3D Aerial Highway: The Key Enabler of the Retail Industry Transformation
Nesrine Cherif, Wael Jaafar, Halim Yanikomeroglu, Abbas Yongacoglu
33D Aerial Highway: The Key Enabler of the RetailIndustry Transformation
Nesrine Cherif, Wael Jaafar, Halim Yanikomeroglu, and Abbas Yongacoglu
Abstract —The retail industry is already facing an inevitableand significant transformation worldwide, and with the currentpandemic situation, it is even accelerating. Indeed, consumerhabits are shifting from brick-and-mortar stores to online shop-ping. The bottleneck in the end-to-end online shopping experienceremains the efficient and quick delivery of goods to consumers.In this context, unmanned aerial vehicle (UAV) technology is seenas a potential solution to address cargo delivery issues. Hence, thenumber of cargo-UAVs is expected to skyrocket in the next decadeand the airspace to become densely crowded. To successfullydeploy UAV technology for mass cargo deliveries, a challengeremains to provide seamless and reliable cellular connectivityfor command and control of highly mobile and flexible aircraft.There is an urgent need for organized and “connected” routes inthe sky. Like highways for cargo trucks, 3D routes in the airspacewill be required for cargo-UAVs so that they can fulfill theiroperations safely and efficiently. We refer to these routes hereas 3D aerial highways. In this paper, we thoroughly investigatethe feasibility of the aerial highways paradigm. First, we discussthe motivations and concerns of the aerial highway paradigm.Then, we present our vision of the 3D aerial highway framework.Finally, we discuss related connectivity issues and their solutions.
I. I
NTRODUCTION
Unmanned aerial vehicles (UAVs) are gaining momentumin a wide range of applications, such as search-and-rescue,video surveillance, and on-demand cellular connectivity [1].Specifically, UAVs are helping to solve the logistics of thedelivery industry [2], [3]. With consumer behaviour shiftingtowards online shopping, an exponentially increasing amountof cargo has to be delivered in a timely manner. For instance,Amazon, FedEx, and UPS are delivering approximately 2.5,3, and 4.7 billion US packages every year, respectively . Withalready congested roads, truck deliveries can be delayed bytraffic, which poses a problem for tight delivery deadlines.By contrast, UAV technology is seen as a safe, reliable,and environmentally friendly means of delivery. Indeed, UAVscan support heavy payloads, fly for moderate to long dis-tances using efficient and eco-friendly batteries, and theyare equipped with sophisticated sense-and-avoid techniques toensure collision-free flights.Amazon is among the first to develop a UAV-based platformfor goods delivery in hard-to-reach and remote areas, called“Amazon Prime Air” . Walmart is following Amazon’s lead,and within a few years, UAV-based delivery will be morecommon, not only in remote areas, but also in dense urban centres around the globe. This will create a tremendous volumeof cargo that should be delivered via airspace and in a stringenttime frame. Hence, coordinated path planning for cargo-UAVsis needed to guarantee the safety and fluidity of aerial traffic.Since there are more degrees-of-freedom in planning routes inairspace than on the ground, we define a
3D aerial highway as a set of aerial routes that cargo-UAVs must follow to flyfrom one location to another. For simplicity, the term “route”is similar to a highway, street, or avenue in a road network,whereas “aerial highways” is similar to the road networkitself. The 3D aerial highways will be configured as themain infrastructure that supports the operations of a massivenumber of cargo-UAVs, and this will be achieved by settingup specific aerial routes to guarantee cargo-UAV requirements,such as ubiquitous cellular connectivity, maximum missiontime, and cargo priority. In essence, 3D aerial highways willaddress the issue of coordinating a significant number ofcargo-UAV deliveries by splitting the traffic among differentaerial routes. Unlike conventional routes, 3D aerial highwayswill be reconfigurable to leverage the in-homogeneity of thecargo-UAV traffic. This will lead to an extremely agile andflexible framework where the supply in terms of 3D aerialhighways always match the demand in terms of cargo-UAVoperations.The “creation” of 3D aerial highway in the sky dependsmainly on the cellular connectivity of its routes. Multiplediscussions in the research community have shed light onpotential network architectures to connect UAVs [4]–[7]. Forinstance, reutilizing terrestrial networks to provide aerial cov-erage has been extensively investigated, and realistic channelmodels for cellular ground-to-air channels have been studiedand presented by the Third Generation Partnership Project(3GPP) in its Release 15 [8]. Channel model simulationresults suggest that terrestrial networks may not be adequatefor providing ubiquitous cellular connectivity to cargo-UAVs.Subsequently, different network architectures were proposed,including vertical heterogeneous networks (VHetNets) [4] andstandalone aerial networks enabled by UAV base stations(UAV-BSs). It is not clear yet which type of networks wouldfully support the cellular connectivity requirements of cargo-UAVs in 3D aerial highways while also being economicallyviable.To ensure the safe operations of large-scale cargo-UAVsystems, a massive amount of data about cargo-UAVs has to becontinuously collected and analyzed, including locations, bat-tery levels, flight speeds, and environmental conditions (e.g.,weather conditions, obstacles). Hence, artificial intelligence a r X i v : . [ ee ss . SP ] S e p ocking stationRetailer Warehouse City mapCargo-UAV 3D aerial highway Fig. 1. A 3D aerial highway design. (AI) can be leveraged to analyze the data and provide controlcommands to prevent failures, aerial traffic congestion, andcollisions, to name few.To fully exploit the potential of UAVs for cargo deliveries,we propose here a complete framework for 3D aerial highwaydesign. In addition, we discuss the motivations and concernsrelated to the 3D aerial highway paradigm, our vision, and thesuitability of different cellular networks for providing seamlessand reliable connectivity to cargo-UAVs.II. 3D A
ERIAL H IGHWAYS : M
OTIVATIONS AND C ONCERNS
A. Motivations
Aerial highways are projected to become regulated routesfor cargo-UAVs, where seamless connectivity will be guar-anteed by several aerial communications advancements, e.g.,UAV-BSs, high altitude platforms station (HAPS) systems,and low earth orbit (LEO) satellites. Fueled by the shift inconsumer habits towards online shopping, UAVs are beingpromoted as the future pillar of cargo delivery. Indeed, relyingon traditional delivery systems involving a driver and a vehiclemeans more traffic congestion, and pollution, and health risksto the most vulnerable. This is especially pronounced in timesof disasters and pandemics, such as the current COVID-19pandemic.Also, given the exponentially increasing number of onlineorders, last-mile delivery is becoming a bottleneck for theconsumer’s online shopping experience. This issue is criticalnot only for customers, but also for the survival and prosperityof online businesses and retailers. Hence, adopting this newcargo-UAV delivery mode would be an immense boost foronline stores, besides being faster, more eco-friendly, and morecost-effective than conventional delivery modes.Nevertheless, the massive use of cargo-UAVs would bringtheir own share of issues, such as planning and organizationof aerial routes, control and command (C&C) connectivity, and aerial incidents avoidance. Since very similar concernswere raised in the old days of the car industry boom, whenroad networks were built to enable new use cases for vehicles,we envision that similar steps will need to be followed forcargo-UAVs in supporting the transformation of the retailindustry. Specifically, 3D aerial highways should be carefullydefined and planned on the basis of several criteria, includingregulatory restrictions, layout of the ground/sky area, cellularconnectivity for C&C operations, and cargo priorities.Unlike conventional highways, which are built from scratchand maintained for vehicular traffic, 3D aerial highways defineroutes in the open sky between an origin and a destination.Such routes can be designed for urban areas as well as forrural or hard-to-reach areas. For example, in Fig. 1, we see 3Daerial highways above an urban area, where cargo-UAVs maybe coming to and from a retailer warehouse. The aerial routesare strategically planned at different altitudes according tospecific criteria, e.g., cargo-UAV type, properties, and payload.Moreover, vertical routes are designed to facilitate transitionsbetween highways at different altitudes and also ensure thecontinuous and efficient cellular connectivity of cargo-UAVsto their control centre.3D aerial highways present attractive characteristics. First,they are very flexible and easily reconfigurable. Such qualitiesare handy in situations of cellular network failures or out-of-control ground/sky layout modifications. For instance, ina disaster situation or harsh weather conditions, alternativeroutes may be rapidly configured and sent to the cargo-UAVs,which would allow service continuity and efficient deliveryof sensitive supplies and medicines. Second, energy-efficientC&C can be achieved using cargo-UAV swarms. When acargo-UAV fleet is deployed in a particular area, the UAVsheading in the same direction can delegate a cargo-UAV tomonitor all C&C data exchange with the cellular network, thusreducing the fleet’s cellular communications to a minimum.Third, aerial highways can be shared and organized to supportdelivery operations of different retailers. Since the retailersknow in advance the deliveries they will be making, a centralaerial traffic monitoring system compiles all retailer missionrequests on an hourly/daily basis and plans the routes for alldeliveries.Obviously, achieving accurate organization and coordinationof massive cargo-UAV systems cannot be realized withoutthe involvement of governmental regulations, high safety andprivacy standards, and public endorsement.
B. UAV Regulations
UAV regulatory authorities have developed guidelines forUAV usage. These directives define the procedures to berespected when flying UAVs. These directives establish max-imum UAV weight, maximum altitude allowed, purpose, andminimum spacing from individuals and sensitive services[9, Table I]. However, most guidelines apply to recreationalUAVs and do not specify procedures for commercial UAVs.Regulations for the latter may be more stringent for cargo-UAVs, especially when carrying heavy loads for relatively longistances, since a malfunction may jeopardize the safety ofindividuals on the ground. For this reason, 3D aerial highwaydesign above dense urban areas requires them to adhereto several restrictions, such as no-fly zones, specific pathsto avoid malfunctioning over pedestrians, cars, or sensitivebuildings, and limited cargo payloads.
C. Public Safety
UAV technology raises serious concerns in terms of publicsafety. Unlike well-maintained and secure traditional aviationoperations, UAVs can be inadequately sustained, and thusexperience errors that affect them. Studies have shown thataccident rates involving UAVs is very high, due to collisionswith structures, other aircraft, or trees [10]. Thus, UAVs areseen as a serious risk to communities as they may result inphysical and material damage. To reduce these risks, regu-latory authorities have limited the maximum UAV payload(less than 10 kg) and flying altitude (less than 122 m).When maliciously used, UAVs can easily trigger public servicedisruptions. For instance, in December 2018, more than 1000flights were disrupted and 140 thousand travellers blocked dueto the sighting of a UAV at Gatwick Airport in the UK. Thisincident revealed the extent to which UAV technology canendanger the safety of aviation operations and persons. Forthis reason, extended no-fly zones have been imposed on UAVnear airports and critical services.
D. Privacy and Security
With the development of any new technology, privacyconcerns of individuals and communities must be reassuredand addressed in order for the technology to be deployed suc-cessfully. As UAV technology develops new ways of operating(i.e., aerial capabilities previously reserved for aviation pur-poses and monitored by governmental agencies), the privacyof individuals and communities are under threat. Cargo-UAVsequipped with sophisticated sensors and cameras are contin-uously sensing and collecting data, notably sensitive data,such as location addresses, and neighborhood aerial photos.This data can be hacked or stored in offshore unsecured datacentres. Retailers who operate their online deliveries via cargo-UAVs are responsible for securing all collected data duringdelivery operations and locking it away from cyber-attacks.Also, since C&C operations are expected to go throughcellular networks, they have to be protected against wirelesscontrol hijacking and denial-of-service attacks, which mayresult in UAV failures, causing significant retailer losses.
E. Social Acceptability
The emergence of UAV-based applications has generateda range of responses from the public. Depending on theUAV use case, the public perception and approval has variedsubstantially. Specifically, risk assessment, privacy concerns,and the impact on job security, are the main factors influencingthe social acceptability of the UAV technology. For instance,UAVs can be positively perceived by farmers as they helpensuring food security. However, the use of UAVs in urban areas can be unwelcome as they may be connected with joblosses (e.g., loss of jobs for delivery drivers) and risk toproperties and lives.In a recent survey conducted on the public perception ofUAVs [10], the authors found that the respondents did notoverrate the risk and threat of UAVs, compared to manned air-craft. However, privacy issues, military use, and UAV misusesemerged as the most prevalent public concerns. The authorsconcluded that the public perception of UAVs has yet to beformed and that as UAV technology becomes more mature, itssocial acceptability is expected to evolve positively for variousapplications, including cargo-delivery.III. 3D A
ERIAL H IGHWAY : O UR D ESIGN V ISION
Enabling massive cargo-UAV delivery operations requiresrigorous 3D aerial highway design, where several parametershave to be taken into account, including payload, cargo prior-ity, and cellular connectivity requirements. All these parame-ters shape the demand for efficient aerial routes, characterizedby route coordinates, cellular connectivity level, and autho-rized flying speed, to name few.In Fig. 2, we depict the envisioned 3D aerial highway designframework, including a description of the input parameters,processing unit, and output metrics.
A. Inputs
Prior to defining a 3D aerial highway, a number of inputparameters are collected and processed. The objective is to usethis information to exactly match the supply (i.e. the highwaysand routes) to the demand (i.e., cargo-UAV delivery missions).These parameters include the following: • Cargo weight:
Depending on the cargo weight, the latterwill be assigned to a specific type of UAV that handlesthat weight and will follow a specific itinerary composedof routes dedicated to this range of weights. • Cargo priority level:
Cargos may have different prioritylevels (e.g., standard, premium, or urgent), causing thedelivery to be scheduled differently during the day, and/orput on a different priority itinerary. • Cargo confidentiality level:
Cargo content may have dif-ferent levels of confidentiality. For instance, delivering of-ficial documents to citizens (e.g., passports, government-issued ID) should be treated with high end-to-end securitymeasures. Consequently, secured aerial routes should beavailable to support critical delivery operations. • Cargo drop-off locations:
Aerial routes are expectedto link the retailer warehouse to any possible shippingaddress within the UAV’s flying range. With this infor-mation, drop-off locations can be clustered to streamlineUAV fleet assignments and minimize C&C data exchangein beyond visual line-of-sight (BVLoS) operations. • Cargo maximum delivery mission time:
Consumerstypically expect very fast deliveries, e.g., the same dayand/or less than two hours, according to Amazon. Conse-quently, significant pressure is put on the delivery process,where the maximum cargo-UAV delivery time, i.e., the
D Aerial Highway Design Framework
Cellular connectivity requirements Cargo weight Cargo priority level Cargo confidentiality level Cargo drop-off locations Cargo maximum delivery mission time
Inputs Cargo-UAV maximum payload Cargo-UAV battery autonomy
City centre airspace map
Data extractionData extraction
3D aerial highways processing unit
Objective:
Design efficient cargo-UAV delivery routes
Outputs
Routes description:
3D coordinates of routes Authorized cargo weight range per route Authorized flying speed per route Cargo priority per route Cellular connectivity KPIs per route Number of lanes per routeEmergency routes Real-world environmentCargo-UAVs missions execution
Feedback data (quality of routes, errors, unexpected events..)
Artificial intelligence agent Cargo-UAV traffic patternsShopping ordershistory
Logistics data
Fig. 2. Our vision of the 3D aerial highway design framework.
Integrated Airspace
Low-Speed Localized TrafficHigh-Speed Transit -61 meters-122 meters -152 meters
No-Fly Zone P r e d e f i n e d L o w R i s k L o c a t i o n Fig. 3. Amazon’s airspace segregation model. time between leaving the warehouse with the order untildelivering it to the customer, becomes crucial to the end-to-end shopping experience. • Cargo-UAVs traffic patterns:
From shopping order his-tory and logistics data, traffic patterns can be extracted,which characterize the density of order traffic by area.This information is crucial in accurately designing aerialroutes. For instance, an area where the number of shop-ping orders is higher than average should be supplied bya higher number of routes to avoid aerial congestion. • Cargo-UAV maximum payload:
Available UAVs foraerial cargo transportation have different payload capa-bilities, which currently range from a few hundred gramsto 10 kilograms. For accurate 3D aerial highway design,routes for different ranges of cargo weights, shapes, andsolidity have to be defined, while guaranteeing the safetyof individuals and properties on the ground. • Cargo-UAV battery autonomy:
The cargo-UAV on-board energy lifetime is a critical limitation that shouldbe considered in the design of aerial routes. Intuitively, farther drop-off locations would be carried by cargo-UAVs with higher flying autonomy. • Cellular connectivity requirements:
For BVLoS op-erations, reliable cellular connectivity is required. Thelatter would allow C&C data exchange and cargo-UAVcontinuous localization to prevent any issue during themission. The connectivity requirements mainly includeend-to-end communication delays with the control sta-tion, in the order of tens of milliseconds, and the tolerabledisconnectivity rate of the itinerary, defined as the ratioof the flight duration without any cellular connection tothe total flight time. • City centres airspace map:
The airspace 3D city map,where cargo-UAVs perform their delivery missions, is amixture of the city’s urban topography (i.e., buildings,streets) and the defined areas of the airspace. These areascan be delimited as suggested by Amazon and shown inFig. 3. Specifically, cargo-UAVs operating beyond visualline-of-sight travel between altitudes of 61 m and 122m in the “High-Speed Transit Zone”, while recreationaland non-transit activities, which require VLoS control,fly below 61 m in the “Low-Speed Localized Traffic”area. The airspace between 122 m and 152 m is apermanent “No-Fly Zone”, except for emergencies. Fi-nally, Amazon’s model includes a “Predefined Low RiskLocations” area, such as designated by the Academy ofModel Aeronautics.
B. 3D Aerial Highway Design Processing Unit
The processing unit is the core of the 3D aerial routedesign system. After extracting the input parameters from theshopping order history and retailer logistics data, it compilesthem to achieve an optimized 3D highway design, e.g., itderives the 3D coordinates of aerial routes, associates eachroute to a specific cargo weight range, and assigns prioritylevels, confidentiality levels, and cellular connectivity keyperformance indicators (KPIs) to routes. The outputs of therocessing unit, described below, can be fed back and re-processed to improve the route design. Also, given that theairspace environment changes, for instance, due to differentweather conditions, varying shopping traffic patterns, etc.,an advanced artificial intelligence agent can be leveraged toupdate and reconfigure new and efficient aerial routes.
C. Outputs
The processing unit produces several metrics, which aresummarized as follows: •
3D coordinates of routes:
As in road networks, aerialroutes have to be identified mainly by their 3D coordi-nates. In a more sophisticated system, it is expected thatroutes will be given identifications such as aerial streetnames and highway numbers. • Cargo priority per route:
Each aerial route is assigneda priority level that, for convenience, would allow flyingcargos with the same priority or higher. • Authorized cargo weight range per route:
Each aerialroute is envisioned to support a range of cargo weightsas it depends on its altitude and regulations in place. • Authorized flying speed per route:
UAVs may movealong aerial highways at different speeds due to theircharacteristics. To reduce collision risks, routes can bedivided for different speed ranges, e.g., very fast, fast,moderate, slow, and very-slow, within the regulationlimits. • Cellular connectivity KPIs per route:
Each route willbe characterized by cellular connectivity KPIs, that areexpected to exceed the requirements provided by inputs. • Number of lanes per route:
Dense neighborhoods mayhave a massive number of cargo deliveries and thus aerialroutes may need to support a substantial number of cargo-UAVs simultaneously. Hence, several lanes on the sameroute may be designed to make significant gains in thedelivery times. • Emergency routes:
The 3D aerial highway design shouldencompass emergency routes for unexpected situationsthat may be faced by the cargo-UAVs during their mis-sions, such as longer than usual cellular disconnectivitytimes, and cargo-UAV malfunctioning.IV. C
ELLULAR C ONNECTIVITY FOR
3D A
ERIAL H IGHWAYS : A C
LOSE L OOK
In this section, we focus on the cellular connectivity issueof cargo-UAVs in 3D aerial highways. Specifically, we dis-cuss different network architectures that may support massivecargo-UAV operations.
A. Existing Terrestrial Network
Since conventional terrestrial networks were designed anddeployed for the sole purpose of covering terrestrial users,their aerial coverage is somewhat limited and unreliable asthere are many gaps in coverage in the sky. Indeed, with itsantennas tilted down towards terrestrial users, terrestrial-BSs
500 1000 1500 2000 2500 300050010001500200025003000
Fig. 4. Terrestrial-BS aerial coverage in an area of 3000 x 3000 metersusing UMA (urban macro) environment channels model [8] and 3D antennaradiation pattern [11] to connect a cargo-UAV at altitude of 80 meters.Terrestrial-BSs are identified by the black filled circles with their IDs, andthe color-map designates the coverage color for the terrestrial-BS identifiedby its ID. ID 0 in the color-map refers to the color of the coverage holes. provide limited coverage in the sky with their antenna side-lobes. Moreover, connectivity of UAVs may be affected by thestrong line-of-sight (LoS) interference from other terrestrial-BSs [8]. Thus a UAV may be connected to a distant BS thatprovides better signal quality than a nearby BS.In Fig. 4, we depict the aerial coverage heat map of terres-trial BSs for a cargo-UAV flying at altitude 80 m, where thedownlink communication is considered successful when thesignal-to-interference (
SIR ) is above a fixed threshold of 5 dB,i.e., cellular coverage is provided. First, we can see that thereare most likely gaps in aerial coverage (disconnectivity areasare identified using the color ID 0). Moreover, to corroboratethe view that a BS serving a cargo-UAV may not be the closestone, we see for instance that the coverage region of BS 10 (inyellow) is far from it, since it is provided with the side-lobes.
B. Dedicated Terrestrial Network
As in aviation, where all communications are supportedby dedicated terrestrial networks, cargo-UAVs traveling in 3Daerial highway can be served by a similar cellular networkdesign. To this end, terrestrial-BSs with antennas tilted up tothe sky can be deployed to cover aerial routes. However, thecoverage of such a dedicated network may face significantchallenges when the aerial routes are dynamically reconfigureddue to traffic pattern changes or other unforeseen events.For instance, when a route starts to be frequently reusedand congested with a high number of cargo-UAVs due to achange in customer shopping patterns, new lanes and/or routeshave to be rapidly configured to support this additional trafficload. The cellular connectivity for the new routes has to beguaranteed, which means that terrestrial-BS coverage of thesky has to be potentially reconfigured. Such flexibility may notbe available with a dedicated terrestrial network. Furthermore,this option may not be the most economically attractive dueo the need for high capital and operational expenditures.
C. UAV-BS Aerial Network
Recently, the use of UAV-BSs to provide cellular connec-tivity has gained much attention since they were promoted ascomplementary to terrestrial networks in several use cases. Forinstance, when a spike in data rate demand occurs due to atemporary event, e.g., a concert or sporting event, UAV-BSscan be easily deployed to support the extra traffic load [12].In cargo-UAV systems, connectivity can also be supported byUAV-BSs. More precisely, the latter can be placed along aerialroutes at strategic locations to provide cellular connectivity tocargo-UAVs. In such a design, the UAV-BS antenna main-lobeshave to be aligned with the cargo-UAV routes, i.e., horizon-tally or vertically [13]. When new routes are configured, themobility of the UAV-BS allows it to move freely to a moreadequate location, thus guaranteeing connectivity for cargo-UAVs along the new routes.Although UAV-BSs have extra degrees-of-freedom, i.e.,deployment flexibility and mobility, they suffer from limitedflying times, typically ranging from 30 minutes to few hours,which complicates their utility. To bypass these constraints,researchers and industry players are investigating several op-tions, including on-demand deployments depending on cargo-UAV traffic variations, on-the-fly UAV-BS swapping, batteryswapping, laser charging, and threaded UAV-BSs.
D. LEO Satellite Network
With the evolution of space technology today, the costs ofsatellite production and deployment are significantly reduced.This has made LEO satellites more attractive for providingubiquitous and low-latency communications. SpaceX has takenthe lead with its Starlink project, which aims to deploythousands of LEO satellites to provide Internet connectivityover the globe. In the context of 3D aerial highways, sucha network design would be beneficial, especially in ruraland hard-to-reach areas. However, for UAV applications, LEOcommunications face several challenges that need to be re-solved. For instance, satellite pointing loss due to satellitevibrations or imperfect tracking-and-stabilization mechanismsmay affect the quality of communications. In addition, the RFchannel strength can be affected by several factors, such asweather condition attenuation, frequency, absorption and scat-tering due to various gas molecules and aerosols in the Earthsatmosphere, and atmospheric turbulence. Consequently, LEOsatellite networks can be seen, for now, as a complementarynetwork when the main network that connects the cargo-UAVsfails partially or totally.
E. HAPS Aerial Network
An interesting alternative to LEO satellites is a HAPSsystem, which offers similar performance but with fewerconstraints. In recent publications, HAPS have been proposedto act as super macro-BSs with large coverage footprints (upto 100 km) [14]. HAPS systems operate in the stratosphere ata typical altitude of 20 km, fueled mainly by solar panels and rechargeable batteries. They can stay aloft at a quasi-stationarylocation, thus providing significant benefits over LEO satellitesto achieve the goal of ubiquitous connectivity. The deploy-ment of HAPS systems was initially planned for rural areasand disaster relief applications, as in the case of Google’sLoon project. However, HAPS use cases are currently beingextended to advanced wireless communications services withultra-wide coverage and high capacity for all areas, urban andrural [15]. Owing to these capabilities, HAPS systems can actas an adequate cellular connectivity platform for cargo-UAVstraveling in 3D aerial highways, since a HAPS can guarantee areliable wide coverage with relatively low-latency, especiallyin densely-populated areas where thousands of cargo-UAVsare expected to be flying around daily.In summary, each of these networks has its advantages anddrawbacks in providing cellular connectivity for the 3D aerialhighway paradigm. Nevertheless, we envision that a practicaldeployment for cargo-UAVs will be supported by at leasttwo different types of networks, which will provide reliableconnectivity and safe operation in the airspace. For the sakeof clarity, we provide the pros and cons of these solutions inTable I below. V. C
ONCLUSION
In this article, we discussed our vision of a 3D aerialhighway paradigm, which will be the main enabler of theretail industry transformation. First, we discussed the mainmotivations and concerns related to our vision. Then, wedetailed our 3D aerial highway framework that is designedto enable the coordinated and dynamic planning of routes forcargo-UAVs. Finally, we discussed the related issue of cellularconnectivity and evaluated possible solutions. For 3D aerialhighways to operate safely and effectively, we recommendsupporting cargo-UAVs with at least two types of wirelessnetworks. A
CKNOWLEDGEMENTS
This work is funded in part by Huawei Canada and inpart by the Natural Sciences and Engineering Council Canada(NSERC). The authors would like to thank Dr. GaminiSenarath, Huawei Canada, for insightful discussionsR
EFERENCES[1] M. Mozaffari, W. Saad, M. Bennis, Y.-H. Nam, and M. Debbah, “Atutorial on UAVs for wireless networks: Applications, challenges, andopen problems,”
IEEE Commun. Surveys Tuts.
IEEE Access , vol. 7, pp. 49 191–49 200, 2019.[4] N. Cherif, M. Alzenad, H. Yanikomeroglu, and A. Yongacoglu, “Down-link coverage analysis of an aerial user in vertical heterogeneousnetworks,” in
Proc. IEEE Glob. Commun. Conf. (Globecom) , Waikoloa,HI, USA, Dec. 2019, pp. 1–6.[5] M. M. Azari, F. Rosas, A. Chiumento, and S. Pollin, “Coexistence ofterrestrial and aerial users in cellular networks,” in
Proc. IEEE Glob.Commun. Conf. (Globecom) Workshops , Singapore, Dec. 2017, pp. 1–6.ABLE IC
OMPARISON BETWEEN DIFFERENT TYPES OF CELLULAR NETWORKS FOR CARGO -UAV
CONNECTIVITY
Type of network Pros Cons
Existing terrestrialnetwork • $0 deployment cost• Mature technology• Reliable backhaul links • Limited cellular aerial coverage• Fixed deployment of BSs• Complex management for aerial and ground usersDedicated terrestrialnetwork • Strong cellular aerial coverage• Mature technology• Reliable backhaul links • Costly and fixed BS deployment• Complex management for dynamic 3D aerial routesUAV-BS aerial network • Easy and quick on-demand deployment• Reliable cellular coverage for aerial routes• Flexible reconfiguration fordynamic 3D aerial routes • Potentially unstable backhaul links• Limited on-board processing power on UAV-BSs• Limited flight duration of UAV-BSsLEO satellite network • Ubiquitous cellular coverage• Potentially supports low-latency links(LEO round trip delay between 2.66 msand 13.33 ms [14, Table I]) • Several challenges still unresolved• Potentially unstable backhaul linksHAPS aerial network • Quasi-static location and operation for longperiods of time (up to 6 months)• Reliable wide cellular coverage• Easy location update• Potentially supports low-latency links(HAPS round trip delay between 0.13 msand 0.33 ms [14, Table I]) • Potentially unstable backhaul links [6] R. Amer, W. Saad, H. ElSawy, M. M. Butt, and N. Marchetti, “Cachingto the sky: Performance analysis of cache-assisted CoMP for cellular-connected UAVs,” in
Proc. IEEE Wireless Commun. Net. Conf. (WCNC) ,Apr. 2019, p. 16.[7] N. Cherif, W. Jaafar, H. Yanikomeroglu, and A. Yongacoglu, “On theoptimal 3D placement of a UAV base station for maximal coverage ofUAV users,” in
Proc. IEEE Glob. Commun. Conf. (Globecom) , Taipei,Taiwan, Dec. 2020, pp. 1–6.[8] 3GPP, “Study on enhanced LTE support for aerial vehicles (release 15),TR 36.777,” Jun. 2018.[9] A. Fotouhi, H. Qiang, M. Ding, M. Hassan, L. G. Giordano, A. Garcia-Rodriguez, and J. Yuan, “Survey on UAV cellular communications:Practical aspects, standardization advancements, regulation, and securitychallenges,”
IEEE Commun. Surveys Tuts. , vol. 21, no. 4, pp. 3417–3442,Firstquarter 2019.[10] R. A. Clothier, D. A. Greer, D. G. Greer, and A. M. Mehta, “Riskperception and the public acceptance of drones,”
Risk Analysis , vol. 35,no. 6, pp. 1167–1183, Feb. 2015.[11] 3GPP, “Study on channel model for frequencies from 0.5 to 100 GHz(release 16), TR 38.901,” Dec. 2019.[12] I. Bor-Yaliniz and H. Yanikomeroglu, “The new frontier in RANheterogeneity: Multi-tier drone-cells,”
IEEE Commun. Mag. , vol. 54,no. 11, pp. 48–55, Nov. 2016.[13] V. V. Chetlur and H. S. Dhillon, “Coverage and rate analysis of downlinkcellular vehicle-to-everything (C-V2X) communication,”
IEEE Trans.Wireless Commun. , vol. 19, no. 3, pp. 1738–1753, Mar. 2020.[14] M. S. Alam, G. K. Kurt, H. Yanikomeroglu, P. Zhu, and N. D.Dao, “High altitude platform station based super macro basestation (HAPS-SMBS) constellations,” 2020. [Online]. Available:https://arxiv.org/abs/2007.08747[15] G. Kurt, M. G. Khoshkholgh, S. Alfattani, A. Ibrahim, T. S. J. Darwish,M. S. Alam, H. Yanikomeroglu, and A. Yongacoglu, “A vision andframework for the high altitude platform station (HAPS) networks ofthe future,” 2020. [Online]. Available: https://arxiv.org/abs/2007.15088 B IOGRAPHIES
Nesrine Cherif [S] ([email protected]) is the recipient of theFull Admission Scholarship for International Students for her PhDstudies at uOttawa. Her main research interests include performanceanalysis and optimization of wireless and non-terrestrial networks.
Wael Jaafar [SM] ([email protected]) is an NSERC Post-doctoral Fellow in the Systems and Computer Engineering Depart- ment of Carleton University. His research interests include wirelesscommunications, edge caching and computing, and machine learning.
Halim Yanikomeroglu [F] ([email protected]) is a professorat Carleton’s University, Canada. His research interests cover manyaspects of 5G/5G+ wireless networks. He is a Fellow of IEEE, theEngineering Institute of Canada (EIC), and the Canadian Academyof Engineering (CAE), and he is a Distinguished Speaker for IEEECommunications Society and IEEE VT Society.