outlined integration chal-lenges and highlighted candidate 5G/B5G innovations to ad-dress some of those challenges for cellular-connected UAV.Despite of several studies, there is still considerable areasof open problems that needs to be investigated. The currentsection aims to bring out such future opportunities for re-searchers and shed light on interesting open research topics.
UAVs are expected to be deployed in a wide variety ofindoor and outdoor environments such as stadiums, urban,rural, sub-urban, industrial, over water, highways etc. Allthese environments require unique air-ground propagationconditions, measurement campaigns and channel models forcellular-connected UAVs. However, accurate modelling andcharacterizing each of these unique channels is a non-trivialtask that remains largely unexplored. In addition, introduc-tion of 5G-oriented technologies such as mmWave and mas-sive MIMO systems bring additional factors and scope for improvements in design and development of effective chan-nel models for cellular-connected UAVs.
Onboard energy is a bottleneck in UAV. Recent devel-opments to rechargeable battery cells and use of solar cellsare some of the ways to extend the flying time of UAV. TheUAVs require continuous power source to operate as they usea huge percentage of battery power in flying. Different ac-tions such as transmission, reception, execution of softwarefunction, path planning optimizations consume the batterypower. Most of the existing works on cellular-connectedUAV do not factor this energy limitation during the study.There exists a lot of scope to investigate the performanceof cellular-connected UAVs considering the limited energyconstraints especially in areas of VNF deployment on UAVsfor automated operation, trajectory optimization, learning-based methods, and longevity calculation of the mission ina use case.
Due to open communication links, the cellular-connected UAVs are vulnerable to cyber-physical ormalicious attacks to spoof the control signals. Such at-tempts pose tremendous threat to the UAV system in termsof loss/stealing of confidential information or failure ofmission. The signal spoofing of control signal might haveadverse effect on the UAV mission and making it difficult tobring it back online. Hence, in order to avoid such maliciousmodifications, a relevant open issue is to improvise thesecurity and privacy aspect of UAV cellular communicationthat demands in-depth study of security issues spanning alllayers of the protocol stack.
The UAVs are typically served by the side lobes of cur-rent base stations in 4G and 5G cellular networks and there-fore may easily establish cell selection with far away basestations. Furthermore, the high mobility of UAVs increasesthe failure rates of radio link due to more frequent handovers.The design of antennas to supports flying UAV in high alti-tude and enhanced solutions for radio network planning arerequired by network operators and equipment vendors.
Typically, ground UEs are mobile in 2D space and basestation antennas are optimally designed for ground users.UAVs are typically served by the side lobes of current basestation and their aerial patterns are different resulting inunique handover characteristics, different from ground UEs.The frequent handover pattern is largely dependent on block-age, mobility, altitude variations. There is a need for im-proved handover mechanisms that suit the characteristicsof high mobility UAVs. Additionally, the cell selectionschemes based on nearest or strongest RSRP may no longerbe an appropriate method for cellular-connected UAVs.
D. Mishra et al.:
Preprint submitted to Elsevier
Page 26 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements
Simulations and measurement campaigns are not suf-ficient to fully characterize the performance and workingprinciples of cellular-connected UAVs. The proposed solu-tion approaches to the integration challenges must be com-plemented by extensive field trials and real-world testbed-based evaluation. However, there are not enough workingprototypes to study and evaluate the behaviour of cellular-connected UAVs from practical standpoint. Additionally,there are no experimental or open-source simulator platformavailable till date to assist in a wide range of functionalitytesting of UAVs. Such works need to be pursued in futureto fill in the gap between theoretical proposal and practicalevaluation.
AI and ML-based approaches have been considered aspowerful tools to solve many real-world wireless network-ing problems revolutionizing 5G/B5G networks. On thisadvent, the UAV cellular communication has opened upnew possibility for autonomous UAV operation consider-ing the security, performance and dynamic complex deploy-ment scenarios. Numerous research issues exist in study-ing and evaluating AI/ML-empowered techniques to solvechallenges of UAV-ground interference management, powercontrol, multi-UAV cooperation etc. Moreover, Q-learningapproaches are helpful for UAVs to tackle security issues byadaptively controls the UAV transmission as per the mali-cious type of attacks.
Due to moderate computational capabilities and limitedonboard energy of UAVs, MEC can be helpful for offload-ing computationally heavy tasks from cellular-connectedUAV to edge nodes to improve endurance and life time ofUAV. Some examples of such intense tasks are real-timeface recognition in a crowd surveillance use case. For suchuse case, leveraging MEC capabilities along with UAV, therecognition task can be offloaded to complete the job in atimely manner. Additionally, to save the information fromeavesdropper, proper security mechanisms needs to be inte-grated to the MEC-UAV platform for optimum performance.These research areas are largely unexplored so far and nu-merous scopes exists for design and development of UAV-MEC frameworks for cellular-connected UAVs consideringits diverse use case.
9. Conclusions
In this work, we provided a comprehensive study on theCellular-assisted UAV communication paradigm (Cellular-connected UAV) where UAV is integrated to existing5G/B5G cellular systems as a new aerial UE. The detailedtaxonomy of various application domains with emerginguse cases as well as the technical synergistic challenges ofUAV integration with cellular network are discussed first.Then, we focus on the promising network architectures and physical layer improvements in 5G/B5G systems consider-ing the hardware and software design challenges of Cellular-connected UAVs. The key innovative 5G technologies areelaborated enabling the seamless integration and supportof UAV communication over cellular spectrum. In orderto characterize the design performance benefits and studythe realistic deployment issues, we also highlighted the ef-forts to develop working prototypes as well as the field tri-als and simulations. The progress on standardization ac-tivities by 3GPP, national and international regulations andconcerns pertaining to socio-economic barriers are also dis-cussed which must be accounted before successful adoptionthis new technology. We believe this work will be a veryuseful and motivating resource for researchers working oncellular-connected UAVs in order to unlock a holistic viewand to exploit its full potential.
References [1] K. P. Valavanis, G. J. Vachtsevanos, Handbook of unmanned aerialvehicles, Springer, 2015.[2] R. W. Beard, T. W. McLain, Small unmanned aircraft: Theory andpractice, Princeton university press, 2012.[3] M. Asadpour, B. Van den Bergh, D. Giustiniano, K. A. Hummel,S. Pollin, B. Plattner, Micro aerial vehicle networks: An experimen-tal analysis of challenges and opportunities, IEEE CommunicationsMagazine 52 (7) (2014) 141–149.[4] L. Godage, Global Unmanned Aerial Vehicle Market (UAV) Indus-try Analysis and Forecast (2018-2026), Montana Ledger (2019).[5] S. Hayat, E. Yanmaz, R. Muzaffar, Survey on unmanned aerial vehi-cle networks for civil applications: A communications viewpoint,IEEE Communications Surveys & Tutorials 18 (4) (2016) 2624–2661.[6] R. Shakeri, M. A. Al-Garadi, A. Badawy, A. Mohamed, T. Khat-tab, A. K. Al-Ali, K. A. Harras, M. Guizani, Design challenges ofmulti-UAV systems in cyber-physical applications: A comprehen-sive survey and future directions, IEEE Communications Surveys &Tutorials 21 (4) (2019) 3340–3385.[7] H. Wang, H. Zhao, J. Zhang, D. Ma, J. Li, J. Wei, Survey on un-manned aerial vehicle networks: A cyber physical system perspec-tive, IEEE Communications Surveys & Tutorials (2019).[8] M. Mozaffari, W. Saad, M. Bennis, Y.-H. Nam, M. Debbah, A tu-torial on uavs for wireless networks: Applications, challenges, andopen problems, IEEE communications surveys & tutorials 21 (3)(2019) 2334–2360.[9] Y. Zeng, R. Zhang, T. J. Lim, Wireless communications with un-manned aerial vehicles: Opportunities and challenges, IEEE Com-munications Magazine 54 (5) (2016) 36–42.[10] A. Fotouhi, H. Qiang, M. Ding, M. Hassan, L. G. Giordano,A. Garcia-Rodriguez, J. Yuan, Survey on uav cellular communica-tions: Practical aspects, standardization advancements, regulation,and security challenges, IEEE Communications Surveys & Tutori-als 21 (4) (2019) 3417–3442.[11] P. Marsch, I. Da Silva, O. Bulakci, M. Tesanovic, S. E. El Ayoubi,T. Rosowski, A. Kaloxylos, M. Boldi, 5G radio access network ar-chitecture: Design guidelines and key considerations, IEEE Com-munications Magazine 54 (11) (2016) 24–32.[12] P. Marsch, I. Da Silva, Ö. Bulakci, M. Tesanovic, S. E. El Ayoubi,M. Säily, Emerging network architecture and functional design con-siderations for 5G radio access, Transactions on Emerging Telecom-munications Technologies 27 (9) (2016) 1168–1177.[13] S. D. Muruganathan, X. Lin, H.-L. Maattanen, Z. Zou, W. A.Hapsari, S. Yasukawa, An overview of 3GPP release-15 studyon enhanced LTE support for connected drones, arXiv preprintarXiv:1805.00826 (2018).
D. Mishra et al.:
Preprint submitted to Elsevier
Page 27 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements [14] B. Li, Z. Fei, Y. Zhang, UAV communications for 5G and beyond:Recent advances and future trends, IEEE Internet of Things Journal6 (2) (2018) 2241–2263.[15] I. Bor-Yaliniz, M. Salem, G. Senerath, H. Yanikomeroglu, Is 5Gready for drones: A look into contemporary and prospective wirelessnetworks from a standardization perspective, IEEE Wireless Com-munications 26 (1) (2019) 18–27.[16] S. A. R. Naqvi, S. A. Hassan, H. Pervaiz, Q. Ni, Drone-aided com-munication as a key enabler for 5G and resilient public safety net-works, IEEE Communications Magazine 56 (1) (2018) 36–42.[17] R. J. Kerczewski, J. D. Wilson, W. D. Bishop, Frequency spectrumfor integration of unmanned aircraft, in: 2013 IEEE/AIAA 32ndDigital Avionics Systems Conference (DASC), 2013, pp. 6D5–1.[18] 3GPP TR 36.777, Technical specification group radio access net-work:study on enhanced LTE support for aerial vehicles (V15.0.0,Dec, 2017).[19] Y. Zeng, Q. Wu, R. Zhang, Accessing from the sky: A tutorial onUAV communications for 5G and beyond, Proceedings of the IEEE107 (12) (2019) 2327–2375.[20] E. Vinogradov, H. Sallouha, S. De Bast, M. M. Azari, S. Pollin, Tu-torial on UAV: A blue sky view on wireless communication, arXivpreprint arXiv:1901.02306 (2019).[21] A. Chakraborty, E. Chai, K. Sundaresan, A. Khojastepour, S. Ran-garajan, SkyRAN: a self-organizing LTE RAN in the sky, in: Pro-ceedings of the 14th International Conference on emerging Network-ing EXperiments and Technologies, ACM, 2018, pp. 280–292.[22] K. Sundaresan, E. Chai, A. Chakraborty, S. Rangarajan, SkyLiTE:End-to-End Design of Low-Altitude UAV Networks for ProvidingLTE Connectivity, arXiv preprint arXiv:1802.06042 (2018).[23] Y. Zeng, J. Lyu, R. Zhang, Cellular-connected UAV: Potential, chal-lenges, and promising technologies, IEEE Wireless Communica-tions 26 (1) (2018) 120–127.[24] M. M. Azari, G. Geraci, A. Garcia-Rodriguez, S. Pollin, CellularUAV-to-UAV communications, in: IEEE 30th Annual InternationalSymposium on Personal, Indoor and Mobile Radio Communications(PIMRC), 2019, pp. 1–7.[25] H. Shakhatreh, A. H. Sawalmeh, A. Al-Fuqaha, Z. Dou, E. Almaita,I. Khalil, N. S. Othman, A. Khreishah, M. Guizani, Unmanned aerialvehicles (UAVs): A survey on civil applications and key researchchallenges, IEEE Access 7 (2019) 48572–48634.[26] A. A. Khuwaja, Y. Chen, N. Zhao, M.-S. Alouini, P. Dobbins, Asurvey of channel modeling for UAV communications, IEEE Com-munications Surveys & Tutorials 20 (4) (2018) 2804–2821.[27] W. Khawaja, I. Guvenc, D. W. Matolak, U.-C. Fiebig, N. Schnecken-berger, A Survey of Air-to-Ground Propagation Channel Modelingfor Unmanned Aerial Vehicles, IEEE Communications Surveys &Tutorials (2019).[28] L. Gupta, R. Jain, G. Vaszkun, Survey of important issues in UAVcommunication networks, IEEE Communications Surveys & Tuto-rials 18 (2) (2015) 1123–1152.[29] M. E. Mkiramweni, C. Yang, J. Li, W. Zhang, A Survey of GameTheory in Unmanned Aerial Vehicles Communications, IEEE Com-munications Surveys & Tutorials 21 (4) (2019) 3386–3416.[30] C. Yan, L. Fu, J. Zhang, J. Wang, A comprehensive survey on UAVcommunication channel modeling, IEEE Access 7 (2019) 107769–107792.[31] N. H. Motlagh, T. Taleb, O. Arouk, Low-altitude unmanned aerialvehicles-based internet of things services: Comprehensive surveyand future perspectives, IEEE Internet of Things Journal 3 (6) (2016)899–922.[32] M. M. Azari, F. Rosas, S. Pollin, Reshaping cellular networks for thesky: Major factors and feasibility, in: IEEE International Conferenceon Communications (ICC), 2018, pp. 1–7.[33] G. Geraci, A. Garcia-Rodriguez, M. Hassan, M. Ding, UAV CellularCommunications: Practical Insights and Future Vision (2018).[34] M. M. Azari, F. Rosas, S. Pollin, Cellular connectivity for UAVs:Network modeling, performance analysis, and design guidelines,IEEE Transactions on Wireless Communications 18 (7) (2019) 3366–3381.[35] H. Wang, J. Wang, J. Chen, Y. Gong, G. Ding, Network-connectedUAV communications: Potentials and challenges, China Communi-cations 15 (12) (2018) 111–121.[36] O. Alvear, N. R. Zema, E. Natalizio, C. T. Calafate, Using UAV-based systems to monitor air pollution in areas with poor accessibil-ity, Journal of Advanced Transportation 2017 (2017).[37] O. Alvear, C. Calafate, N. Zema, et al., A discretized approach toair pollution monitoring using uav-based sensing, Mobile NetworkApplications 23 (2018) 1693–1702.[38] M. Erdelj, O. Saif, E. Natalizio, I. Fantoni, UAVs that fly forever:Uninterrupted structural inspection through automatic UAV replace-ment, Ad Hoc Networks 94 (2019) 101612.[39] A. Trotta, F. D. Andreagiovanni, M. Di Felice, E. Natalizio, K. R.Chowdhury, When UAVs ride a bus: towards energy-efficient city-scale video surveillance, in: IEEE Conference on Computer Com-munications (INFOCOM), 2018, pp. 1043–1051.[40] P. Grippa, D. A. Behrens, F. Wall, C. Bettstetter, Drone delivery sys-tems: job assignment and dimensioning, Autonomous Robots 43 (2)(2019) 261–274.[41] E. Milan and K. Michał and N. Enrico, Wireless sensor networksand multi-UAV systems for natural disaster management, ComputerNetworks (2017) 72–86.[42] E. Milan and N. Enrico and C. Kaushik R and A. Ian F, Help fromthe sky: Leveraging UAVs for disaster management, IEEE PervasiveComputing (2017) 24–32.[43] X. Lin, V. Yajnanarayana, S. D. Muruganathan, S. Gao, H. Asplund,H.-L. Maattanen, M. Bergstrom, S. Euler, Y.-P. E. Wang, The sky isnot the limit: LTE for unmanned aerial vehicles, IEEE Communica-tions Magazine 56 (4) (2018) 204–210.[44] M. Erdelj, B. Uk, D. Konam, E. Natalizio, From the Eye of theStorm: An IoT Ecosystem Made of Sensors, Smartphones andUAVs, Sensors 18 (11) (2018) 3814.[45] N. H. Motlagh, M. Bagaa, T. Taleb, UAV-based IoT platform:A crowd surveillance use case, IEEE Communications Magazine55 (2) (2017) 128–134.[46] P. Boccardo, F. Chiabrando, F. Dutto, F. Tonolo, A. Lingua, UAV de-ployment exercise for mapping purposes: Evaluation of emergencyresponse applications, Sensors 15 (7) (2015) 15717–15737.[47] D. He, S. Chan, M. Guizani, Drone-assisted public safety networks:The security aspect, IEEE Communications Magazine 55 (8) (2017)218–223.[48] M. Salhaoui, A. Guerrero-González, M. Arioua, F. J. Ortiz,A. El Oualkadi, C. L. Torregrosa, Smart industrial IoT monitoringand control system based on UAV and cloud computing applied to aconcrete plant, Sensors 19 (15) (2019) 3316.[49] T. Lagkas, V. Argyriou, S. Bibi, P. Sarigiannidis, UAV IoT frame-work views and challenges: Towards protecting drones as “Things”,Sensors 18 (11) (2018) 4015.[50] J. Chakareski, Aerial UAV-IoT sensing for ubiquitous immersivecommunication and virtual human teleportation, in: IEEE Confer-ence on Computer Communications Workshops (INFOCOM WK-SHPS), 2017, pp. 718–723.[51] H. Ullah, N. G. Nair, A. Moore, C. Nugent, P. Muschamp,M. Cuevas, 5G Communication: An Overview of Vehicle-to-Everything, Drones, and Healthcare Use-Cases, IEEE Access 7(2019) 37251–37268.[52] R. Lyu, Jiangbin; Zhang, Network-Connected UAV: 3-D SystemModeling and Coverage Performance Analysis, IEEE Internet ofThings Journal 6 (8 2019).[53] M. Mozaffari, A. T. Z. Kasgari, W. Saad, M. Bennis, M. Debbah, 3Dcellular network architecture with drones for beyond 5G, in: IEEEGlobal Communications Conference (GLOBECOM), 2018, pp. 1–6.[54] X. Xu, Y. Zeng, Cellular-Connected UAV: Performance Analysiswith 3D Antenna Modelling, in: IEEE International Conference onCommunications Workshops (ICC Workshops), 2019, pp. 1–6.[55] M. M. Azari, F. Rosas, A. Chiumento, S. Pollin, Coexistence of ter-restrial and aerial users in cellular networks, in: IEEE Globecom
D. Mishra et al.:
Preprint submitted to Elsevier
Page 28 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements
Workshops (GC Wkshps), 2017, pp. 1–6.[56] W. Mei, Q. Wu, R. Zhang, Cellular-Connected UAV: Uplink Asso-ciation, Power Control and Interference Coordination, IEEE Trans-actions on Wireless Communications 18 (11) (2019) 5380–5393.[57] L. Zhou, Z. Yang, S. Zhou, W. Zhang, Coverage probability analy-sis of UAV cellular networks in urban environments, in: IEEE In-ternational Conference on Communications Workshops (ICC Work-shops), 2018, pp. 1–6.[58] 3GPP TR 38.901, Study on channel model for frequencies from 0.5to 100 GHz (V14.0.0).[59] J. Stanczak, I. Z. Kovacs, D. Koziol, J. Wigard, R. Amorim,H. Nguyen, Mobility challenges for unmanned aerial vehicles con-nected to cellular LTE networks, in: IEEE 87th Vehicular Technol-ogy Conference (VTC Spring), 2018, pp. 1–5.[60] R. Arshad, H. ElSawy, S. Sorour, T. Y. Al-Naffouri, M.-S. Alouini,Handover management in 5G and beyond: A topology aware skip-ping approach, IEEE Access 4 (2016) 9073–9081.[61] C. Zhang, W. Zhang, W. Wang, L. Yang, W. Zhang, Research Chal-lenges and Opportunities of UAV Millimeter-Wave Communica-tions, IEEE Wireless Communications 26 (1) (2019) 58–62.[62] A. Fakhreddine, C. Bettstetter, S. Hayat, R. Muzaffar, D. Emini,Handover challenges for cellular-connected drones, in: Proceedingsof the 5th Workshop on Micro Aerial Vehicle Networks, Systems,and Applications, ACM, 2019, pp. 9–14.[63] U. Challita, W. Saad, C. Bettstetter, Deep reinforcement learningfor interference-aware path planning of cellular-connected UAVs, in:IEEE International Conference on Communications (ICC), 2018, pp.1–7.[64] S. Zhang, Y. Zeng, R. Zhang, Cellular-enabled UAV communica-tion: A connectivity-constrained trajectory optimization perspec-tive, IEEE Transactions on Communications 67 (3) (2018) 2580–2604.[65] N. Senadhira, S. Durrani, X. Zhou, N. Yang, M. Ding, UplinkNOMA for Cellular-Connected UAV: Impact of UAV Trajectoriesand Altitude, arXiv preprint arXiv:1910.13595 (2019).[66] E. Bulut, I. Guevenc, Trajectory optimization for cellular-connectedUAVs with disconnectivity constraint, in: IEEE International Con-ference on Communications Workshops (ICC Workshops), 2018,pp. 1–6.[67] C. Rani, H. Modares, R. Sriram, D. Mikulski, F. L. Lewis, Securityof unmanned aerial vehicle systems against cyber-physical attacks,The Journal of Defense Modeling and Simulation 13 (3) (2016) 331–342.[68] G. Choudhary, V. Sharma, I. You, K. Yim, R. Chen, J.-H. Cho, In-trusion detection systems for networked unmanned aerial vehicles:a survey, in: 14th International Wireless Communications & MobileComputing Conference (IWCMC), IEEE, 2018, pp. 560–565.[69] U. Challita, A. Ferdowsi, M. Chen, W. Saad, Machine learningfor wireless connectivity and security of cellular-connected UAVs,IEEE Wireless Communications 26 (1) (2019) 28–35.[70] B. Nogales, V. Sanchez-Aguero, I. Vidal, F. Valera, J. Garcia-Reinoso, A NFV system to support configurable and automatedmulti-UAV service deployments, in: Proceedings of the 4th ACMWorkshop on Micro Aerial Vehicle Networks, Systems, and Appli-cations, 2018, pp. 39–44.[71] B. Nogales, V. Sanchez-Aguero, I. Vidal, F. Valera, Adaptable andautomated small uav deployments via virtualization, Sensors 18 (12)(2018) 4116.[72] J. Czentye, J. Dóka, Á. Nagy, L. Toka, B. Sonkoly, R. Szabó, Con-trolling Drones from 5G Networks, in: Proceedings of the ACMSIGCOMM Conference on Posters and Demos, 2018, pp. 120–122.[73] F. Zhou, R. Q. Hu, Z. Li, Y. Wang, Mobile edge computing in un-manned aerial vehicle networks, IEEE Wireless Communications(2020).[74] X. Cao, J. Xu, R. Zhang, Mobile edge computing for cellular-connected UAV: Computation offloading and trajectory optimiza-tion, in: IEEE 19th International Workshop on Signal ProcessingAdvances in Wireless Communications (SPAWC), 2018, pp. 1–5. [75] C. Grasso, G. Schembra, A fleet of mec uavs to extend a 5g networkslice for video monitoring with low-latency constraints, Journal ofSensor and Actuator Networks 8 (1) (2019) 3.[76] W. Ejaz, M. A. Azam, S. Saadat, F. Iqbal, A. Hanan, Unmannedaerial vehicles enabled IoT platform for disaster management, Ener-gies 12 (14) (2019) 2706.[77] P. Royo, J. López, C. Barrado, E. Pastor, Service abstraction layer forUAV flexible application development, in: 46th AIAA AerospaceSciences Meeting and Exhibit, 2008, p. 484.[78] A. Koubâa, B. Qureshi, M.-F. Sriti, A. Allouch, Y. Javed, M. Ala-jlan, O. Cheikhrouhou, M. Khalgui, E. Tovar, Dronemap planner: Aservice-oriented cloud-based management system for the internet-of-drones, Ad Hoc Networks 86 (2019) 46–62.[79] J. A. Besada, A. M. Bernardos, L. Bergesio, D. Vaquero, I. Cam-paña, J. R. Casar, Drones-as-a-service: A management architectureto provide mission planning, resource brokerage and operation sup-port for fleets of drones, in: IEEE International Conference on Per-vasive Computing and Communications Workshops (PerCom Work-shops), 2019, pp. 931–936.[80] G. ETSI, 002: Network Functions Virtualisation (NFV); Architec-tural Framework, Group Specification (2014).[81] R. N. Mitra, D. P. Agrawal, 5G mobile technology: A survey, ICTExpress 1 (3) (2015) 132–137.[82] A. Garcia-Rodriguez, G. Geraci, D. López-Pérez, L. G. Gior-dano, M. Ding, E. Bjornson, The essential guide to realizing 5G-connected UAVs with massive MIMO, IEEE Communications Mag-azine (2019).[83] G. Geraci, A. Garcia-Rodriguez, L. G. Giordano, D. López-Pérez,E. Björnson, Understanding UAV cellular communications: fromexisting networks to massive MIMO, IEEE Access 6 (2018) 67853–67865.[84] W. Xia, M. Polese, M. Mezzavilla, G. Loianno, S. Rangan, M. Zorzi,Millimeter Wave Remote UAV Control and Communications forPublic Safety Scenarios, in: 16th Annual IEEE International Confer-ence on Sensing, Communication, and Networking (SECON), 2019,pp. 1–7.[85] W. Khawaja, O. Ozdemir, I. Guvenc, UAV air-to-ground channelcharacterization for mmWave systems, in: IEEE 86th VehicularTechnology Conference (VTC-Fall), 2017, pp. 1–5.[86] L. Bertizzolo, T. X. Tran, B. Amento, B. Balasubramanian, R. Jana,H. Purdy, Y. Zhou, T. Melodia, Live and let Live: Flying UAVsWithout affecting Terrestrial UEs, in: Proceedings of the 21st In-ternational Workshop on Mobile Computing Systems and Applica-tions, 2020, pp. 21–26.[87] M. F. Sohail, C. Y. Leow, S. Won, Non-orthogonal multiple accessfor unmanned aerial vehicle assisted communication, IEEE Access6 (2018) 22716–22727.[88] W. Mei, R. Zhang, Uplink cooperative NOMA for cellular-connectedUAV, IEEE Journal of Selected Topics in Signal Processing 13 (3)(2019) 644–656.[89] A. Rahmati, Y. Yapıcı, N. Rupasinghe, I. Guvenc, H. Dai,A. Bhuyany, Energy efficiency of RSMA and NOMA incellular-connected mmwave UAV networks, arXiv preprintarXiv:1902.04721 (2019).[90] P. J. Burke, A Safe, Open Source, 4G Connected Self-Flying PlaneWith 1 Hour Flight Time and All Up Weight (AUW)< 300 g: To-wards a New Class of Internet Enabled UAVs, IEEE Access 7 (2019)67833–67855.[91] L. Sundqvist, et al., Cellular controlled drone experiment: Evalua-tion of network requirements (2015).[92] G. N. Solidakis, F. M. Tsokas, M. C. Batistatos, N. C. Sagias, G. V.Tsoulos, D. A. Zarbouti, G. E. Athanasiadou, An Arduino-basedsubsystem for controlling UAVs through GSM, in: IEEE 6th Inter-national Conference on Modern Circuits and Systems Technologies(MOCAST), 2017, pp. 1–4.[93] D. Brodn , evs, Development of a Flexible Software Solution for Con-trolling Unmanned Air Vehicles via the Internet, Transport andAerospace Engineering 6 (1) (2018) 37–43. D. Mishra et al.:
Preprint submitted to Elsevier
Page 29 of 30 Survey on Cellular-connected UAVs: Design Challenges, Enabling 5G/B5G Innovations, and Experimental Advancements [94] X. Lin, R. Wiren, S. Euler, A. Sadam, H.-L. Maattanen, S. Muru-ganathan, S. Gao, Y.-P. E. Wang, J. Kauppi, Z. Zou, et al., MobileNetwork-Connected Drones: Field Trials, Simulations, and DesignInsights, IEEE Vehicular Technology Magazine 14 (3) (2019) 115–125.[95] R. Muzaffar, R. Christian, F. Aymen, L. José López, E. Driton,C. Bettstetter, First Experiments with a 5G-Connected Drone, arXivpreprint arXiv:2004.03298 (2020).[96] L. Qualcomm, Unmanned Aircraft Systems—Trial Report (2017).[97] R. Amorim, P. Mogensen, T. Sorensen, I. Z. Kovács, J. Wigard,Pathloss measurements and modeling for UAVs connected to cellu-lar networks, in: IEEE 85th Vehicular Technology Conference (VTCSpring), 2017, pp. 1–6.[98] S. Hayat, C. Bettstetter, A. Fakhreddine, R. Muzaffar, D. Emini, Anexperimental evaluation of LTE-A throughput for drones, in: Pro-ceedings of the 5th Workshop on Micro Aerial Vehicle Networks,Systems, and Applications, 2019, pp. 3–8.[99] R. Amorim, J. Wigard, H. Nguyen, I. Z. Kovacs, P. Mogensen,Machine-learning identification of airborne UAV-UEs based on LTEradio measurements, in: IEEE Globecom Workshops (GC Wkshps),2017, pp. 1–6.[100] R. Amorim, H. Nguyen, J. Wigard, I. Z. Kovács, T. B. Sorensen,P. Mogensen, LTE radio measurements above urban rooftops foraerial communications, in: IEEE Wireless Communications andNetworking Conference (WCNC), 2018, pp. 1–6.[101] H. Marques, P. Marques, J. Ribeiro, T. Alves, L. Pereira, Experi-mental Evaluation of Cellular Networks for UAV Operation and Ser-vices, in: IEEE 24th International Workshop on Computer AidedModeling and Design of Communication Links and Networks (CA-MAD), 2019, pp. 1–6.[102] C. Stöcker, R. Bennett, F. Nex, M. Gerke, J. Zevenbergen, Reviewof the current state of UAV regulations, Remote sensing 9 (5) (2017)459.
D. Mishra et al.: