Survey on Aerial Radio Access Networks: Toward a Comprehensive 6G Access Infrastructure
Nhu-Ngoc Dao, Quoc-Viet Pham, Ngo Hoang Tu, Tran Thien Thanh, Vo Nguyen Quoc Bao, Demeke Shumeye Lakew, Sungrae Cho
11 Survey on Aerial Radio Access Networks: Toward aComprehensive 6G Access Infrastructure
Nhu-Ngoc Dao, Quoc-Viet Pham, Ngo Hoang Tu, Tran Thien Thanh,Vo Nguyen Quoc Bao, Demeke Shumeye Lakew, and Sungrae Cho
Abstract —Current network access infrastructures are char-acterized by heterogeneity, low latency, high throughput, andhigh computational capability, enabling massive concurrent con-nections and various services. Unfortunately, this design doesnot pay significant attention to mobile services in underservedareas. In this context, the use of aerial radio access networks(ARANs) is a promising strategy to complement existing ter-restrial communication systems. Involving airborne componentssuch as unmanned aerial vehicles, drones, and satellites, ARANscan quickly establish a flexible access infrastructure on demand.ARANs are expected to support the development of seamlessmobile communication systems toward a comprehensive sixth-generation (6G) global access infrastructure. This paper providesan overview of recent studies regarding ARANs in the literature.First, we investigate related work to identify areas for furtherexploration in terms of recent knowledge advancements andanalyses. Second, we define the scope and methodology of thisstudy. Then, we describe ARAN architecture and its fundamentalfeatures for the development of 6G networks. In particular, weanalyze the system model from several perspectives, includingtransmission propagation, energy consumption, communicationlatency, and network mobility. Furthermore, we introduce tech-nologies that enable the success of ARAN implementations interms of energy replenishment, operational management, anddata delivery. Subsequently, we discuss application scenariosenvisioned for these technologies. Finally, we highlight ongoingresearch efforts and trends toward 6G ARANs.
Index Terms —6G, access infrastructure, aerial radio accessnetwork, unmanned aerial vehicles.
I. I
NTRODUCTION
The Internet of things (IoT) requires most electronic devicesused in our daily lives to have Internet capability. In recentdecades, macro shifts in the technological requirements of mo-bile services have occurred with each network generation [1].The first wave realizes service heterogeneity transformationfrom voice-only services in the first generation (1G) to mul-timedia services in the third generation (3G). The secondwave focused on improvements in quality of service (QoS)
N.-N. Dao is with the Department of Computer Science and En-gineering, Sejong University, Seoul 05006, Republic of Korea (e-mail:[email protected]).Q.-V. Pham is with the Korean Southeast Center for the 4th Industrial Rev-olution Leader Education, Pusan National University, Busan 46241, Republicof Korea (e-mail: [email protected]).N. H. Tu and T. T. Thanh are with the Department of Computer Engineering,Ho Chi Minh City University of Transport, Ho Chi Minh City 710372,Vietnam (e-mail: [email protected], [email protected]).V. N. Q. Bao is with the Posts and Telecommunications Institute of Technol-ogy, Ho Chi Minh City 710372, Vietnam (e-mail: [email protected]).D. S. Lakew and S. Cho are with School of Computer Science andEngineering, Chung-Ang University, Seoul 06974, Republic of Korea (e-mail:[email protected], [email protected]). in terms of peak throughput maximization, end-to-end latencyminimization, network connectivity maximization, and userserviceability, which have been the key challenges for thirdto fifth-generation (5G) networks [2], [3]. In recent years, athird wave has begun to address concerns regarding accessto ubiquitous mobile technology services which have becomeintegral to our lives, developing from a foundation beingestablished by the IoT paradigm in (sub)urban areas with 5Gtoward Internet globalization in the sixth generation (6G) andbeyond. For this purpose, the International TelecommunicationUnion (ITU) has launched the Network-2030 initiative todiscuss potential technologies and innovation [4].Internet globalization aims to equip all users with Internetcapability [5]. Primary examples include event-based commu-nications where vast numbers of users assemble in one location(e.g., concerts, festivals, stadiums, and squares), which requirea tremendous increase in network service connections fora short time, thereby potentially overloading existing com-munication infrastructure. Another important case is disas-ter communications in which network infrastructure may bedamaged or destroyed and thus rendered unable to provideservice, nonetheless communication remains critical to supportsearch and rescue (SAR) operations. From another perspective,scheduled communications may iteratively establish servicesas planned periodically, such as environmental sensing reportsin intelligent agriculture and aerial forest surveillance. A fixedinfrastructure is not necessarily suitable for such scenarios,which call for a mobile portable infrastructure instead. Finallyyet importantly, sparsely populated communities (e.g., in iso-lated areas and at sea) especially would benefit from improvedInternet access capabilities for emergency and informationalcommunications [6].
A. Motivation
Unfortunately, current access infrastructure designs havenot comprehensively considered network serviceability in theaforementioned scenarios. Such serviceability has not beenincluded in the requirements for the most recent telecommuni-cation technology, 5G [1], [7]. In particular, the ITU IMT-2020specification defines that standard 5G infrastructure is to targeteither network performance improvements or QoS such asconnectivity, throughput, latency, reliability, energy efficiency,and mobility [8]. However, the responsibility of broadeningaccess to Internet service is commonly seen as a businessconcern instead of a crucial technical requirement. As a result,mobile services in underserved areas have not been given a r X i v : . [ c s . N I] F e b significant attention to accommodate these communicationneeds.In reality, these scenarios have been only partially andincompletely addressed, in an unsystematic fashion. The fun-damental advantages of aerial networking infrastructures suchas mobility, better channel access, improved coverage, anda higher probability of line-of-sight (LoS) signal propaga-tion compared to terrestrial networks serve as a promisingfoundation to overcome the current challenges. For instance,owing to their quick and flexible three-dimensional (3-D)deployment capabilities, unmanned aerial vehicles (UAVs)have been involved in wireless systems for multiple pur-poses such as communication relaying, coverage expansion,and traffic alleviation [9]. To support various applicationscenarios, multiple UAVs may collaborate and establish dif-ferent networking topologies including mesh, star, bus, andhierarchical architectures with respect to the communicationrequirements and the capabilities of the UAVs. Conversely,satellite communications have been exploited to provide userswith internetworking services in remote areas for the lastdecade [10]. In this system, ground stations typically act astransfer nodes that connect to the satellite constellation andrelay communications to user devices. Recently, an ambitiousproject (namely Starlink ) being constructed by SpaceX Corp.began establishing a constellation of thousands of small satel-lites in low Earth orbit (LEO) to deliver high-speed broadbandInternet directly to end users across the globe. Individually,these solutions have supported several specific applicationswith a basic degree of success. However, there has not beena systematic integration of such infrastructures into mobilenetworks as native components that can seamlessly provideglobal communications.The above observation leads to the necessity of study-ing current proposals and reorganizing such aerial accessarchitectures toward a comprehensive access infrastructure for6G networks. Because airborne objects are involved in thesesystems as primary transceiver components, we introduce anappropriate definition for this type of infrastructure, i.e., aerialradio access network (ARAN) (see Fig. 1). ARANs providea radio access medium from the sky to end users for Internetservices using aerial base stations (ABSs). Typical ABSs in-clude aircraft and airships such as UAVs, drones, balloons, andairplanes equipped with wireless transceiver antennas, whilethe backhaul links may be provided by (miniaturized) satellitesand terrestrial macro base stations. Conceptually, unificationof existing systems to form a multitier and hierarchical ARANis expected to help provide a comprehensive and standardreference model for future research. B. 6G Visions Toward a Comprehensive Access Infrastructure
Specifically, visions of 6G wireless systems are discussedin some recent surveys [11]–[13]. Saad et al. [11] provideda vision of 6G applications and technologies and predictedthat completely integrating terrestrial, airborne, and satellitecommunications will play a vital role in 6G wireless systems.In this context, low-altitude platforms (LAPs) complement underserved areas on the ground with additional connectivity,while both LAPs and terrestrial networks may benefit fromhigh-altitude platforms (HAPs) and satellite communicationsfor backhaul links. Giordani et al. [12] speculated on a numberof important 6G use cases (e.g., augmented/virtual reality,teleoperation, unmanned mobility, and Industry 4.0) and as-sociated technologies (e.g., full-duplex communications, out-of-band channel estimation, and sensing-based localization).To provide Internet connectivity to rural areas, Yaacoub etal. [13] reviewed technologies for front/back-end interfaceconnectivity (e.g., 5G, drones/UAVs, satellite, and free-spaceoptics) and the challenges of efficiently deploying electricalgrids.Motivated by the successes of artificial intelligence (AI) inmany engineering disciplines, many believed that AI wouldrevitalize 6G wireless systems by providing intelligent solu-tions [14], [15]. Holographic beamforming, orbital angularmomentum multiplexing, and the Internet of (Bio-)Nano-Things were also identified as promising 6G technologies [16].To accelerate 6G research, the Finnish 6G Flagship program recently published 12 specific white papers on various researchthemes, for instance, applications of machine learning for 6Gin [17], edge intelligence in [18], 6G broadband connectivityin [19], and massive IoT connectivity in [20]. To efficientlyprovide operational environments for these foundational tech-nologies, existing mobile systems will need to be supple-mented toward a comprehensive infrastructure in 6G [21].With an expected extraordinary popularization of Internetcapability on daily-life devices in future 6G networks, verticaland horizontal communication coverage has never stoppedexpanding to improve serviceability to accommodate massivenumbers of users and services. To this end, a ubiquitous3-D coverage model was introduced in [16] and [22] inte-grating different aspects of the architecture such as aerial,terrestrial, and underground communications into a unifiedaccess platform. A combination of existing infrastructures andsupplemented components forming such a unified platform de-fines comprehensive 6G access infrastructures. Unfortunately,the aforementioned technical recommendation documents havenot yet focused on standardizing requirements toward these 6Ginfrastructures.Directly interacting with end users, radio access networks(RANs) play important roles in infrastructure evolution. Somesurveys over the last decade have been dedicated to reviewingRANs, and recently there have been some speculative studieson 6G. However, they fail to provide an up-to-date survey onARANs. Several aspects of RAN in 5G networks have beenreviewed in the existing literature, for example, radio resourcemanagement in [23] and control-data separation architecturein [24]. Considered a key enabling technology to enhance 5Gperformance and cost efficiency, cloud RANs (CRANs) werewell-reviewed in some survey articles such as [25]–[27]. Inparticular, Peng et al. [25] presented various CRAN archi-tectures, key techniques (e.g., front-end interface compressionand collaborative processing), and cooperative resource allo-cation studies for CRAN systems. A comprehensive tutorial ComSatCubeSat AircraftUAVUnderground RANsTerrestrial RANs Aerial RANs Low Earth orbit (LEO) communications High-altitude platform (HAP) communicationsLow-altitude platform (LAP) communicationsTerrestrial communicationsUnderground(/water) communicationsThe core of the Earth 500 ~ 1500 km20 ~ 50 km0 ~ 10 km0 ~ 100 m– 1500 ~ 0 m
Fig. 1. ARANs in a comprehensive 6G access infrastructure. on CRAN optical fronthaul in 5G cellular networks wasprovided in [26]. As 5G-and-beyond (B5G) networks providecommunication, computation, caching, and control [28], thepairing of edge computing and RAN is a promising solutionand has been investigated in many research works [29], [30].In particular, a Fog-RAN architecture was proposed in [29],and the integration of multiaccess edge computing (MEC) andRAN was investigated in [30]. Moreover, Because end devicesare increasingly capable of accessing multiple RANs, hetero-geneous RANs have been studied in many research works [31].Recently, various technologies and network architectures havebeen proposed for B5G and 6G RANs, for example, millimeterWave (mmWave) and terahertz (THz) communications in [32],[33], new radio and unlicensed spectrum in [34], blockchain-enabled RAN (BRAN) in [35], and AI-defined RAN in [36].There have been surveys on aerial communications, butthey are limited and/or specific to particular topics. For in-stance, satellite and CubeSat communications were discussedin [37] and [38], respectively, opportunities and challengesfrom the integration of terrestrial and aerial communicationswere reviewed in [39], UAV wireless communications andnetworking in [40], hierarchical airborne (i.e., satellites, HAPs,and LAPs) wireless communications in [41]. Although thedemand for a comprehensive 6G access infrastructure in thenext 10 years is a matter of urgency, we are not aware ofany surveys on ARANs, and this motivates us to conduct thepresent survey. To summarize, the state-of-the-art surveys onaerial communications and major contributions offered by thiswork are provided in Table I. • ARANs in 6G access infrastructure • System architecture and reference model • Fundamental features • Transmission propagation • Energy consumption • Communication latency • Mobility analysis • Energy refills • Operational management • Data delivery • Event-based communications • Scheduled communications • Permanent communicationsI.B. Related work investigationII. Network designIII. System model analysisIV. Enabling technologiesV. Application scenariosVI. Research challenges
Fig. 2. Survey framework and major topic organization.
C. Research Scope and Methodology
To provide a comprehensive overview of ARAN technologyin the context of 6G, this study focuses on clarifying a multitierand hierarchical ARAN architecture from multiple perspec-tives. The survey framework and major topic organization areillustrated in Fig. 2. In brief, our survey discovered: • Network design:
An extensive review of similar accessinfrastructure designs was conducted. To provide a conve-nient reference for readers with different backgrounds andinterests, we adopt a logical narrative structure, initially
TABLE IS
UMMARY OF EXISTING SURVEYS ON AERIAL WIRELESS COMMUNICATIONS AND
RAN S . Reference Main theme Contributions5G/6G RANs [11] (cid:88)
A discussion of emerging applications and key technologies in 6G wireless systems[12] (cid:88)
Speculation about a set of 6G use cases and technologies[13] (cid:88)
Opportunities and challenges of providing Internet connectivity for rural areas in 6G[14], [15] (cid:88)
Roles and applications of AI for 6G wireless systems[17]–[20] (cid:88)
White papers from the Finnish 6G Flagship program on various research aspects of 6G[23] (cid:88)
Radio resource management in 5G RANs[24] (cid:88)
Separation of data and control planes in 5G RANs[25], [27] (cid:88)
Various aspects of CRAN, including network architectures, key techniques, and radio resource allocation[29], [30] (cid:88)
Integration of edge computing and RANs, Fog-RAN in [29], and MEC-RAN in [30][32]–[36] (cid:88)
Technologies and architectures for B5G and 6G RANs such as mmWave RANs [32], THz communications [33],blockchain RANs in [35], and AI RANs in [36][37]–[41] (cid:88)
Aerial communications in 5G/B5G/6G wireless systems such as satellite communications in [37], UAV communicationsin [40], and airborne communications in [41]Ours (cid:88) (cid:88)
A survey on aerial RANs toward a comprehensive 6G access infrastructure providing a high-level view of ARAN contribution to 6Gaccess infrastructure. Here, the positions, roles, and rela-tions of ARANs are investigated. Afterward, we discusstechnical ARAN characteristics regarding multitier andhierarchical system architecture as well as its standardreference model. From these analyses, the fundamentalfeatures that distinguish ARANs from other RANs arederived. Details are provided in Section II. • System model:
We analyze four key technical aspects ofARANs including transmission propagation, energy con-sumption, communication latency, and system mobility.In particular, transmission propagation modeling dealswith path loss formulation in a 3-D wireless environment.As energy efficiency is essential to maintain sustainableflight operations, we analyze energy consumption inbehaviors such as data transmission, data computation,an airborne object’s motion, and hovering. In addition,multihop communication latency and on-the-move datacomputation time are jointly investigated to identify po-tential end-to-end service delay in ARAN model. Finally,we study system mobility and trajectory scheduling forARANs, which is expected to be helpful for futureresearch regarding learning algorithms for system opti-mization. These details are covered in Section III. • Enabling technologies:
Regarding ARAN realization, wereview technologies that could feasibly support ARANimplementations on three planes: energy replenishment,operational management, and data delivery. In an ARAN,energy replenishment is a critical first step toward systemsustainability. The flight components must have either aself-recharging or a wireless recharging capability besidesthe basic charging station method. Accordingly, thesetypical approaches to charging are discussed. On the op-erational management plane, we focus on recent advancesin foundational pillars including network softwarization,mobile cloudization, and data mining technologies sup-porting ARANs. As the airborne components in ARANsinterconnect via different wireless spectra on the frontand back-end interfaces, enabling technologies for data delivery on these segments have distinguishable features.The frequency spectrum, communication protocol, andmultiaccess schemes especially are within the scope ofour survey. Details are provided in Section IV. • Application scenarios:
We envision emerging applicationscenarios that effectively exploit ARAN infrastructuresin the context of 6G. The applications are classifiedinto three categories, including event-based, scheduled,and permanent communications. Here, we clarify therequirements of these applications and describe how theadvances of ARANs can support them well. In addition,we note the common problems inherent in these applica-tion scenarios that may attract research and interest fromspecialist industrial communities. Details can be found inSection V. • Research challenges:
To drive potential research on 6GARANs, we highlight essential challenges for the suc-cessful development and implementation of ARANs. Asa native component of the comprehensive 6G accessinfrastructure, ARANs face challenges in terms of intelli-gent radio design, extremely high spectrum exploitation,and network stability to enable emerging technologiesand applications in 6G networks. Moreover, security andprivacy issues are considered as critical issues to protectuser data and system operations. Finally, reputable andefficient evaluation tools are needed to validate proposalsfor ARANs. Details are discussed in Section VI.In summary, the main contributions of this paper are asfollows. The survey serves as a reference framework for inter-ested readers, providing state-of-the-art knowledge and studiesregarding the emerging ARAN model. Technical foundationsof ARANs are taxonomized systematically from differentperspectives such as network design, system models, enablingtechnologies, and application scenarios. Further, an appropri-ate lesson learned is derived from related work analyses toconclude each perspective survey. Finally, future challengesare highlighted to illustrate research trends in ARANs towarda comprehensive 6G access infrastructure. A flowchart of ourresearch methodology is shown in Fig. 2. Common acronyms
TABLE IIL
IST OF ABBREVIATIONS
Abbreviation Description used in this survey are summarized in Table II.II. N
ETWORK D ESIGN
In this section, we position ARANs in a comprehensive 6Gaccess infrastructure and analyze them from the perspectivesof system architecture and reference model.
A. ARANs in 6G Access Infrastructure
In the context of comprehensive 6G access infrastructure,ARANs are positioned in the aerial communication layer toserve high-altitude and terrestrial users. As illustrated in Fig. 1,ARANs encompass three systems including LAP communica-tions at the altitude of 0–10 km, HAP communications at thealtitude of 20–50 km, and LEO communications at the altitude of 500–1500 km above relative sea-level [55], [56]. Comparedto terrestrial and underground(water) RANs, ARANs serve alarge coverage space with highly dynamic and mobile usersfor in-flight infotainment, aerial surveillance, flying vehicularcontrol, and isolated populations.The literature review reveals several aerial communicationclasses that are closely related to ARANs. The terms relatedto aerial communications that support mobile networks aresummarized in Table III. Briefly, these are drone/UAV assistedradio access networks (DA-RANs) focusing on LAP commu-nications [42]–[44], flying radio access networks (FlyRANs)aiming at LAP/HAP communications [45]–[48], and satellite-terrestrial integrated networks (STINs) involving satellite com-munications [51]–[54]. Because each of the similar termsrepresents a partial ARAN tier that is used for special appli-cations, the interconnection among these systems is weak andasynchronous. Hence, the introduction of an ARAN definitionis a significant contribution toward unifying the aforemen-tioned systems into a common model.
B. System Architecture
Figure 3 shows a wide perspective to demonstrate a detailedARAN architecture in the context of a complete user-corepath. Typically, an ARAN architecture comprises ( i ) a mainsegment that is a cross-tier networking infrastructure sharedamong ABSs at the LAP, HAP, and LEO altitudes, ( ii ) a front-end interface providing terrestrial and aerial access points thatgather user connections, and ( iii ) a back-end interface bridgingARAN infrastructure to the terrestrial core networks. Note thatLEO communication systems use satellite links to contact theterrestrial network through ground stations, while LAP andHAP systems use mobile (wireless) links to contact ABSs andthe terrestrial base stations (e.g., gNBs and eNBs) directly.In the main segment of an ARAN, LAP systems are atthe lowest tier. Drones and UAVs in LAP systems act asABSs, providing connectivity directly to aerial and terrestrialend users or remote base stations using wireless technologiessuch as 5G new radio (NR) and Wi-Fi. Conversely, theseABSs may utilize either satellite technology to connect withLEO communication systems or wireless technologies to theterrestrial macro base stations for backhaul transmission tothe core networks. LAPs can be classified by size, range,speed, and endurance. According to the US Department ofthe Army [57], LAPs are classified into five categories, i.e.,small < medium < large < larger < largest groups. MostLAPs are relatively lightweight and cost-effective devices thatcan be deployed quickly and flexibly. However, they haverelatively low endurance with limited energy and networkingresources. To mitigate these issues, tethered technology canbe considered a feasible approach to either establish a reliablebroadband backhaul through a fiber cable or energize LAPsby a powerline connection between the LAPs and terrestrialstations [58]–[60]. Nevertheless, LAPs effectively supporttime-sensitive and event-based scenarios such as emergencyand rescue, aerial surveillance, traffic offloading, and mobilehotspots at public gatherings.In the middle tier of the ARAN main segment, HAPsare defined by the ITU Radio Regulations (RR) as radio TABLE IIIR
ELATED TERMS REGARDING AERIAL COMMUNICATION ASSISTED MOBILE NETWORKS . Name Access infrastructure Definition ApplicationsLAP HAP LEO
DA-RAN (cid:88)
DA-RAN stands for a drone/UAV assisted radio access network, where drones help toextend the coverage area and capacity of terrestrial access infrastructure. In this context,drones can connect either directly to terrestrial base stations or via a head node in(multihop) mesh, star, tree, and chain topologies. Unlicensed/licensed spectrum can beexploited by drones in DA-RANs [42]–[44]. Aerial surveillance,smart farm, smartcity, environmentstudies, temporarycapacity boost, etc.FlyRAN (cid:88) (cid:88)
FlyRAN stands for flying radio access network, where aircraft and airships equippedwith radio transceivers are utilized as ABSs to provide mobile services in underservedareas. In FlyRAN, the LAP tier exploits the unlicensed/licensed spectrum, while the HAPtier typically exploits the licensed spectrum to avoid unmanaged conflicts with terrestrialsystems [45]–[48]. The term FANET is covered by this definition [49], [50]. Remote populations,rescue & surveillance,environment studies,temporary capacityboost, etc.STIN (cid:88)
STIN represents a satellite-terrestrial integrated network by which satellite communica-tions supplement the terrestrial systems to offer global seamless and ubiquitous Internetservices to users in isolated areas. STIN utilizes the licensed spectrum for ground-airconnections. The airborne infrastructure includes multiple tiers such as LEO, mediumEarth orbit (MEO), and geostationary Earth orbit (GEO) satellite constellations [51]–[54]. Isolated populations,emergency & alarm,global roaming, large-scale Earth studies,etc.ARAN (cid:88) (cid:88) (cid:88)
Referred to the definition in Section I-A, ARAN is a multitier and hierarchical aerialaccess infrastructure combining the FlyRAN and LEO communication systems of STINto provide radio access medium from the sky to end users for Internet services throughABSs equipped with heterogeneous wireless transceivers. The ABSs can be UAVs, drones,balloons, and airplanes. The definition of ARAN was first introduced by Song et al. [55]with the original name AAN (aerial access network) at the IEEE International Conferenceon Communications (ICC) in June 2020. Smart farm/city,temporary capacityboost, underservedpopulations, rescue& aerial surveillance,emergency & alarm,scalable environmentstudies, etc.
ComSatCubeSatUAV Inflight servicesAerial servicesRemote IoT and user servicesRemotebase stationGround stationAerial base stationCore network Ground station Macrobase station Aircraft UserMain segmentLEO layer 2LEO layer 1HAPLAPSatellite linkMobile link Front-end interfaceBack-end interfaceCore Tethered link
Fig. 3. System architecture of ARANs. stations located at a specified, nominal, fixed point relativeto the Earth . 2, 6, 27/31, and 47/48 GHz frequency bandswere assigned for HAP communications in bidirectional HAP-terrestrial links [61] at three world radio communication con-ferences (WRC-97, WRC-2000, and WRC-12). HAP systemsserve aerial and remote terrestrial end users with wirelesstechnologies in a wide coverage area from high altitude.For backhaul transmission to the terrestrial core networks, HAP systems mostly utilize satellite technology via the LEOcommunication systems. In practical situations, some indus-tries have implemented trial projects using lightweight solar-powered aircraft and airships to provide stable broadbandservices in rural and remote areas [62].At the top of ARAN main segment, two-layer LEO commu-nication systems consist of miniaturized satellites below (i.e.,CubeSats) and communication satellites above (i.e., ComSats),orbiting at an altitude of 500–1500 km. CubeSats aim toprovide low-latency and high-throughput Internet services, while ComSats are designed for high coverage and serviceavailability. These two LEO layers interact with each otherthrough interlinks, which provide redundancy, backup, andcollaboration interfaces for dynamic network organization.Compared with other satellite classes, LEO satellites arecharacterized by lower latency, cost-efficiency, and quickproduction and deployment. Most LEO satellites operate onKu, Ka, and V bands to provide Internet connection toaerial and ground stations within tens of ms latency [63].Unlike the LAP/HAP systems, LEO communication systemstypically do not provide services to the end users directly.Both fronthaul to the end users and backhaul to the corenetworks are satellite transmissions through ground stationsand very small aperture terminals (VSATs). To orchestrateintertier networking operations in an ARAN, LEO communi-cation systems additionally support backhaul tunnels allowingLAP/HAP systems to connect to the core networks. Recentyears have witnessed several emerging commercial satelliteprojects on LEO communication systems such as OneWeb,Telesat, and Starlink, with hundreds of satellites successfullylaunched into orbit and thousands of satellite launches plannedfor the near future [64].In ARANs, the front-end interface includes ABSs at theLAP/HAP tiers, remote base stations, and ground stationsenabling end users network access the networks. The ABSsand remote base stations are ARAN access points that providewireless links directly to aerial and terrestrial users. Mean-while, the ground stations are end-points of satellite links fromLEO communication systems that deliver traffic to and fromthe remote base stations. Depending on prevailing circum-stances, the ABSs and remote base stations may cross-serveboth aerial and terrestrial services, as illustrated on the left sideof Fig. 3. Conversely, the back-end interface includes macrobase stations and ground stations to accommodate backhaultraffic toward the core networks. Typically, the macro basestations help transfer traffic flows from LAP systems, while theground stations support traffic forwarding from (HAP systemsto) LEO communication systems; see the right side of Fig. 3.
C. Reference Model
The ARAN reference model is built by jointly adoptingthe 3GPP TS 23.501 (version 16.5.0 Release 16) [65] andthe ETSI TR 103 611 (V1.1.1) standards [66] that werereleased recently by the 3rd Generation Partnership Project(3GPP) organization and the European TelecommunicationsStandards Institute (ETSI), respectively. Fig. 4 depicts theARAN reference model in detail. This reference model isbased on the 3GPP 5G standard, where ABSs perform thefunctions of the next-generation Node B (gNB), and the LEOcommunication tier is considered a trusted non-3GPP compo-nent [67]. It is worth noting that this model does not covernon-trusted LEO system interaction because LEO networksare considered critical components in ARAN architecture. Inparticular, mobile service management and control throughthe LEO networks must be handled by the mobile networks.For this purpose, only trusted LEO networks are consideredsatisfactory for network integration. In Fig. 4, the description of the 5G core components is abstracted significantly, exceptthe user plane function (UPF), which is the contact pointto access the core networks and other systems (i.e., intranetand Internet). Mapped on the system architecture illustrated inFig. 3, the components of the reference model are representedas follows: • User equipment (UE): inflight, aerial, remote IoT, andterrestrial user services. • ABS: LAPs and HAPs • VSAT: ground stations at the front-end interface • Gateway (GW): the ground stations at the back-endinterface • LEO systems: CubeSats, ComSats, and their controlsystems • gNB: remote and macro base stations.In the control plane, a common 5G network managementsystem is in charge of managing, controlling, and monitoringoperations of almost all the ARAN components such asABS, VSAT, GW, and gNB via C-interfaces of correspondingreference points. In the user plane, ABSs are consideredas 5G gNB-functional components. Therefore, the NR Uu reference points are utilized between ABSs and UEs, whilethe Xn reference points are defined for ABSs’ interactions andbetween ABSs and gNBs. As a standard 5G reference model,gNBs and UPFs are connected via N3 reference points. Adetailed description of the NR Uu , Xn , and N3 reference pointscan be found in the 3GPP TS 23.501 (version 16.5.0 Release16) [65] and the 3GPP TS 38.300 (version 16.2.0 Release16) [68] standards. Conversely, to standardize the translationbetween 3GPP messages and non-3GPP messages (i.e., IPpackets) on interfaces gNB/ABS–VSAT and GW–UPF, the Ymx and
Ygw reference points are defined in the ETSI TR103 611 (V1.1.1) standard [66].
D. Fundamental Features
The fundamental features that distinguish ARANs from theother RAN architectures and help to reinforce emerging 6Gservices are described as follows: • Ubiquity:
The LEO communication tier guarantees ser-vice continuity across the globe with three advantages,including wide coverage, networking backup/resilience,and emergency broadcast, for which the terrestrial infras-tructure has limited capacity. • Mobility:
The dynamicity of aerial LAP/HAP topol-ogy implementation and the networking overlay amongthe LAP, HAP, and LEO communication systems helpARANs to adapt flexibly to the requirements of end usersanywhere on the ground and in the air. • Availability:
Operating in the air at various altitudes,ARANs are not commonly affected by natural and man-made disasters capable of rendering terrestrial communi-cation infrastructures vulnerable and interrupting service.In other words, better service availability is providedregardless of the recipient’s terrain (e.g., mountain, sea,desert, etc.). • Simultaneity:
Multitier LAP/HAP/LEO communica-tion systems can adaptively self-organize to forward gNBVSATLEO systemsGWUPFOther network functionsData packet networks UENetwork management systemsCore network LEO tiergNB LAP/HAP tiers
NRUuYmxYgw XnXnN3
U-planeC-planeABSABS ABS
NRUuXnXn Ymx
C-interfaceU-interface Xn Fig. 4. Reference model of ARAN adopting the 3GPP 5G standards. information-centric services effectively across discrete(aerial and terrestrial) locations on simultaneous multi-cast and broadcast streams using various wireless accesstechnologies. • Scalability:
Because ABSs interlink to each other us-ing aerial wireless ad hoc technologies and there arehierarchical networking overlays among LAP, HAP, andLEO communication tiers, ARANs can quickly establishscalable topologies for local sites without service inter-ruptions. III. S
YSTEM M ODEL
In this section, we discuss four key aspects of ARANs:transmission propagation, energy consumption, latency analy-sis, and system mobility.
A. Transmission Propagation
A key challenge in aerial communications is accuratelymodeling radio propagation. Indeed, channel modeling is nec-essary for designing correct waveforms, resource allocation,modulation order, multiantenna techniques, and interferencemanagement [69]. It is widely known that channel characteri-zation for cellular networks has been modeled and empiricallyverified. Propagation channels for satellite communicationsat various frequency bands (i.e., Ku, Ka, and V) were alsowell-studied [70]. Moreover, many HAP projects for wirelesscommunications were undertaken in the 1990s and 2000s. Forinstance, the theoretical model of small-scale fading for HAPpropagation channels was investigated in [71], and the appli-cation of HAPs for broadband communications, e.g., channelmodeling, interference, coding techniques, and resource allo-cation was presented in [72]. Owing to recent technologicaladvancements, investment in the industry, and the involvementof major tech corporations like Google and Facebook, asignificant amount of research has been conducted to integrateHAPs into B5G wireless systems. As the characterization ofHAP and satellite communications channels has been wellestablished, in this subsection we focus on discussing channelmodeling for LAP-enabled communications.Depending on the channel model, there are two mainmodeling approaches, including empirical channel models formulated on field measurements, and analytical models thatanalyze the transmission channel under certain conditionsand/or assumptions. Measurement campaigns play an impor-tant role in developing empirical channel models. Variousmeasurement campaigns with various environmental and mea-surement settings have been conducted, e.g., the configurationof antenna arrays as either single-input single-output (SISO) ormultiple-input multiple-output (MIMO), channel sounding fornarrow-band and wide-band channels, propagation scenarios(e.g., open space and mountains), and evaluation of eleva-tion angles [69]. Another modeling classification based onpropagation environments includes three approaches [41]: adeterministic model, stochastic model, and geometric-basedstochastic model. In particular, the first approach relies onthe assumption of certain network layouts and is thus suitablefor large-scale fading effects, the second approach takes intoaccount multipath components, and the third approach is tostudy spatio-temporal channel characteristics in 3-D stochasticenvironments.
1) Deterministic Models:
Deterministic channel models arecharacterized by information on the propagation environmentsuch as terrain topography and the composition of buildings orobstacles. Moreover, ray-tracing software is usually employedin the literature to analyze deterministic channel models. Forexample, Feng et al. [73] investigated a statistical model forevaluating the path-loss and shadowing of air-to-ground chan-nels in urban areas. The proposed model is analyzed by ray-tracing simulations over different frequency bands, including200 MHz, 1 GHz, 2 GHz, 2.5 GHz, and 5 GHz. Unlike thelog-distance channel models of terrestrial communications, thechannel model proposed in [73] is dependent not only onthe distance from the ABS (i.e., UAV in LAP systems) butalso on the elevation angle. To generalize the channel modelin [73] to multiple urban environments, Al-Hourani et al. [74]proposed a generic model to estimate the path-loss of air-to-ground channels and facilitate radio frequency (RF) planning.In particular, the path loss, η , for a given ground user iscomputed as η = PL − FSPL , (1)where PL and FSPL are the total power loss and free spacepath loss, respectively. Here FSPL can be calculated from the Angle of depression T e rr e s t r i a l B S ’ s h e i gh t A B S ’ s a ltit ud e G r o u n d - t o - a i r p a t h l o s s m o d e l Fig. 5. Illustration of an experimental setting for ground-to-air path lossmodels.
Friis transmission equation asFSPL = 20 log( d ) + 20 log( f ) − . , (2)where d refers to the distance in 3D space between the ABSand ground user, and f denotes the carrier frequency (inMHz). Bor-Yaliniz et al. [75] proposed a path loss modelfor modern metropolitan scenarios, which may include manyskyscrapers with different heights, and ray-tracing simulationswere conducted to evaluate the performance of the proposedmodel. A ground-to-air channel model for suburban environ-ments was proposed in [76]. In particular, the total path losswas composed of two components, including terrestrial pathloss and aerial excess path loss, which is heavily dependenton the depression angle, as illustrated in Fig. 5. Anotherchannel model for suburban areas can be found in [77]. Themeasurement was carried out at a radio frequency of 3.9 GHz,and the path loss was calculated by a two-ray model.
2) Stochastic Models:
In stochastic channel models, thefading characteristics of multipaths are taken into consider-ation. In the literature, the performance of these stochasticmodels have been evaluated by empirical measurements orby geometric analyses via computer simulations. Comparedto the deterministic model, the accuracy of a stochastic modeldepends on environmental information and the nonstationarypropagation channel. In the multipart papers [78]–[81], dif-ferent air-to-ground channels for LAP communications wereproposed. In [78], a wideband statistical channel model overtwo frequency bands (970 MHz and 5 GHz) was proposed foroff-shore scenarios. For other terrain types, channel modelsfor hilly and mountainous environments can be found in [79],and channel models for suburban and peri-urban environmentscan be found in [80]. In the fourth part [81], shadowingloss, shadowing duration, and small-scale fading statistics wereanalyzed. By using empirical measurement data, it was shownthat shadowing loss can be modeled as a function of anaircraft’s roll angle, but it does not correlate with shadowingduration. For airport environments, Rieth et al. [82] conductedtwo measurement campaigns to characterize small-scale results(e.g., delay and Doppler spread statistics) and large-scaleeffects (shadowing and antenna misalignment) of air-to-groundradio channels. Several interesting results are found by thiswork, e.g., max and min delay resolutions of 2184 and 52 ns, respectively, which correspond to the respective coherencebandwidth of 92 and 3846 kHz (i.e., coherence bandwidth= 1/(5 × delay resolution)). Such findings are necessary forselecting payload symbols, especially, in frequency-selectivefading LAP systems. A recent channel model proposed in [83]was dedicated to residential and mountainous desert scenar-ios. By comparing statistical channel models for these twoenvironments, it was found that the fading effects observed inresidential settings were less than those observed in mountain-ous desert settings. This is reasonable because LoS links aretypically available in residential, and the signals reflected fromthe ground can be obstructed well by residential buildings.
3) Geometric-based Stochastic Models:
Compared to thetwo above modeling approaches, a geometric-based stochasticmodel is suitable for considering spatio-temporal channelcharacteristics and for deriving analytical performance ex-pressions such as coverage radius and channel capacity. Forexample, Cheng et al. [84] investigated a 3-D geometric-based stochastic model for MIMO nonstationary propagationchannels. Particularly, the nonstationarity of LAP communi-cations was overcome by integrating time-varying angles ofarrival and departure into the reference model and by consid-ering both line-of-sight and non-line-of-sight components. Aninteresting observation from this work is that ABS-specificparameters (e.g., moving direction and antenna setup) andaltitude significantly affect channel stability, demanding moreefficient control strategies for the ABS trajectory to render theproposed channel model more stable and less nonstationary.Recently, Jiang et al. [85] considered the presence of interfer-ing objects in designing air-to-ground channels and developeda computationally-efficient method to estimate the angularparameters such as azimuth/elevation angles of departure andarrival. Unlike [84], [85], the work in [86] does not fix themoving speed of the ABS, thus complicating the channelmodel by the directions of transceivers’ motion.
B. Energy Consumption
Energy-efficient communications have been one of the mostimportant design requirements in B5G wireless systems. It wasrecently reported in [87] that the use of drones for packagedelivery is relatively energy-efficient compared with ground-based delivery provided that the warehouse system is designedproperly and the drones are used within their enduranceand ferry ranges. Despite many potentials and applications,airborne objects, especially those working in the LAP tiers,are typically powered by onboard batteries that usually havelimited capacity and lifetime. Several promising technologieshave been developed to mitigate the impact of limited energyand battery lifetimes such as wireless power transfer andenergy harvesting (EH). However, the implementation of thesetechnologies to massive airborne objects is in practice stillquestionable and will not be available in the foreseeable future.Owing to the importance of energy efficiency in maintainingsustained flying operations for ARANs, the modeling of en-ergy consumption during flight is a critical challenge.Various factors must be considered when developing energyconsumption models for aerial communications. The obvious examples are UAV types, flying speed, acceleration, payload,and external factors such as weather conditions. The firstfactor is reasonable because UAVs usually travel in highlydynamic environments (e.g., weather and wind conditions)that significantly affect flight capabilities. These factors mayenhance system performance in some cases. For instance,when a UAV is flying with the wind, it can fly faster butconsumes less energy. Moreover, external temperatures canalso directly affect performance by causing battery lifetime anddrain issues. As reported in [88, Fig. 4], UAV consume lessenergy when they fly with a tailwind, but they consume moreenergy when flying into a headwind. Moreover, within a windspeed threshold, the higher the wind speed, the more energy-efficient the UAV operation. Second, the energy consumptionof a UAV significantly depends on its flying state, e.g., hover-ing, horizontal, and vertical motion [89]. Finally, as the weightof a UAV increases according to the payload, the energyconsumption also increases as a function of the payload. Asempirically tested in [88, Fig. 3] when a UAV is hovering,the power consumption increases linearly with increments inthe payload. Because various factors may affect the energyconsumption of aerial systems, it is very challenging to modelenergy consumption for all environmental conditions.
1) Energy Consumption in Hover and Vertical Movement:
To model the power consumed during hovering for deliveryservices, Dorling et al. [90] considered modeling the powerconsumption of a multirotor helicopter as a function of its totalweight, which includes the frame weight, battery, and payload.The power, P , required for single-rotor copters to hover wasmathematically established in [91] as P = T / √ σA , (3)where σ is the fluid density of the air, A is the area of the rotordisk, and T = ( M + m ) g is the rotor thrust where M is theframe weight, m is the battery and payload weight, and g isthe gravitational acceleration. Based on this model, Dorling etal. [90] proposed a new power-consumption calculation formultirotor copters on the assumption that each rotor bears anequal amount of frame, battery, and payload weight. Thus, thepower consumed by a rotor is modeled as P = ( MN + mN ) / g / √ σA , (4)where N is the total number of rotors. As a result, the totalpower consumed by the multirotor copter can be given as P = N (cid:18) MN + mN (cid:19) / g / √ σA = ( M + m ) / (cid:114) g σAN . (5)To simplify the model further (5), Dorling et al. [90] con-sidered a linear approximation composed of two parts: onerepresents the power consumed to bear the battery and payloadweight and another accounts for the power needed for thevehicle to hover. These models are also used to calculatepower consumption during takeoff and landing, that is, thepower consumed during takeoff and landing is set to beequivalent to that of hovering [89]. Many models have been Moving direction ThrustLiftDrag Weight
Fig. 6. Four forces occur in UAVs. proposed to consider other factors, for example, motor andpropeller efficiencies were integrated in [92], and multipledynamics of UAVs (e.g., electrical dynamics of the battery andaerodynamics of the rotor-propeller assembly) were studiedin [93].
2) Energy Consumption in Horizontal Movement:
The fourprincipal forces lift, weight, thrust, and drag should be consid-ered for a drone’s flight. Fundamentally, drag consists of lift-induced and parasitic drags and power is needed to overcomethem. The former is directly prosegmental to the air density, σ , the square of the drone velocity, V , and the square of thespan loading (i.e., the ratio between the lift and wingspan inN/m) as [94] D i = κ ( L/b ) / (cid:18) σπV (cid:19) , (6)where κ is a non-dimensional coefficient, L is the lift gener-ated by the drone, b is the wingspan, and V is the forwardvelocity. Because air density decreases at higher altitudes,more lift-induced drag is generated, and accordingly, morepower and a greater wingspan are required for the drone tooperate. Another form of drag is parasitic drag, which can beestimated as [94] D p = C D SσV / , (7)where C D is the parasitic drag coefficient, and S is the wingarea. Thus, the respective power required to overcome lift-induced and parasitic drags can be given as P i = D i V = κL / (cid:18) σπV b (cid:19) , (8) P p = D p V = C D SσV / . (9)As a result, the total power needed for the drone to stay afloatis calculated as P = P i + P p = κL σπV b + C D SσV . (10)The power consumption model (10) has been employed byZeng and Zhang in [95] to optimize UAV flight trajectorysubject to various design constraints imposed by the initialand final locations as well as maximum and minimum velocityand acceleration values. This work has played a seminal role inresearch on UAV wireless communications. We note that the power consumption models above are for fixed-wing UAVs,while those of rotary-wing UAVs were investigated in [96].For more details, we invite interested readers to refer to thiswork and the references cited therein. C. Latency Analysis
Along with propagation models and energy consumption,communication latency is another key feature of aerial com-munications requiring a thorough study.
1) Communication Latency:
In addition to on-demandcommunications, aerial communications have also been im-plemented to support low-latency services. An example ofthis is the use of ABSs as aerial caching servers. In thisscenario, ground users can directly request content from anaerial caching server instead of sending the request to aremote server, thus reducing latency and avoiding a networkbottleneck. Conversely, achieving low latency is also importantin aerial communications. Along with high reliability andhigh security, low latency is a critical requirement for controland non-payload communication (CNPC) links for supportingthe management of LAP/HAP systems and avoiding crashes.Since LEO satellites typically have lower altitude orbits com-pared to GEO and MEO, LEO communication is highlyattractive to the industry owing to its ability to provide latencyon the order of tens of ms. The OneWeb system can provideInternet services of 400 Mbps with an average latency of32 ms, while the Telesat system promises to have latencybetween 30 and 50 ms, and the Starlink system can offerbroadband Internet with a latency from 15 to 35 ms, asreported in [63]. These aerial platforms (i.e., LAP, HAP, andLEO) should be integrated into ARANs to provide Internetservices with different latency ranges. While low-tier LAPsystems aim to provide for event-based scenarios and low-latency services, high-tier LEO communication systems aredesigned to provide Internet access across the globe. In themiddle tier of the ARAN model, HAP systems maintain abalance between service availability, latency, and deploymentcost.In general, end-to-end latency can be approximated as twicethe sum of radio, backhaul, core, and transport latencies. Radiolatency is composed of queuing latency, frame alignmentdelay, transmission latency, and processing latency. In long-distance transmission, LEO communications take advantageof the speed of light in space, whereas in optical cables, thespeed of light is relatively lower, typically 180,000 to 200,000km/s. Using queuing theory, Horani et al. [97] showed thatthe use of ABSs can reduce the queuing delay compared tothat of terrestrial communications and that an ABS’s altitudesignificantly affects latency by contributing to the likelihoodof line-of-sight communication links. An analysis of latencyin uplink mmWave-caching networks was provided in [98],on the assumption that queuing latency accounts for themajority of total latency. Based on theoretical analyses, it wasshown that end-to-end latency is reduced when more ABSsare deployed in the network, while it increases when userdensity increases. Wu et al. [99] leveraged the concept of“physical layer security” to guarantee low latency for secure content sharing services in aerial communications. To achievethis objective, the ABSs’ trajectories and user associations areoptimized jointly via an iterative algorithm. The aforemen-tioned studies agreed that lower latency could be achievedby an appropriate deployment of ABSs compared to baselineschemes with fixed deployment.To improve the performance of CNPC links, several studieshave integrated ultra-reliable and low-latency communications(URLLC) into LAP/HAP systems. She et al. [100] proposedimproved distributed multiantenna systems to maximize hor-izontal communication distance (a.k.a. available range) toguarantee the end-to-end delay and total packet loss proba-bility in URLLC-based LAP systems. This work emphasizedthat the number of antennas at each access point should beoptimized to maximize the available range of the ABS. Ren etal. [101] characterized the achievable data transmission ratesof CNPC links. However, the analysis is complicated by the3-D deployment of the aerial platforms and the design ofshort-packet transmission in URLLC. A notable observationis that the CNPC’s achievable link rate increases as blocklength increases, but this comes at the cost of greater latency.This work also indicated that leveraging the Shannon formuladirectly to compute the achievable rate is somewhat inefficient,especially when the block length is small. Another study onURLLC-based LAP communication was conducted in [102],where an ABS was deployed to transfer URLLC packetsamong IoT devices and the block length is optimized.
2) Computation Latency:
Considered as a key technologyin B5G, edge computing (e.g., fog computing and multiaccessedge computing (MEC)) has recently received significantattention from both academia and industry. Edge comput-ing markedly enables the emergence of delay-sensitive andcomputationally intensive applications (e.g., human activityrecognition and autonomous driving) as computing resourcesare moved from the cloud to the network edges closest toend users [103]. MEC nodes can be deployed at variousRAN locations, including both fixed points (e.g., macro basestations and WiFi access points) and mobile points (e.g.,moving vehicles and smart phones) [104]. Corresponding totwo types of aerial communications, there are also two MECdeployment scenarios, as shown in Fig. 7. The first scenarioconsiders aerial computing servers processing computationaltasks migrated from ground users, whereas the second scenarioconsiders aerial users, who may offload their computationaltasks to terrestrial computing servers for remote processing.Different latency sources in MEC systems may include com-puting, transmission, queuing, and backhaul latencies [105],[106].To minimize the weighted sum energy of the ABS and allthe ground users under constraints on completion time, Hu etal. [107] considered an MEC system to process computationaltasks of energy-limited devices and devised an iterative al-gorithm. This work reported a trade-off between energy con-sumption and task completion time, i.e., the network consumesless energy as the completion time increases and vice versa.In [108], an aerial relay was deployed to collect and thenforward data generated by IoT devices to an MEC serverfor remote processing. In this case, the challenges caused by UAV-assistedMEC(a) UAV-assisted edge computingUEs C o m pu t a ti ono ff l o a d i ng Aerial users C o m pu t a ti ono ff l o a d i ng Obstacle (e.g., high-rise building and mountain)(b) Edge computing assisted UAV communicationsABSMacro base station equipped with MEC
Fig. 7. Illustration of two UAV-MEC scenarios. limited computing capability at the MEC server and unstabletransmission links were addressed jointly to enhance systemendurance and task completion time. For IoT service provi-sioning in aerial communications, the work in [109] proposedthat UAVs can serve as both computing servers and relaysto minimize the average latency of all the IoT devices. Thesimulation results of this work revealed that completion time isreduced when the number of UAVs is increased, and the dualuse of both aerial computing and aerial relaying significantlydecreases the completion time compared to a case in whichUAVs are used only for relaying purposes. In [110], integratingMEC into LEO communications was shown to be a promisingsolution to provide computing services for various scenarios,especially when ground users are sparsely distributed overdifferent regions. An illustrative example of a satellite MECsystem is shown in Fig. 7, where UAVs can connect to MECservers located at ground station via satellite connections. Insuch scenarios, cooperative computation offloading problemscan be optimized to reduce both energy consumption andlatency.
D. Mobility Analysis
While terrestrial infrastructures are constructed at prede-fined locations or follow constrained investment policies,ARANs can be designed for dynamic deployment to optimizebenefits to infrastructure providers as well as to efficientlyimprove QoS for end users. Many studies have shown that thedynamic mobility and trajectory control of ARANs would of-fer significant advantages such as improving energy efficiency,supporting data collection applications, and powering energy-limited IoT devices. We explicate the benefits of and state-of-the-art studies pertaining to the optimization of mobility andtrajectory in aerial systems as follows.
1) System Mobility:
Aerial communications play an impor-tant role in the IoT, typically consisting of a large numberof small, energy- and computation-limited devices. To reducethe power consumption of an IoT network and satisfy QoSrequirements of IoT devices, in [111], multiple UAVs weredeployed for IoT data collection. Since the locations of IoT devices vary over time and an IoT device may be underthe service coverage of multiple UAVs, the joint problem ofUAV deployment, IoT device association, and power controlis extremely challenging. Simulation results demonstrated thatmobile deployment of UAVs can reduce the average totaltransmission power by 45% compared to that of the stationaryABSs. As analyzed by Chetlur et al. [112], increasing thenumber of UAVs can increase network spectral efficiency;however, this is reduced when the number of deployed UAVsis sufficiently large. Therefore, optimizing UAV placement isa critical problem and has been investigated in many researchworks. For example, Lyu et al. [113] considered the UAVplacement problem to guarantee that each ground user isserved by at least one ABS. In aerial relay communications,one significant challenge in optimizing UAV placement is thatshadowing on the receiver side depends on the environment.Chen et al. [114] addressed this dependence by developinga blockage-adaptive algorithm, which they found to be moresuitable for propagation models than a statistical channelmodel for all environments in existing studies.Much research has focused on analyzing the performanceof LAP/HAP systems with UAV mobility, e.g., coverageprobability, secrecy rate, and outage probability. In [115], aset of ABSs was deployed to serve a ground user, and thecoverage performance was analyzed in two scenarios. Thefirst considers that the user randomly associates with oneamong multiple ABSs, whereas the user associates with theclosest ABS in the second scenario. The use of UAVs forrelaying satellite signals to ground users was considered [116].In particular, aerial relays are to be deployed in the case inwhich direct communications links between the satellite andground users are not well established (e.g., indoor users). Thisstudy reported that the deployment of 3-D aerial relays couldachieve better performance in terms of outage probability.In [117], the secrecy capacity and secrecy outage probabilityof a hybrid satellite-terrestrial network with 3-D aerial relayswas analyzed. These metrics are necessary if eavesdroppersexist in the network. To generalize the secrecy performance,this work considered three criteria for selecting aerial relay: closest, random, and maximum signal-to-noise ratio.In the high tier of ARANs, the design of satellite con-stellations is important to ensure global coverage, that is,anywhere on Earth being covered by at least one satellite.Generally, constellation types and designs are dependent onseveral factors such as coverage requirements, propagationlatency, launch cost, and designated missions. The reasonfor the first factor is that a GEO satellite has much greatercoverage than a LEO satellite, and thus multiple LEO satellitesare required to cover the same region of interest. Propagationlatency is determined mainly by the satellites’ orbiting altitudeand can be quite pronounced for satellite constellations. Whenthe respective orbiting altitudes of LEO and GEO are set as900 and 36,000 km, the propagation latency is 3 and 120 ms,respectively (the speed of light in space is × m/s). Launchcost, the third factor listed above, is related to the number ofsatellites to be deployed. As aforementioned, LEO satellitesare attractive to the industry owing to their lower costs and de-ployment complexity compared with GEO and MEO satellites.Finally, the design of satellite constellations should be basedon the designated mission, and there are multiple constellationdesigns for specific missions. These may include circularorbit constellations, elliptical orbit constellations, flower con-stellations, and Walker constellations. Several studies haveinvestigated the design of satellite constellations from a com-munication perspective. For example, two classes (with andwithout inter-satellite links) of LEO constellation designs tosupport IoT applications were reviewed in [118]. The formeris suitable for delay-tolerant IoT applications, while the latterprovides seamless connectivity. Papa et al. [119] integratedsoftware-defined networking (SDN) into LEO communicationsystems to separate the control plane from the data plane insatellite communications, and they further proposed a three-tier SDN-based LEO architecture in which SDN controllers areco-located on satellites. A technical comparison among threeLEO satellite constellation systems (i.e., SpaceX, Telesat, andOneWeb) in terms of system throughput and satellite efficiencywas conducted in [120].
2) Trajectory Schedule:
Besides mobility, the trajectory ofABSs can be optimized for various purposes. For instance, incollecting IoT data in smart cities and agricultural scenarios,the deployment of ABSs can assure line-of-sight communica-tion links for mobile users in emergency rescue operations,thus increasing data collection efficiency and saving more ofthe energy required by mobile devices to transmit data. In suchcases, UAVs can be regarded as mobile hubs for data collectionand processing. Another scenario may include situations whereUAVs are deployed as energy sources to wirelessly power IoTdevices. Recently, significant research efforts in the field ofwireless communications have been dedicated to optimizingUAV trajectories.UAV trajectory optimization was investigated in the seminalwork [95]. After developing an energy consumption model,Zeng et al. considered two UAV trajectory scenarios: un-constrained and constrained. An interesting result for theunconstrained case is that the network was energy inefficientregardless of the design objective, thus demanding trajectoryschemes with practical constraints such as initial and final
Initial location Final location
Angle of elevation(Rician fading)IoT devices
Fig. 8. Illustration of UAV-enabled data harvesting. locations, minimum and maximum velocity, and permittedaltitude ranges. In contrast to [95], You et al. [121] consid-ered optimizing the joint UAV scheduling and 3-D trajectoryproblem in Rician fading channels, as illustrated in Fig. 8.Particularly, at each time instance, one sensor device can bescheduled to transmit its data to a UAV traveling between twopredefined locations to maximize the minimum data collectionrate. The use of UAVs for wirelessly powering IoT devices andimproving the freshness of IoT data was investigated in [122].To address the non-convexity of the problem, a decompositiontechnique was applied, and a joint dynamic programmingmetaheuristic approach was adopted to solve the trajectorysub-problem. Numerical simulations in this work demonstratedthat data freshness increases linearly with increments in UAVs’altitudes and the size of collected data. The application of AIfor optimizing a UAV trajectory has been considered recentlyin various research studies. For instance, echo state networksand multiagent Q-learning were adopted to calculate effectiveUAVs’ trajectories and predict ground users’ locations [123],and a deep reinforcement learning approach was proposedto select UAVs’ trajectories in UAV-enabled edge computingsystems [124].
E. Summary and Discussion
This section presents four key theoretical aspects of anARAN architecture, including transmission propagation, en-ergy consumption, latency analysis, and system mobility. InTable IV, we summarize key points of the ARAN systemmodel as well as representative references and their contribu-tions. In particular, three transmission propagation approachesfor ARANs (i.e., deterministic, stochastic, and geometric-based stochastic models) are reviewed in Section III-A. Weobserve that existing studies mainly focus on channel mod-elling in sub-6G GHz frequency bands, while almost ignoringconsideration of THz and non-RF models (e.g., visible light,neural, and molecular links). As the first attempt, the THzchannel model is investigated in [125] assuming that theplacement and power allocation of UAV are jointly optimizedto minimize the total latency between the UAVs and ground TABLE IVS
UMMARY OF
ARAN
REFERENCES ON SYSTEM MODEL . Aspect Sub-aspect Highlights Ref. Contributions
TransmissionPropagation DeterministicModels Deterministic models assume certain networklayouts such as terrain topography and obsta-cles. [74] The path-loss is comprised of power loss and freespace path loss. The latter is dependent on the carrierfrequency and transmission distance (urban areas).StochasticModels Stochastic models consider multipath fadingeffects. [82] Both small-scale and large-scale effects are consideredfor an air-to-ground channel. Respective max and mindelay resolutions are 2184 ns and 52 ns.Geometric-basedStochasticModels Geometric-based stochastic models considerspatio-temporal channel characteristics, and arethus suitable for deriving analytical perfor-mance metrics. [84] To cope with the non-stationarity of ABS in aerialcommunications, the channel model considers time-varying arrival and departure angles.EnergyConsumption Hover andVertical Moving Most of the existing studies model the totalpower in hover and vertical motion as a func-tion of the total weight, including the frameweight, battery, and payload. [90] The power consumption of a multirotor drone is P = ( M + m ) / (cid:115) g σAN , where M is the frame weight, m is the battery andpayload weight, A is the rotor disc area, N is thenumber of rotors, and g is the gravitational acceleration.HorizontalMoving Among the four principal forces (i.e., lift,weight, thrust, and drag), drag is mainly an-alyzed to compute the energy consumed byhorizontal motion, focusing on lift-induced dragand parasitic drag. [94] The power needed for the drone to remain aloft is P = 2 κL σπV b + C D SσV , where main parameters are the wingspan, b , velocity, V , air density, σ , and wing area, S .LatencyAnalysis CommunicationLatency • LEO communications typically have an aver-age latency of a few tens of ms, e.g., 15–35 msin Starlink and 32 ms in OneWeb. [97] The establishment of LoS links in UAV communi-cations can reduce the queuing latency compared toterrestrial communications. • Low latency is of importance for CNPC linksLAP/HAP communications. [100] A multiantenna system was proposed to guaranteelatency and loss probability in URLLC LAP systems. • In ARANs, the low tiers provide lower la-tency, while high tiers provide better coverage. [101] The achievable rate of CNPC links, which becomeshigher at the price of larger latency, was analyzed.ComputationLatency Computation latency would be a critical com-ponent in ARANs as many compute-intensive [107] An aerial relay is deployed to facilitate the computationof end users, considering the energy-delay tradeoff.applications are emerging, e.g., virtual realityand data analytics in the sky. [110] MEC is integrated into satellite communications tobetter serve sparsely distributed users.MobilityAnalysis SystemMobility • The system mobility is mainly reflected bythe 3-D deployment of ABSs and constellationdesigns in (LEO) satellite communications. [111] UAVs are 3-D deployed for IoT data collection, whichcan reduce the energy consumption by 45% comparedto fixed BSs in terrestrial communications. • ARANs can be dynamically configured andmobilized to improve the QoS of end users andto benefit service/infrastructure providers. [119] The integration of SDN into LEO communications wasproposed to reduce the (migration and reconfiguration)costs of constellation designs.TrajectoryOptimization In the lower tier (LAP) of ARANs, the tra-jectory of ABSs can be optimized for variouspurposes, e.g., for reducing energy consumptionand minimizing the completion latency. [121] The UAV trajectory and the problem of user schedulingwere considered to maximize the data collection rate.The problem was solved using a convex approximationalgorithm and the decomposition technique. users. Consequently, more studies should be conducted onARAN channels to characterize potential scenarios and fre-quency bands in 6G wireless systems. In Section III-B, wereview energy consumption models for UAVs in two modes:hovering as well as vertical and horizontal movements. Itis observed that these models are well-established for bothfixed-wing and rotary-wing UAVs. However, because morepotential technologies and network scenarios that can beintegrated with ARAN infrastructures will be available in 6G,more studies should be conducted. For example, the energyconsumption model may vary if UAVs are equipped withwireless power transfer and/or energy harvesting capabilities.Next, communication and computational latency analysis ofARANs is provided in Section III-C. Similar to energy con-sumption models, latency analysis of ARANs is currentlyperformed in typical 5G scenarios, and thus it should becarried out in potential network scenarios of future 6G wirelesssystems such as massive URLLC and zero-touch services. Finally, system mobility and trajectory scheduling of ARANplatforms are reviewed in Section III-D. Owing to the factthat ARANs would be configured dynamically and controlledadaptively, system performance can be significantly improvedwhen compared with conventional terrestrial/stationary RANs.IV. E
NABLING T ECHNOLOGIES
To realize ARANs in 6G networks, we dedicate this sec-tion to reviewing the enabling technologies, including energyrefills, operational management, and data delivery.
A. Energy Refills
As aforementioned in the ARAN access infrastructure’sdescription, typical ABSs face battery storage capacity lim-itations during flight, which is the primary issue to be ad-dressed. [126]. Therefore, research into energy replenishmentstrategies is essential. Aside from energy replenishment atcharging stations, (self-)recharging through energy harvesting (EH) and RF wireless charging technologies are the mostpromising methods, with several advantages in terms of contin-uous provision of energy for ABSs working aloft. Conversely,we can consider the capability of ABSs to use wireless powerto serve massive low-power ground devices, especially inisolated and remote areas. An illustration of three primaryscenarios for energy replenishment is shown in Fig. 9.
1) Charging Station:
This approach can be divided intothree main categories of traditional charging stations, tetheredtechnology, and near-field wireless charging. a) Traditional Charging Stations:
For extended mis-sions, terrestrial intermediate charging stations are the basicconventional approach to replenish ABSs’ energy. Kantor etal. proposed a patent for a UAV-assisted recharging sta-tion [127]. In this work, a UAV can calculate a flight pathand identify a terrestrial intermediate charging base stationas the most suitable, considering the route characteristics tostop and recharge its batteries and then travel to the nextmission location. In addition to the conventional chargingstation model, Sharma et al. proposed the concept of chargingstations where a UAV’ exhausted batteries are replaced byfully charged ones from terrestrial macro base stations [128].To secure the charging process, several recent works havesuccessfully incorporated blockchain technology to a systemcomprised of an ABS swarm and multiple charging stationsas investigated in [129], [130]. In an ARAN context, theblockchain technology allows secure peer-to-peer transac-tions among multiple networking components such as ABSsand user devices at different geographical locations. In theblockchain model, transaction information is stored in chainedblocks. Every node in the committed chain can see all thetransactions, and thus no node can fabricate or change the com-mitted data in the chain [131]. In particular, the recent studyin [132] deployed a novel application for ABS systems andcharging stations by implementing an advanced blockchain.The proposed solution is based on a tangled data structure thatis equally secure, distributed as conventional blockchain, andsimultaneously reduces power consumption and latency [133].The numerical results in [132] revealed that the solution caneventually provide an ABS swarm and charging stations withoptimal cost and significantly outperform the conventionalstrategies. Although blockchain has been proven as a highlyefficient technology for ARANs, these approaches suffer fromseveral fundamental limitations, including a consensus mech-anism consuming significant energy, considerably substan-tial latency from transaction confirmation, and constrainedscalability [134]. Nevertheless, traditional methods of energytransfer have several limitations such as service interruptionsand small operational areas. b) Tethered Technology:
Recently, tethered technologyhas been considered as a potential solution for ABS en-ergy replenishment and system performance problems inARANs [58]–[60], [135], [136]. Assisted by the tetheredconnections, ABSs can not only replenish their limited batteryonboard via a stable power supply from ground stations butcan also extend the reliability of their backhaul links [59].Wrapped into a tether, two typical cables can be utilizedseparately, such as power and data transmission cables. In reality, various commercial products empowered by tetheredtechnology have been launched by many companies aroundthe world. Tethered UAV (T-UAV) systems are designed withdiverse models, tether lengths, and fight time, as summarizedin Table V.In academic studies, the work in [136] analyzed and com-pared performance between T-UAVs and conventional UAVs,referred to as untethered UAVs (U-UAVs). The numericalresults in this research show that T-UAVs can achieve betterperformance for maximum cellular coverage than U-UAVs.Similar observations have been revealed in [59], where T-UAVs with a 120-meter tether length can extend coverageprobability up to 30% compared to U-UAVs. The studyobserved that the relative mission availability and rechargingtimes of U-UAVs are approximately 70% and 30%, respec-tively, during operations. Meanwhile, 100% available timeis achieved by T-UAVs because of the usage of continuouspower and data cables. Nevertheless, it is necessary to furtherconsider the mobility and endurance tradeoff to optimallyselect whether tethered or untethered solutions should be usedbecause T-UAVs have several native drawbacks such as tetherlength limitation and strict inclination angle consideration toavoid tangling the tether on surrounding buildings or otherobstacles. c) Near-field Wireless Charging:
Near field wirelesscharging techniques are commonly categorized into two ap-proaches, including inductive charging and magnetic reso-nance charging [145]. These approaches can be entirely in-tegrated into a single charging station for ABSs. Owing tothe short range of charging, it is harmless to the humanbody. Although the boundary between the two strategies is notsharply delineated, they usually have distinctive characteristicsthat can be described as follows. • Inductive Charging:
Inductive charging is a wirelesscharging method in which a device uses magnetic induc-tion to deliver electrical energy between two coils. Theinductive charging method operates in the kHz frequencyband with an effective charging distance generally within20 cm [146]. Despite the limited transmission range,at a suitable distance, charging efficiency can be veryhigh (up to 90% within 17.5–26.5 cm [147]). In thecontext of ARANs, the technology is entirely suitablefor deployment in charging stations to recharge the bat-teries of ABSs. For instance, a planar coil misalignmentapproach [148] was proposed for an inductive wirelesscharging system to recharge UAV batteries and reduce theweight of onboard components, leading to a cost reduc-tion and providing a light-weight solution. Meanwhile,Obayashi et al. [149] elaborated a recharging prototypedesign with up to 450-W inductive power in the 85-kHzfrequency band for a fast wireless charging port on a largeairborne component like HAPs. • Magnetic Resonance Charging:
Magnetic resonancecharging utilizes magnetic resonance between the trans-mitter and receiver antennas and operates in the MHz fre-quency bands. Because two resonant coils operate at thesame resonant frequency, the energy transfer efficiencycan be significantly high with only slight leakage and Macrobase station(c) RF wireless charging(a) Charging station (b) Energy harvestingLAPHAP Sun LAP UEsTerrestrial stationsTether
Fig. 9. Primary scenarios of energy refills in ARANs. TABLE VC
OMMERCIAL TETHERED -UAV
SYSTEMS . Company Series of products Maximum tether length Flight time
Lifeline [137] Tethered Phantom 4, Mavic Pro, Mavic 2 Pro, Inspire 1 & 2, etc. 60 m 1.5–2.5 hAcecore [138] Neo Tethered System 60 m hZoe Tethered SystemZiyan [139] Ziyan’s Tethering System 100 m 24 hTethered Drone System [140] Tethered Drone System’s T-UAV 120 m 24 hEquinox Systems [141] Falcon heavy, medium, and light 120 m 30 dHoverfly [142] LiveSky TM Sentry 120 m 20–30 dBigSkyElistair [143] Orion 2 110 m 24 hSafe-T 2 130 m UnlimitedLight-T 4 70 m UnlimitedEagle Sky Light [144] Aquila 100 100 m Unlimited immunity to ambient frequency while meeting LoS trans-fer requirements. Magnetic resonance wireless charginghas also proven its ability to transfer power over longerdistances with higher energy efficiency than the inductivecharging approach [146]. In the context of ARANs,Wang et al. deployed a wireless magnetic resonant powertransfer charging station for UAVs, solving the problemsof endurance [150]. Meanwhile, Qiong et al. [151] pro-posed an optimal design for magnetic resonance wirelesscharging in UAVs. The experimental results in [151] showthat the energy transmission efficiency of their proposalcan be kept stable despite the horizontal movement of thetransmitting and receiving coils.
2) Energy Harvesting:
EH is another energy replenishmentapproach that is especially useful in the context of ARANsgiven their use of typical airborne components. The EHstrategy is notably efficient for a swarm of ABSs instead ofbeing limited to standalone devices as in the aforementionedtraditional energy exchange methods. At present, EH is con-sidered to be one of the best potential solutions for energyconstraint problems, especially in self-powered systems likeARANs, because it allows the harvest of energy from externalambient sources such as solar, thermal, wind, vibration, andkinetic and then converts this energy into electrical current tosupport specific energy demands [152]. Among these ambient sources, solar EH is the most popular approach, where LAP,HAP, and LEO airborne components can be powered by solarenergy via photovoltaic (PV) cells.The amount of available solar energy mainly depends onaltitude, geographic location, radiation, the number of daylighthours, and the day of the year. The solar EH power resourceunder various conditions such as cruising altitude, cloud types,and solar positions can be estimated as [153] P EH ( y ) = SGε ( y ) if y ≥ H up ,SGe − β c ( H up − y ) ε ( y ) if H low < y < H up ,SGe − β c ( H up − H low ) ε ( y ) if y ≤ H low , (11)where y is the cruising altitude of the ABS, S and G correspond to the PV cell area onboard the solar-powered UAV,and the average solar radiation intensity on Earth. β c denotesan absorption coefficient modeling the optical characteristicsof the surrounding clouds. While H up and H low are the upperand lower boundary altitudes of the clouds and ε ( y ) is theenergy harvesting efficiency as a function of y , which can beestimated as ε ( y ) = ϕ ( y ) cos θ, (12)where ϕ ( y ) and θ represent the atmospheric transmittance ofthe cloud at altitude y depending on the cloud type, and the angle between the solar light rays and the PV cell dependingon the solar position, respectively.Derived from (11), it is seen that the altitude of an ABSinfluences the solar EH power significantly. Hence, it is nec-essary to consider a trade-off between EH and communicationbecause more energy can be harvested at higher altitudes,whereas the communication path-loss is also larger as analyzedin Section III-A. There is a variety of research contributionsregarding these issues. In particular, Jashnani et al. emphasizedthat the altitude, payload changes, and flight duration of solar-powered UAVs can be affected by its physical dimensions,weight, area of PV cells, the aspect ratio of the wing, and max-imum battery recharging capacity [154]. The authors estimatedthe available solar power in two typical models with an altitudebelow approximately 2.5 km for one and a higher altitude forthe other. The experimental results on charging capabilitiesfor an engineering ground model with test UAVs in [154]showed excellent efficiency for more than 24 h of continuousoperation. Meanwhile, the solar-based UAV prototypes devel-oped in [155], [156] demonstrated a continuous flight potentialof up to 28 h. In [157], Dwivedi et al. experimented with aday flight for a solar-based UAV on April 1, 2017, where thetest time was from 9:30 to 18:00. From these measurementresults, it is recognized that the generated power is less thanthe power required by the UAV system from approximately15:30 to 17:10. This can be explained by the throttle beingcut and the UAV being directed into an extremely slow glidemaneuver at 0.15 m/s while the available solar power was stillbeing generated normally. In addition, the power generatedduring this period was utilized to charge the batteries. In [158],Wang et al. studied a simulation model of solar cell behaviorfor a solar-powered UAV via the MATLAB/Simulink platformwhere the test time was from 6:00 to 18:00 over three daysfrom July 10 to 13, 2017. This included light rain, cloudy, andsunny days as three typical weather types. The experimentalresults demonstrated the impact of light intensity and ambienttemperature on solar EH efficiency. Meanwhile, Rajendran etal. investigated a maximum power point tracker for the optimaloperation of solar-based UAVs by demonstrating the impactof temperature and solar irradiance intensity on various solarmodule angles [159]. The results showed that approximately45 ◦ C is the optimal operating temperature, and the solar powerrose almost linearly along the tilt angle of this solar module.From the standpoint of optimization problems, Hosseini etal. considered optimal path scheduling and power allocationfor UAVs, where the UAVs’ wings are equipped with PV cellsto harvest solar energy and recharge the batteries [160]. Thecrucial objectives in [160] were to optimize energy storagecoupled with a flight path trajectory based on the allocationof available power. However, a sophisticated path-schedulingalgorithm was called for by the constraints of strict moving tra-jectories and flight altitudes. To this end, Sun et al. examineda design with optimal joint resource allocation including 3-Dposition, power, and subcarrier for a solar-powered UAV [161].Unfortunately, the assumption of constant aerodynamic power-consumption in [161] is unrealistic because it still dependson flight velocity. For this issue, they further investigated acommunication system with a multicarrier solar-based UAV by jointly optimizing the amount of solar EH, aerodynamicpower consumption, capability of the onboard energy storage,and QoS requirements of ground users [162]. The experimentalresults revealed that EH efficiency may be higher when flyingabove the clouds. Furthermore, Zhang et al. [153] exploitedan intelligent power approach by combining a reinforcementlearning mechanism with the impact of a solar-based UAV’sflight altitude. Simulation results demonstrated that this ap-proach could simultaneously improve both communicationperformance and EH efficiency.Conversely, EH strategies from other ambient sources in-cluding thermal, wind, vibration, and kinetic, can be seen ascomplementary ways of utilizing more energy. For instance,Wang et al. published a patent of a multirotor aerial dronewith thermal energy scavenging capability in [163]. The patentpresented a thermoelectric generator that harvests the wastethermal energy from a processor in a UAV, and then theUAV batteries are recharged to prolong flight time. Bonnin etal. in [164] described a comprehensive understanding ofprinciples and mathematical formulae for a dynamic soar-ing technique. This technique is inspired by albatross flightbehavior, with which they can fly against the flow withoutflapping their wings, i.e., without having to waste energy.Meanwhile, Anton is a pioneer in investigating and designingnovel piezoelectric elements based on UAV platforms [165].Koszewnik et al. in [166] proposed a common system forvibrational energy scavenging using a piezoelectric harvesterintegrated into UAV platforms. Experiment results showed thatit is possible to extend flight duration using integrated piezopatch elements with flexible batteries for each UAV wing.Moreover, Anton et al. in [167] developed an onboard hybridmodel using solar and vibrational EH. This hybrid approachtakes advantage of both EH methods and supports higherenergy efficiency.
3) RF Wireless Charging:
RF wireless charging (a.k.a. far-field wireless charging) operates in various frequency bandsfrom 300 MHz to 300 GHz [146]. From the standpoint ofestimation, the amount of RF power, P r , received from thisRF wireless charging strategy is dependent on several typicalparameters such as transmission power, wavelength, and thedistance between transmitter and receiver and can be estimatedfollowing the Friis formula as [168] P r = P t G t G r (cid:18) λ πd (cid:19) , (13)where P t is the total power of the transmitter, λ is thewavelength of the RF signals, d denotes the distance betweentransmitter and receiver, while G t and G r are the transmitterand receiver gains, respectively.The charging distance of this approach can be much furtherthan that of near-field wireless charging, i.e., up to several tensof km. Furthermore, the power inversion efficiency obtainedis significantly high at up to 84% at 5.8 dBm cumulativereceived RF power [169]. In particular, the energy efficiencyparameter, δ , can be estimated by the ratio of the output ofusable electrical power, P DC , and the received RF power, P r , as [168] δ = P DC P r . (14)Next, we invoke two ARAN scenarios applying RF wirelesscharging strategies, where ABSs act as either ( i ) powereddevices recharged by terrestrial macro base stations or ( ii )wireless power supply to user devices. a) ABSs as Powered Devices: This category includes theworks wherein ABSs are powered by terrestrial macro basestations using RF wireless charging technology. For instance,the work in [170] investigated a wireless charging method fora micro-UAV operating at a frequency of 5.9 GHz, where thebatteries were replenished via an integrated rectifier antennawith approximately 5 W of transmit power. A design forthe power receiving side of a wireless charging system forABS applications was proposed in [171]. The results reportedin [171] indicated that the proposed method achieved an en-ergy conversion efficiency of more than 77%. Long et al. [172]developed a framework to energize ABSs using a hybrid ofthe EH and RF wireless charging methods. The frameworkconsists of communication and networking architectures aswell as protocols designed for realizing multidimensionalobjectives of a charging system to significantly prolong thecontinuous operating lifetime of ABSs. The authors in [173]conducted research that investigated the possibility of usingresonant beam charging to replenish the energy for ABSs andachieved the anticipated results. Moreover, they considered thejoint optimization of the ABSs’ trajectory and the rechargingstation’s power. Lahmeri et al. in [174] proved that at least6 laser beaming directors per 10 km are required to ensureprobable energy coverage of above 0.9. Meanwhile, Ouyang etal. solved the problem of jointly optimizing the transmit powerallocation and trajectory of ABSs [175]. Chen et al. in [176]proved the feasibility of this laser approach with an energyconversion efficiency of up to 17.55%. However, this laserbeaming method requires safety measures under Federal Com-munications Commission (FCC) regulations in residential andindustrial areas where human health must be protected andmanaged. b) ABSs as Power Supply: In this category, once ABSshave more available energy, they can be considered as an aerialpower supply to simultaneously energize massive numbersof low-power user devices. For instance, the work in [177]examined optimal energy allocation in the context of a wirelesssensor network energized by dynamic ABSs using RF wire-less charging technology. By exploiting the ABSs’ trajectoryapproach, Xu et al. proposed the first work on characterizingthe energy obtainable by terrestrial users through ABS-enabledwireless power transfer [178]. The authors deployed theirapproach among multiple terrestrial users, where the optimaltrajectory subject to flying velocity constraints was consideredto maximize the energy received by users and minimize energyconsumption. Furthermore, Su et al. [179] recently introduceda dynamic bipartite matching mechanism, where ABSs actas wireless power chargers to replenish the energy of low-power terrestrial devices with maximal charging efficiency.To jointly optimize charging efficiency and communicationthroughput maximization, [180] applied a time and power optimization algorithm to maximize average throughput, withABSs functioning as an original energy source to supplymultiple terrestrial devices. To this end, they designed aharvest-transmit-store strategy.A summary of energy replenishment contributions is pre-sented in Table. VI.
B. Operational Management
This subsection addresses three foundational pillars relatingto operational management planes: network softwarization,mobile cloudization, and data mining.
1) Network Softwarization:
Network softwarization playsa crucial role in harmonizing network and computational re-sources across multiple tiers in ARANs [181], a.k.a. software-defined networking (SDN). To this end, network softwariza-tion improves programmable network management control,enabling global visibility through a central orchestrator [182],[183]. As a result, ARANs are empowered with abilities to( i ) provide network operators with enhanced control, situa-tional awareness, and flexibility, ( ii ) coordinate interferenceavoidance, and ( iii ) facilitate interoperability between networknodes.Typical publications that applied the network softwariza-tion architecture to ARAN were perused. In [184], Zhao etal. proposed an SDN-UAV architecture where SDN-integratedUAV-based radio networks are deployed with separate dataand control planes, and the behaviors of UAVs are controlledby providing network programmability. The results in [184]show that a controller can successfully consider circumstantialinformation related to a global UAV group to select themost suitable flight path route trajectory and avoid collisionsamong multiple UAVs. The work in [185] incorporates SDNdeployment in the LAP/HAP tiers, where an SDN controllerbased on a monitoring platform can make intelligent opti-mization decisions owing to its ability to learn and synthesizenetwork information gathered by ABSs. In considerations ofthe ABSs’ power limitation, a load-balancing algorithm wasproposed. Xiong at al. [186] studied an SDN structure forABS ad hoc topologies, where the network supports flexibledata transmission among payloads, adapts to frequent change,and improves the security of the network topology. In [187],the security management of automatic orchestration, deploy-ment, and configuration in the MEC-LAP/HAP networks wasconsidered. In this context, an integration of SDN orchestrationand network function virtualization (NFV) considered severalcontextual virtual and physical conditions and metrics forcoordination in ABSs.A large-scale implementation of conventional SDNs, knownas Loon SDN, has been exploited to optimally interoperateand coordinate complex networks, especially aerospace net-works [188]. Here integrated terrestrial and aerial segments,corresponding to ground and LAP/HAP/LEO platforms in theARAN network architecture, are considered in view of thepacket routing and physical wireless topology. Regarding LEOcommunication systems, the work in [189] deployed a use-case study of an SDN-enabled LEO satellite space segmentthat considers dynamically changing traffic demands based on TABLE VIS
UMMARY OF CONTRIBUTIONS TO ENERGY REPLENISHMENT FOR
ABS S . Ref. Technology Main contributions [127] Conventional charging station Electrical charging for UAVs from stations[128] Replacing UAV’s exhausted batteries from charging stations[132] Optimized costs, security, power-consumption, and low latency for an ABS swarm[137]–[144] Tethered technology Commercial T-UAVs with the various supported maximum tethered lengths up to 130 m as well as thecapability of unlimited flight duration[59], [136] The significant out-performance is achieved when using the help from T-UAVs’ energy replenishmentperspectives compared to U-UAVs[148] Inductive charging A planar coil misalignment to recharge UAV batteries and reduce weight of UAV onboard components[149] Fast wireless charging prototype for HAP components at 450-W power recharging in the 85-kHz band[150] Magnetic resonance charging Powering UAV batteries and solving the problems of endurance[151] An optimal design for UAVs to maintain stable energy transmission efficiency[154] Solar EH More than 24 h of continuous operation for UAVs[156] Up to 28 h of continuous UAV operation[157] The impact of daytime hours of flight for a solar-powered UAV[158] The impact of light intensity and ambient temperature on solar EH efficiency of UAVs[159] The impact of temperature and solar irradiance intensity on solar-based UAVs for optimal operation[160] Joint optimal path scheduling and power allocation for a UAV with PV cells[161] Joint optimal 3-D position, power, and subcarrier for a solar-powered UAV[162] Joint optimal solar EH, power consumption, onboard energy storage, and users’ QoS requirements[153] The intelligent power mechanism using ML regarding the flight altitude of solar-based UAVs[163] Thermal EH Harvesting the waste thermal energy from a UAV processor to prolong flight time[164] Wind EH A dynamic soaring technique to harvest wind energy and reduce the UAVs energy consumption[166] Vibration EH A piezoelectric harvester integrated into UAV platforms to extend the flight duration[167] Solar and vibration EH An onboard UAV hybrid model to support higher energy efficiency[172] EH and RF wireless charging Prolongation of the continuous operating lifetime of ABSs via combination frameworks[170] RF wireless charging Replenishing batteries via an integrated rectifier antenna with approximately 5-W transmission power[171] More than 77% energy conversion efficiency via power on the receiving side of ABSs[173] Joint optimal ABSs’ trajectory and the recharging station power[174] At least six laser beaming directors are required to ensure the ABSs coverage probability per 10 km [175] Joint optimal transmission power allocation and ABSs’ trajectory[176] Up to 17.55% of energy conversion efficiency for UAVs via the laser beaming approach[177] Optimal energy allocation for dynamic ABS networks[178] Optimal UAV’s trajectory subject to flying velocity constraint for optimal energy efficiency[179] A dynamic bipartite matching mechanism with maximal charging efficiency for ABS networks[180] Jointly-optimized charging efficiency and communication throughput maximization geographical position and end users’ time zones. Recently, thisresearch group has also further investigated an SDN with LEOconstellation based on programmable and reconfigurable con-cepts [119]. Here, programmable SDN controllers that updatethe forwarding rules of data plane devices are integrated withthe control logic. Additionally, the authors in [190] deviseda virtual network based on an orchestration model to realizevirtual resource management for LEO satellite networks.
2) Mobile Cloudization:
Mobile cloudization encompassescomputing resources at all tiers of the ARAN and responds ondemand to the flexibility of network resource allocation [2],[191], [192]. As investigated in Section III-C, the applicationof mobile cloudization includes mutual assistance scenariosbetween ABSs and computing platforms; prime examples areshown in Fig. 7. In these contexts, ARANs benefit from mobilecloudization capabilities for computational offloading in termsof energy efficiency and latency reduction.For instance, the issue of offloading highly intensive compu-tational tasks to LAP/HAP tiers with MEC capability to reduceenergy overhead and execution delay is reviewed in [193].Meanwhile, the work in [194] studied fog-cloud computing cooperation that develops the ability to reduce latency andpower consumption as well as improve the scalability andefficiency of end user information exchange in an ARAN.A similar approach with excellent results between edge andcloud computing and one more feature at high long termperformance in online environments for ABS swarms is alsoexamined in [195]. Furthermore, a joint optimization in dataallocation, trajectory, and energy consumption is achieved forUAV-assisted MEC systems in [196], where a UAV plays therole of a computing server to provide computational offloadingfrom the multiple mobile users’ tasks. One more valuablecontribution is found in [197], where the proposed systemmodel belongs to the category of Fig. 7 (b). In this work,the authors envisioned a visual target tracking experiment testusing deep learning (DL) for a trained convolutional neuralnetwork (CNN) model. The lower layers of a CNN deployedon a UAV can provide sufficient tracking performance in goodimage quality conditions, whereas the higher layers at theMEC server are used in conditions of poor image quality.In addition, other constraints such as resource sharing amongmultiple UAVs and bandwidth for communicating between UAVs and MEC servers are considered as well, showing sig-nificant performance benefits from an MEC implementation.Regarding cloudization in LEO satellite systems, the authorsin [198] investigated an LEO satellite constellation with nano-satellites and CubeSats using fog computing. To realize fogcomputing, satellite computers were included in each of thedistributed satellites [199]. This achieved several advantagesincluding ( i ) delay value of round-trip IoT data transmissiontime, ( ii ) data processing time of IoT devices, ( iii ) orbitaldistributed database transmission time, and ( iv ) computationalload balancing through the LEO satellite IoT System’s orbitaldistributed computing network. An edge computing imple-mentation in LEO satellite networks is reviewed in [110],where a user device without an edge server nearby alsoenjoyed edge computing services via satellite links. Moreover,to achieve parallel computation in LEO–terrestrial networks,a cooperative computation offloading model was designedand simultaneously integrated with network resources via adynamic NFV technique.
3) Data Mining:
This subsection presents the results ofour investigation of data mining techniques as another keyenabler for achieving optimal targets such as reducing en-ergy consumption, improving network security, sharing work-load across an entire network, increasing bandwidth, self-organizing networking configurations, achieving autonomoustraining operations, handling mobile big data, and analyzingmobility [200], [201]. Data mining is defined as the processof applying specific algorithms to extract useful patterns fromdata to feed further activities such as feature prediction andoptimal action decision making [202]. In ARAN architecturemodels, the data is an entire information set located on net-work nodes such as LAP/HAP/LEO components and terrestrialdevices. By exploiting the value of data mining, machinelearning (ML) can provide solutions for simultaneous massivenumbers of user connections in a dynamic, heterogeneous, andunpredictable network resource such as ARANs in the contextof 6G [203].The work in [204] reviewed applications of DL forLAP/HAP components and services where the main appli-cation scenarios were feature extraction, planning, situationalawareness, and motion control. Conversely, the work in [205]surveyed the application of deep reinforcement learning toaddress issues emerging in multitier LAP/HAP systems as wellas communications such as data rate control, dynamic networkaccess, wireless caching, network security, data offloading,connectivity preservation, traffic routing, resource sharing, anddata collection. In [206], a deployment of DL for the UAV-enabled MEC network was exploited to minimize the energyconsumption and the weighted sum of latency. Meanwhile, thework in [40] provided a comprehensive survey on using ABS-assisted cellular communications, where ML algorithms wereused for data popularity analysis for trajectory and placementpurposes. From the standpoint of cybersecurity, the workin [207] provided a comprehensive research on LAP/HAPsystems and communications from a cyber physical securityperspective. Furthermore, the authors in [208] provided a com-prehensive survey of multi-UAV systems, where ML is appliedfor fine-grained cyberphysical applications, highlighting key aspects such as coverage spanning, tracking of targets andinfrastructure objects, energy efficient navigation, and imageanalysis assessment.Regarding LEO tiers, Ferreira et al. [209] devised anddeployed reinforcement learning for LEO satellite communi-cations. ML decisions based on reinforcement learning wereused to configure a satellite link from an LEO constellationto a ground station to promote high throughput, low bit errorrate, bandwidth optimization, and reduced power consump-tion. From another perspective, the work in [210] describedthe use of a LEO satellite dataset recorded at a terrestrialoptical observatory, used along with ML strategies to producehigher resolution image recovery on degraded image setsand perform image interpretability assessment. Subsequently,Chen et al. introduced a novel ML application for estimatingprecipitation based on LEO satellite observations, where anML platform was used to enhance estimation accuracy [211].Especially in [212], multitier LAP/HAP/LEO communicationsystems in ARANs and ground stations were used to inves-tigate DL algorithms and training method implementation toimprove various aspects of network performance.
C. Data Delivery
This subsection provides readers with the enabling tech-nologies that support data delivery in ARANs. In this regard,key technologies for addressing particular issues such asfrequency spectrum, communication protocol, and multiaccessapproaches are introduced.
1) Frequency Spectrum:
As mentioned regarding theARAN architecture above, mobile wireless technologies suchas 5G NR and Wi-Fi are considered to support links amongABSs at LAP/HAP tiers and between ABSs and (terrestrialand aerial) users. 5G NR and Wi-Fi operate at various fre-quency bands including sub-6 and THz frequencies in boththe licensed and unlicensed spectra [213]. Motivated by this,the work in [207] demonstrated that ultra-high-speed wirelessbackhaul can be achieved using 3-D beamforming for ABSsin the mmWave bands, significantly improving flexibility andreducing the comparative cost of wired backhaul in terrestrialnetworks. The author in [214] proposed a mmWave UAV meshnetwork using a fast beam-tracking mechanism to provideultra-high-speed wireless backhaul between UAVs and relaysfor terrestrial base stations. Meanwhile, an efficient channeltracking strategy was proposed for mmWave UAV commu-nications, where the complexity of the downlink channeltracking was significantly decreased by improving both angleand Doppler reciprocities [215]. With regard to LEO satellitesystems, seamless integration of ultra-broadband short-rangewireless network segments into satellite constellations was de-vised to provide multigigabit multimedia to nomadic users viammWave satellite links [216]. The study achieved advantagesof increased throughout and reduced latency in the proposednetwork. The authors in [217] introduced two technologicalenablers applying to mmWave satellite-terrestrial communica-tions, which were a smart antenna approach and programmableintelligent management to enhance mobile broadband accessin denser conditions. Furthermore, THz frequency spectrum bands (i.e., 0.1–10 THz) higher than mmWave are discussed with regard toLAP/HAP/LEO communication links. These THz bands areutilized for LEO satellite links [218], using reconfigurableintelligent surface (RIS) technology to compensate for thehigh path loss at high carrier frequencies, leading to improve-ments in signal-to-noise ratio, i.e., it significantly improvessystem performance. Meanwhile, [219] explored the use ofTHz bands between micro-satellites and ground stations, in ascenario where THz quantum entanglement distribution andTHz quantum key distribution are suitable for deploymentto achieve desirable high performance. In the context ofLAP/HAP communications, the authors in [220] presentedMIMO links in a UAV swarm using orthogonal frequency di-vision multiplexing in the THz band, reaching millimeter-scalepositioning accuracy with respect to separation of dimension,bandwidth, and transmitter-receiver array.
2) Communication Protocol:
As particularly analyzed inSection III-C, URLLC communications are considered oneof the most promising protocols to accommodate ARANs’broad coverage, low latency, and ultra-reliability towards 6Grequirements. Although ARANs and URLLC have a mutualsupportive relation, this section is dedicated to reviewingURLLC communication as an enabler from the perspectiveof ARANs.For instance, the authors in [221] provided an assessmentof short packet communications in a UAV-IoT network, wherespectrum sharing based on intelligent radio strategies pro-grams and dynamically configures the use of the optimizedwireless channels to avoid user interference and congestion.This work invokes the concept of deploying blocklengths asshort as possible to reduce latency and meet URLLC ser-vice expectations for reliable communications between UAVsand IoT devices. In [222], non-orthogonal multiple access(NOMA)-based URLLC communications in ARANs withoutany assistance from terrestrial base stations was analyzed.The authors minimized a block error rate by appropriatelynarrowing the beam width to improve transmission distanceand communication reliability on access links. The beamwidth reduction was adopted via a user grouping approach toreduce coverage of areas with no users. Additionally, the workin [100] investigated a framework for enabling URLLC in thecontext of LAP/HAP systems with control and non-payloadcommunications, where a modified distributed multiantennasystem was adopted to judiciously optimize UAVs’ altitude,the uplink and downlink duration, and the antenna configura-tion. From the standpoint of LEO satellite segments, the workin [223] proposed a power-efficient control link algorithm tominimize total power consumption and response reliabilityby adopting URLLC communications for the satellite linksin the context of ARANs. In addition, the literature reviewin [224] showed that URLLC communications can supportultra-reliability as well as low latency requirements within tensof milliseconds for LEO constellations in ARANs.
3) Multiaccess:
From a multiaccess perspective, massiveMIMO, NOMA, and RIS are considered the main tech-nological enablers to further enhance an ARAN network’sperformance. a) Massive MIMO:
It is well known that a large volumeof data signals can be transmitted and received simultane-ously in the spatial domain by equipping many independentlycontrolled antennas to participate in network nodes, knownas massive MIMO [225]. Additionally, beamforming, basedon a highly directional antenna beam, can be considered asone of the effective solutions to achieve sufficient antennagain to overcome path loss, handle multipath and interfer-ence phenomena, and ensure a high signal-to-noise ratio atoutput [226].Returning to ARAN networks, Chandhar et al. introduced amassive MIMO deployment to a LAP-based (e.g., UAV/droneswarms) network [227], where multiple single-antenna UAVssimultaneously communicate with a multiantenna ground sta-tion. This work established the optimal antenna distanceand demonstrated that the ergodic rate per UAV reacheda maximal value because UAVs are spherically evenly dis-tributed around a ground station with an antenna distancerespecting an integer multiple number of half a wavelength.The authors in [228] provided support for multiuser massiveMIMO systems by UAVs, where the UAV downlink controland command channel was considered under realistic 3GPPassumptions. The results from [228] hold that the massiveMIMO utilized with UAV communications can significantlyimprove system performance across several typical parameterssuch as spatial multiplexing gain, interference mitigation, andcarrier signal strength. Similarly, a UAV network assisted bya cell-free massive MIMO architecture where Rician fadingis examined to obtain a closed-form expression of spectralefficiency was studied in [229]. In addition, user schedulingand power allocation strategies were proposed and shown toprovide superior performance.A massive MIMO channel model for LEO satellite systemswas established in [230], where Doppler and time delaycompensation techniques and user grouping algorithms wereproposed. The study presented a user grouping algorithm aim-ing to schedule access for terrestrial end users using the sametime and frequency resources for various groups. In brief, thedata rate of LEO satellite systems was significantly improvedby massive MIMO model. The work in [231] designed amassive MIMO with downlink transmission for LEO satellitecommunications, where all participating nodes are equippedwith uniform planar arrays. The maximization of the ergodicsum-rate was achieved via a low-complexity algorithm forLagrangian multiplier optimization and provided the expectedperformance improvement. b) NOMA:
The NOMA concept has been widely utilizedin various communication schemes as being advantageous forenhancing network performance by improving spectrum ef-ficiency, increasing capacity, decreasing transmission latency,heightening throughput, and enabling data transmission amonga vast array of UEs that simultaneously request access [232],[233]. There are several publications concerned with theadvantages of the NOMA concept and its integration withLAP/HAP/LEO communications. Chu et al. [234] investi-gated an LEO-terrestrial network of satellites equipped withmultibeam technology, considering massive access using theNOMA scheme. Two robust beamforming algorithms were proposed to minimize total power consumption. Using asimilar approach, Gao et al. [235] contributed to the NOMAtechnique used in LEO satellite communications, achieving asignificantly enhanced performance relating to ergodic capac-ity, outage probability, and mutual information.In the manner of LAP/HAP communications, Nasir etal. [236] deployed a UAV-assisted network to provide a largenumber of terrestrial UEs with multiconnection capabilityusing the NOMA technique, where a path-following algorithmwas proposed to reach the optimal max-min rate. The workin [237] employed beamforming based on the arriving signals’angle and an array of antennas using the NOMA strategy forHAP networks to improve performance by reducing the biterror rate. A contribution to the integration of MIMO andNOMA for UAV-assisted networks can be found in [238], inwhich, as expected, the outage probability and ergodic rateimproved. The diversity order and high signal-to-noise ratioslope were also confirmed. More recently, the integration ofNOMA into UAV-assisted visible light communications wasinvestigated in [239] to maximize the total sum-rate underQoS requirements and various constraints. c) RIS: The novel concept of RIS has been recentlydevised [240], where a joint combination of phase control,angles of incident RF signals, and reflecting phases can bearbitrarily adjusted to create a desirable multipath effect toimprove the received signal power or mitigate interference. Asmentioned in [218] in the high-frequency spectrum subsection,RIS has been integrated into LEO satellite communications atTHz bands with good results. Moreover, an RIS approximatelythe size of two large billboards could significantly improvethe signal-to-noise ratio of LEO-terrestrial satellite links asdemonstrated in [241]. Meanwhile, Hua et al. [242] studied aUAV-assisted RIS radio system, where a passive beamformingtechnique was applied by the RIS to enhance UAV trans-mission. The joint optimization of a UAV’s trajectory, RISscheduling, and RIS phase shift matrix was also consideredin this contribution. In [243], the joint optimization of activebeamforming toward a UAV, passive beamforming toward theRIS, and the trajectory of the UAV were achieved in RIS-assisted UAV communications. The results in this publicationhave demonstrated that this scheme outperforms other bench-mark schemes regarding feasibility and efficacy. Furthermore,the integrating concept among IRS, mmWave beamforming,and ML approaches was invoked in the supporting context ofa UAV network [244]. The preliminary results in [244] hasrevealed that the reflection coefficient of an IRS-based UAVcan be optimized to achieve maximum downlink capacity,and the RL approach can further significantly improve systemperformance compared to a scheme without RL deployment.
D. Summary and Discussion
In brief, three key technological pillars for emergingARANs including energy replenishment, operational manage-ment, and data delivery have been reviewed. In particular,Section IV-A investigates energy replenishment, which is theforemost limiting aspect in ARAN realization to overcomebattery constraints. Three primary categories of energy re-plenishment in ARANs including charging stations, EH, and RF wireless charging are summarized in Table VI. It isobserved that the conventional energy trading techniques suchas a charging station and ABS swapping are inefficient andless dynamic. Although the energy replenishment techniquesof renewable energy sources and RF wireless charging arepromising solutions for ARANs enabling sustainable services,joint optimization of the charging efficiency and communi-cation objectives (e.g., throughput maximization and serviceavailability) considering user mobility and QoS requires fur-ther investigation. Next, Section IV-B presents recent advancesin operational management, which can be listed as networksoftwarization, mobile cloudization, and data mining. Whilenetwork softwarization is beneficial for ARANs with theabilities to ( i ) provide greater control, situational awareness,and flexibility, ( ii ) coordinate interference avoidance, and ( iii )facilitate interoperability among network nodes, ARANs arefurther enabled by mobile cloudization with computationaloffloading in terms of energy efficiency and latency reduction.Additionally, another key enabler for achieving some optimalperformance perspectives in ARANs’ scopes towards 6G ispresented in the data mining subsection. Nevertheless, cooper-ative resource management in 3-D space is needed for efficientutilization of the limited computation, communication, storage,and battery capacity of ABSs. More specifically, importantchallenges such as optimal power control, placement of con-tent on the ABSs, user device association, ABS trajectoryoptimization, and computation resource allocation need furtherinvestigation. Ultimately, data delivery perspectives including( i ) prospective employment of mmWave and THz frequencyspectrums, ( ii ) URLLC protocol utilization, and ( iii ) advancedmultiaccess based on massive MIMO, NOMA, and RIS arewithin the review of Section IV-C. As 6G is expected toexploit higher frequency bands, designing an effective andefficient mobility management scheme for these frequencybands deserves to be investigated. Furthermore, an adaptivebandwidth resource allocation scheme considering the distri-bution of terrestrial users’ traffic, line-of-sight interference,mobility, and positioning of UAVs’ is an important avenueof future research.V. A PPLICATION S CENARIOS
As aforementioned in Section II, hierarchical ARAN ar-chitectures provide heterogeneous communication capabilitiesfrom multiple tiers with distinct functionality and features.As a result, ARAN supports a broad range of emergingapplications and services such as wireless coverage expansion,aerial surveillance, precision agriculture, and commercial de-livery. Depending on the networking requirements, we classifyapplications into three categories: event-based, scheduled, andpermanent communications. The details are described below.
A. Event-based Communications
Event-based communications define application scenariosin which networking infrastructures are temporarily requiredto provide and/or boost communication services for shortduration events. As prime examples in this category, disasterand SAR scenarios are investigated.
1) Disaster:
Emergency management systems currentlydepend on the wireless communication infrastructure. Whenlarge-scale disasters or catastrophes occur in an area, thecommunication infrastructures are often severely damaged.Despite some communication components remaining func-tional, the communication systems have to deal with networkcongestion. This leads to various problems for emergencyservice providers, such as difficulty in acquiring real-timeinformation from users. To this end, the work in [245]demonstrated that UAVs/drones are suitable for deployment,where these networking devices are configured to form animpermanent cluster providing base station functions. Theproposed cluster including multiple UAVs forming hexagonalcells was investigated. Regardless of the initial deploymenttopology, the UAV clusters self-organize and operate as atemporary access infrastructure to provide reliable airbornecommunication links. Furthermore, to enhance the ability todetect and secure emergency services in disasters, proposalson photogrammetry and geocomputing based on UAV-assistedfunctions are investigated in [246]. In these proposals, theUAVs’ main uses are given as ( i ) regular scheduling to monitorand map land features and their transformation over time toanticipate potential hazards and disasters, ( ii ) observing humanactivities during an emergency, ( iii ) altering telecommunica-tions components when they are damaged in disasters, and( iv ) delivering essential materials to disaster-stricken isolatedareas. For instance, the authors in [247] presented severalplatforms (e.g., UAVs/IoT-based or inter-integration strategies)for disaster management, where the information harvestedby UAVs in disaster-affected areas can be analyzed to maketimely and correct decisions on the assistance for people inthese localities. A schedule for data collection that especiallyconsiders the energy-efficiency of UAVs is also considered inthis contribution. Additionally, UAV applications for assessingthe damage to cultural heritage sites after earthquakes havebeen also considered in [248]. As an example, the SulamaniPagoda in Bagan, Myanmar was severely damaged by astrong earthquake on August 24, 2016. This work revealedthe classification of the point cloud strategies for UAV-basedfunctions intended to distinguish damaged parts of the pagodaas well as analyze and quantify the level of damage at thisarchitectural heritage site.With the help of LEO satellite systems, satellite imageanalysis enhances disaster detection accuracy improvementcapabilities to undertake rescue missions, coordinate reliefefforts, and respond to disasters, etc. in a rapid, timely, andefficient manner as well as to identify the extent of thedamage. Amit et al. [249] exploited an automatic disasterdetection system based on DL techniques to analyze imagescollected from LEO satellites. The experimental results mainlyfocus on two disasters (e.g., flood and landslide) in Japanand Thailand and demonstrated an accuracy of 80%–90%in disaster detection for both. Meanwhile, a framework forchange detection in flood and fire disasters was introducedin [250]. Based on DL algorithms using satellite images,81.2% and 83.5% detection accuracy are obtained respectivelyfor flood and fire disasters. Moreover, the work in [251]provided an overview of several communication paradigms (e.g., device-to-device, IoT, UAVs, cellular, mobile ad-hoc,and satellite networks) inter-linked for post-disaster emergencycommunication services. These cooperative networks supportrecovery and ensure failure detection, failure recovery, and thelocalization of failures that occur owing to disasters.
2) SAR:
Regarding SAR application scenarios, the authorsin [252] proposed a multiobjective optimization algorithmfor a UAV swarm aiming to find a targeted object andestablish continuous communication between the target andterrestrial personnel as rapidly as possible. It is recognizedthat overall mission completion times have been significantlyreduced prosegmentally to the number of deployed UAVs.Another work in [253] similarly elaborated on emergencieswhere multiple UAVs are assigned for SAR missions asrapidly as possible such that the maximum number of peopleare saved. The proposed algorithm’s performance has beencompared to other algorithms (e.g., multi-UAV task allocationand opportunistic task allocation) in terms of mission durationand survivor rescue rate, and has demonstrated superior per-formance. Furthermore, the work in [254] described severaluseful mechanisms for sound source enhancement, soundsource localization, and robust communication, where theseapproaches were applied to a UAV conducting SAR missionswith a microphone array to enhance rescue performance inoutdoor conditions.The proposed model in [255] was designed to utilize LEOsatellite systems as follows. ( i ) Life vests are equipped with aglobal positioning system (GPS) receiver to simplify locationservices and send information at low cost with high dataintegrity to remote locations, and with health sensors toquickly detect any health problems as well as maximize QoSand delivery probability. ( ii ) These life vests can be consid-ered an ad-hoc sensor network communicating with an LEOsatellite system via a satellite master node. ( iii ) LEO satelliteterminals are used to provide long-range connectivity betweena satellite master node and a terrestrial satellite gateway, and( iv ) distress signals are forwarded to a mission control center toassist in planning rescue missions accordingly. The proposedsystem’s experimental results demonstrated its capability toimprove performance and ensure the high reliability of datatransmissions for maritime emergency communications. B. Scheduled Communications
Scheduled communications cover scenarios where theARAN components fly on a predefined path to provide userswith networking services of a given duration. Two primaryexamples of this category, aerial surveillance and smart agri-culture, are discussed in this subsection.
1) Aerial Surveillance:
Airborne surveillance usingUAVs/drones is well known as a thoroughly complete,flexible, and unique solution for several special use casessuch as ensuring territorial integrity and national security,resource exploration in dangerous areas, wildfire and oilspill detection, and environmental surveillance. In [256],Rossi et al. proposed a modular embedded sensing systemin which UAVs carried sensors used for measuring reducinggases. Furthermore, a monitoring algorithm for gas leakage localization was proposed, which considers optimizingUAV flight duration and energy consumption. A practicalmodel for an aerial surveillance system using drones wasproposed in [257]. The advantages of this proposed systemare real-time monitoring with audio, extended operatingtime, and multimode operation at an affordable cost. Inspiredby this work, many reports investigated various interestingaspects under considerations including surveillance in remoteareas [258], extended coverage with an energy-efficientrouting path scheme under resolution constraints [259], andmobility-aware control mechanisms for cooperative dronesvia heterogeneous Wi-Fi and satellite networks [260].Specifically, the UAV-based model proposed in [258] canbe used where dense foresting obstructs the detection of risksand hidden activities, e.g., guerilla activities. By using thermalcameras, UAVs can monitor and report minute details ofthese hidden activities. The proposed system is suitable forreconnaissance by army forces or for surveillance in remoteareas that cannot be reached by human beings. In addition,a UAV-based network with the operating scheme proposedin [259] can be deployed in a geometrically complex targetarea. It can capture images of a specified spatial resolution.The authors of this contribution also investigated the solutionsto a routing problem to reduce a UAV cluster’s maximumenergy consumption to attain better coverage. Meanwhile, acontrol problem in a heterogeneous network was examinedunder three scenarios in [260]. In particular, a cooperativeUAVs-based surveillance system operates under the control ofa server via ground base stations (access points) through Wi-Fiand satellite communications. Based on the function of UAVsin a cluster under multiple constraints, the control mechanismautomatically adjusts the distance between the cluster andaccess points in terms of Multipath TCP. The results revealthat the designed system can operate stably and support robustbandwidth.
2) Smart Agriculture:
In smart agriculture, farmers useUAVs not only for crop spraying but also for many functionsincluding crop monitoring and disease detection. The primaryfeature of smart agriculture is to integrate the latest infor-mation and communication technologies (ICT) such as IoT,big data, ML, and UAVs into farming operations [261]–[264].In [265] many kinds of UAVs equipped with suitable sensors,and cases of UAVs being used in smart farming scenarioswere thoroughly reviewed. In particular, multitier ARANs canbe exploited for many potential agricultural applications, suchas weed management, crop health monitoring, plant countingand numbering, pest management, and assessing plant quality.It is clear that an integrated system with the aforementionedfunctions would be very useful to farmers.Furthermore, smart agriculture applications based on accu-rate land monitoring for timely support actions to ensure safefood production need remote sensing and satellite data forcheap and timely paddy mapping. With this as motivation, thework in [266] introduced a contribution to the developmentof an autonomous and intelligent agriculture system builtusing available LEO satellite data. A novel multitemporalhigh-spatial-resolution classification method was applied to atestbed based on a case study of the Landsat 8 data. Landsat 8’s data has been widely utilized in various publicationsincluding satellite field support for smart agriculture becauseof its high spatial resolution. Particularly, in this contribution,the tested model for land monitoring attained an accuracy ofup to 93%. Other works [267], [268] were conducted basedon this Landsat 8 data (i.e., satellite-based data). It furtherencompassed these approaches to LAP/HAP components aswell. Along with the linkage between satellite and drone/UAVdata, larger scale precision monitoring in agriculture has beenachieved.
C. Permanent Communications
Permanent communications involve application scenarioswhere networking services are required continuously overa long period. Prime examples in this category are smartcity applications such as urban monitoring, advanced health-care services, and intelligent transportation systems (ITSs)as well as networking infrastructure for remote and isolatedareas [269].
1) Urban Monitoring:
In contrast to UAV-based aerialsurveillance applications mentioned in the scheduled com-munication category, which focus on control algorithms aswell as detection of unusual movement, research on urbanmonitoring typically focuses on solutions for collecting real-time multimedia data with surveillance systems. A design fora UAV assisted urban monitoring system under tactile Internetconstraints was examined in [270]. This work proposed tactileInternet architecture to determine system elements’ compli-ance with tactile Internet requirements in a 5G ecosystem.In addition, Jin et al. [271] proposed four algorithms: ( i )UAV placement, ( ii ) addressing the bi-objective fragile binpacking problem, ( iii ) obtaining circles for a complete graph,and ( iv ) refining UAV candidate routes. In this network, a UAVcluster functioning as ABSs forms a rectangular boundary inoptimal positions to provide reliable, real-time, and feasibleservices while operating in a heterogeneous communicationenvironment in a smart city. The authors established that theiralgorithms improve throughput, video quality, and delay invideo surveillance systems.
2) Health Care:
Applying advanced networking technolo-gies for healthcare can significantly reduce healthcare ex-penditure for both government and citizens, in particular tothe elderly who need daily assistance. Recently, electronichealthcare (e-health) services provided electronically via theInternet are being gradually formed to provide ubiquitous,continuous, and personalized medical assistance. The authorsin [272] proposed a smart UAV-assisted healthcare architectureto solve the limitations on the Internet of Medical Things inbody area networks (BANs). Particularly, a UAV connects tomultiple BANs via wake-up radio-based communication ina star topology. The UAVs function as data collectors withefficient power consumption. In [273], the UAVs’ positioningin a service area was examined to ensure the minimumnumber of serving UAVs in the network for each patient to bewithin their coverage area. Besides, the positioning problem,UAV services are also considered under the constraint thatthe collected health data must be efficiently processed. The authors proved that their particle-swarm optimization-basedalgorithm could significantly reduce the number of UAVsused for e-healthcare services under data-driven constraints.Subsequently, for decentralized secure and reliable data link-ing between the UAVs and other entities, a scheme thatapplies blockchain-based outdoor medical delivery, namelyVAHAK, was proposed in [274]. Performance analysis pa-rameters relating to scalability, bandwidth, latency, and datastorage cost were investigated to prove the efficacy of usingVAHAK in health care services. Conversely, the researchwork in [275] proposed a framework for exchanging medicalinformation when medical devices are distributed in manyscattered locations. Specifically, medical data from patientsin remote/rural areas are directly transmitted via secure androbust satellite links. Similarly, to improve global access toe-health, especially for remote rural communities, Anema etal. [276] reviewed Canadian space technology applicationsusing LEO connections for e-healthcare.
3) ITSs:
According to development trends in ITSs, vehiclecomponents with autonomous properties will be integratedinto systems forming a large-scale ITS. This leads to manyopportunities for new services as well as new applicationsto be introduced with the support of 6G wireless networks,e.g., flying taxi service. In the next-generation ITSs, trans-port automation needs to be fully implemented, not onlyautomating the vehicles but also including the automation ofremaining components such as roads and end-to-end transportsystems, e.g., terminal communications, field support teams,traffic police, road surveys, and rescue teams. This can berealized by the large-scale use of intelligent and reliableARAN components [277]. In some congestion predictionproblems, the proposed algorithms must count the number ofvehicles circulating in many places in a large region [278].UAVs with cameras used for traffic monitoring as well asfor data collection produce better results than fixed recordingdetectors permanently installed at traffic intersections becausethe targeting mechanism of UAV is based on the true density.Furthermore, UAVs can act as intermediate agents to in-terconnect either among vehicles or to roadside units in theITSs to improve the overall quality of experience [279]. Inparticular, a joint optimization of trajectory planning andresource allocation was investigated for critical data deliv-ery in UAV-assisted vehicular networks [280]. Adapting tosituational changes, a low-complexity trajectory design withminimum communication resources is achieved to fully serveall vehicles. By contrast, the work in [281] considered datacaching and trajectory design of UAVs to support ITSs. Aconvolutional neural network-based approach was utilized tomaximize network throughput for reliable content delivery.In addition, a dynamic positioning technique for 3-D UAVpositions was investigated in [282] to optimize vehicularcommunications’ QoS. Next, the minimum number of power-limited UAVs in a swarm involved in the optimal routingscheme with strict considerations of communication delayand energy consumption was addressed in densely crowdedvehicular environments [283].Additionally, to meet the stringent requirements for ITSservices with improved reliability, positioning accuracy, and continuous location updates, more complex navigation strate-gies must be adopted. Global navigation satellite systemsare the potential candidates that demonstrated two majoraspects of ITSs and location-based applications to provideaccurate global position, velocity, and time data where theyare supported by LEO satellite systems. The authors in [284]introduced critical problems that can be expressed as follows.In addition to classical inertial measurement devices (e.g.,accelerometers or high/low-grade gyroscopes), there are sev-eral other types of signals and sensors such as barometers,magnetometers, cameras, mobile network signals, and signalsof opportunity to be considered. Such signals and sensorsthat are integrated into navigation systems can be assessedand used for autonomous mobility to deploy ITS applicationsappropriately.
4) Networking in Underserved Areas:
In [285], UAVs wereassigned as intermediate aerial nodes to improve coverage aswell as boost the system capacity. In this system, UAVs createmultiple intermediate links to connect users in macro cells andsmall cells. The proposed system is evaluated according to sev-eral performance parameters: network delay, throughput cov-erage, and spectral efficiency. Compared to systems withoutUAV assistance, it can improve efficiency by up to 38% andreduce delays by up to 37.5%. Therefore, this model is suitablefor areas with high demand for connections. Similarly, thefoundation of an airborne wireless cellular network was firstintroduced in [286]. By cooperating with local base stationsand LEO satellite communications, the network establishesa large coverage to provide Internet services for sparselypopulated communities (in forests, deserts, at sea, etc.). Insuch circumstances, the LEO satellites help to transfer databetween these localities and the Internet [287].
D. Summary and Discussion
This section presents three emerging application scenar-ios based on networking requirements including event-based,scheduled, and permanent communications. In particular, Sec-tion V-A represents existing studies primarily focused onapproaches to allow clusters of UAVs to self-organize andautomatically form network replacements when the currentnetwork needs to be supported in providing and/or boostingcommunication services for short duration events. In addition,UAVs/IoT-based or inter-integration strategies are proposedfor post-disaster management. Moreover, in this model, LEOsatellite systems implementing advanced deep learning tech-niques are taken into account in detecting and managingnatural disasters. For SAR application scenarios, related worksfocus on optimization algorithms to control a UAV swarm inseeking targeted objects and in establishing transparent andimmediate communication links between the targets and ter-restrial personnel. In most event-based application scenarios,LEO satellite systems are utilized as an auxiliary supportsystem to enhance the accuracy of the UAVs assigned toSAR missions. With scheduled communications mentioned inSection V-B, the proposed works mainly focus on detectingunusual activities in cases of aerial surveillance and methodsfor utilizing UAVs to autonomously perform smart farming jobs for the other sub-group application. Next, Section V-Cprovides researches relevant to smart city applications thatrequire permanent communications to be deployed. In contrastto applications classified as the scheduled communicationsapplication group (Section V-B), smart city applications (Sec-tion V-C) must deal with tremendous amount of data forstorage and analysis. Consequently, related works in thisSection V-C are primarily designed for UAV-assisted modelsunder tactile Internet constraints, UAV position trajectory op-timization approaches, algorithms to improve throughput andvideo quality, methods to allow UAVs to function as sufficientdata collection systems to decrease operational costs. Finally,some articles in underserved areas are mentioned to provethat the multi-tier ARAN networks can improve coverage andreduce delay in data transmission, especially in sparsely pop-ulated areas. Besides the significant advantages of exploitingARAN in cost-efficiency 6G scenarios as discussed in [288],optimal trade-offs between cost efficiency and technologicalfacilitation still need further realistic studies. In addition,telecommunication subscribers may have concerns regardingtheir privacy (e.g., urban surveillance), personal informationleakage (UAVs often collect data by design), and high-techeducation (e.g., doctors or farmers need to be well-equippedwith high-tech knowledge to efficiently use complicated 6G-driven applications). These concerns can be used as guidelinefor future investigations on 6G networks.VI. R ESEARCH C HALLENGES
This section specifies and discusses the challenges and openresearch issues to spur further investigation of ARANs in 6Gcontexts.
A. Intelligent Radio
The emerging AI chip revolutions empower communicationdevices with high computing capabilities for adaptive reactionto environmental changes in real time. In this circumstance,the 6G envisions RANs growth from the current NR (of 5G)to the next generation: intelligent radio (IR) [289]. IR definesRAN capabilities of exploiting advanced AI chip powersat both user and network devices to intelligently select themost appropriate algorithms for radio frequency planning,spectrum sharing, channel modulation/coding/estimation, andmultiaccess schemes. Considered as a native component ofcomprehensive 6G access infrastructures, ARANs have tobe designed to embed a flexible federated learning modelto orchestrate networking behaviors between the ABSs andmultiple user devices interactively. In addition, the varietyof AI chip classes in user devices is a significant challengeto building effective learning models according to differentservice requirements. Moreover, the locality of user traffic andbehaviors such as service type, volume, and reliability as wellas mobility should to be studied and exploited owing to theirsignificant impact on channel communication between ABSsand user devices.
B. Extremely High Spectrum Exploitation
Continuing the success of 5G in exploiting high spectrumfor Gbps data rates, 6G access networks will expand broader and higher spectrum utilization to achieve Tbps connections.In this regard, THz and visible light communications are themost promising candidates [22], [290]. Although literaturehas witnessed some breakthroughs in this extremely highspectrum exploitation for intra-tier and inter-tier communi-cations in ARAN, there exist inevitable challenges aheadto realize effective and optimal communications. First, highspectrum communications are sensitive to attenuation becauseof environmental condition changes, especially within the 3-Dspace between ABSs and end users. Therefore, an adaptivechannel propagation model and estimation should be carefullyconsidered to react to any environmental changes in real time.In addition, super-narrow beamforming techniques with highdirection degree and fast transformation should be furtherstudied to improve transmission efficiency. Moreover, eventhough ARANs have advantages in high probability of LoSsignal propagation, a relay solution through RIS cannot beignorable in the future research.
C. Network Stability
In ARANs, ABSs interconnect with each other using aerialad hoc technologies and have hierarchical networking overlaysamong the LAP, HAP, and LEO communication tiers. Al-though the high mobility of ABSs and hierarchical networkingoverlays assist ARANs to adapt flexibly to the requirementsof airborne and terrestrial end users globally, guaranteeingnetwork stability for such a highly dynamic and resource-constrained platform is difficult. More specifically, the highlydynamic network topology and intermittent connectivity dueto the high mobility of ABSs in 3-D space, limited resources,and varying QoS requirements of applications are challengingfor the network organization design of both intra-tier and inter-tier communications in ARANs. Thus, the routing protocolsand network organization designs for ARANs should considerdifferent mobility patterns in 3-D space, traffic characteristics,available resources, and load balancing among networkinginfrastructures as well as multiple backup routes for faulttolerance and reliable packet delivery.
D. Security and Privacy Issues
Security and privacy are the most critical issues to beresolved for ARANs. The shared wireless links, potential line-of-sight links among real platforms and aerial platforms andground-users, limited resources (energy, computation, etc.)at some typical aerial nodes, mobility features, a massivedevice communications, among other factors, contribute to thechallenge of meeting ARAN security requirements. Crypto-graphic solutions are mainly considered to combat networksecurity problems at the higher communication protocol layers.However, conventional cryptographic techniques, for instance,public-key cryptography involving high computation and delaycosts in the encryption and decryption processes may not befeasible for dynamic ARANs using resource-constrained aerialplatforms. The high mobility and deployment of UAVs inopen areas can also render achieving physical layer securitysolutions utilizing the inherent features of wireless channelsmore challenging. In addition, the potential line-of-sight links and high mobility features of UAVs in ARANs can be anotherthreat to ground communication networks as they can be easilyjammed or eavesdropped on by malicious UAVs once they arepart of the network. Thus, designing a secure communicationprotocol for ARANs that considers the limited resource andmobility features of typical ABSs is crucial and worth furtherinvestigation. UAVs fitted with video cameras in the LAP/HAPtiers of ARAN are perceived to be privacy hazards owingto their ability to capture videos from unanticipated anglesor areas. Hence, a privacy-preserving mechanism may berequired for UAVs that capture videos. E. Simulation Tools
The performance of the proposed solutions for ARANscan be evaluated using either real experiments or software-based simulations. Real experiments enable the evaluation andanalysis of the proposed algorithms, systems, and protocolsin real environments. It is, however, very difficult to conductexperiments with aerial platforms in many cases because of thehigh cost, large space, strict regulations, difficulties related toscenario repetition, and the complexity of building large-scalenetworks with varying topologies. Consequently, simulation-based performance evaluation is considered to be a viableoption by most researchers owing to its flexibility and lowercost. However, the existing simulation-based evaluations areoften conducted with the assumption that aerial platformssuch as UAVs or drones can move in any direction at anytime with no mechanical and aerodynamic constraints due todifferent UAV types and/or any specific constraints occasionedby environmental obstacles. Hence, the simulation resultsmay not accurately reflect the real environments in manyspecific scenarios. Thus, it would be highly beneficial to designsimulation tools for aerial platforms that can fully characterizemechanical and aerodynamic constraints and capture envi-ronmental factors, allowing researchers to precisely simulatevarious types of UAVs based on their hardware specificationsand subject to environmental factors.VII. C
ONCLUSION
This paper has presented a thorough survey of publicationson network design, system models, enabling technologies, andapplication of ARANs toward a comprehensive 6G accessinfrastructure. First, ARANs are positioned in the 6G archi-tecture to demonstrate their roles, features, and relationshipswith other elements of the networks. Thereafter, an ARANreference model is derived from recent ETSI and 3GPPstandards released for 5G networks and STINs. Subsequently,we analyzed key technical aspects of the systems regardingtransmission propagation, energy consumption, communica-tion latency, and network mobility, followed by enablingtechnologies for improving the performance of these aspects.Finally, research challenges are detailed to offer readers theexpected future trends in ARAN studies in the context of 6Gnetworks. R
EFERENCES[1] K. David and H. Berndt, “6G vision and requirements: Is there anyneed for beyond 5G?”
IEEE Vehicular Technology Magazine , vol. 13,no. 3, pp. 72–80, Jul. 2018. [2] N.-N. Dao, W. Na, and S. Cho, “Mobile cloudization storytelling:Current issues from an optimization perspective,”
IEEE Internet Com-puting , vol. 24, no. 1, pp. 39–47, Jan. 2020.[3] N.-N. Dao, J. Lee, D.-N. Vu, J. Paek, J. Kim, S. Cho, K.-S. Chung, andC. Keum, “Adaptive resource balancing for serviceability maximizationin fog radio access networks,”
IEEE Access , vol. 5, pp. 14 548–14 559,Aug. 2017.[4] ITU-T Rec. Y.3172, “New services and capabilities for Network 2030:Description, technical gap and performance target analysis,” Oct. 2019.[5] Q. Bi, “Ten trends in the cellular industry and an outlook on 6g,”
IEEECommunications Magazine , vol. 57, no. 12, pp. 31–36, Dec. 2019.[6] S. Chen, Y.-C. Liang, S. Sun, S. Kang, W. Cheng, and M. Peng,“Vision, requirements, and technology trend of 6G: how to tackle thechallenges of system coverage, capacity, user data-rate and movementspeed,”
IEEE Wireless Communications , vol. 27, no. 2, pp. 218–228,Feb. 2020.[7] K. Samdanis and T. Taleb, “The road beyond 5G: A vision and insightof the key technologies,”
IEEE Network , vol. 34, no. 2, pp. 135–141,Feb. 2020.[8] ITU-R Report ITU-R M.2410, “Minimum requirements related totechnical performance for IMT-2020 radio interface(s),” Nov. 2017.[9] H. Zhang, L. Song, and Z. Han, “UAV assisted cellular commu-nications,” in
Unmanned Aerial Vehicle Applications over CellularNetworks for 5G and Beyond . Springer, 2020, pp. 61–100.[10] M. Abo-Zeed, J. B. Din, I. Shayea, and M. Ergen, “Survey on landmobile satellite system: Challenges and future research trends,”
IEEEAccess , vol. 7, pp. 137 291–137 304, Sep. 2019.[11] W. Saad, M. Bennis, and M. Chen, “A vision of 6G wireless systems:Applications, trends, technologies, and open research problems,”
IEEENetwork , vol. 34, no. 3, pp. 134–142, May/Jun. 2020.[12] M. Giordani, M. Polese, M. Mezzavilla, S. Rangan, and M. Zorzi,“Toward 6G networks: Use cases and technologies,”
IEEE Communi-cations Magazine , vol. 58, no. 3, pp. 55–61, Mar. 2020.[13] E. Yaacoub and M. Alouini, “A key 6G challenge and opportu-nity—connecting the base of the pyramid: A survey on rural connec-tivity,”
Proceedings of the IEEE , vol. 108, no. 4, pp. 533–582, Apr.2020.[14] R. Shafin, L. Liu, V. Chandrasekhar, H. Chen, J. Reed, and J. C.Zhang, “Artificial intelligence-enabled cellular networks: A critical pathto beyond-5G and 6G,”
IEEE Wireless Communications , vol. 27, no. 2,pp. 212–217, Apr. 2020.[15] H. Song, J. Bai, Y. Yi, J. Wu, and L. Liu, “Artificial intelligence enabledInternet of things: Network architecture and spectrum access,”
IEEEComputational Intelligence Magazine , vol. 15, no. 1, pp. 44–51, Feb.2020.[16] Z. Zhang, Y. Xiao, Z. Ma, M. Xiao, Z. Ding, X. Lei, G. K. Kara-giannidis, and P. Fan, “6G wireless networks: Vision, requirements,architecture, and key technologies,”
IEEE Vehicular Technology Mag-azine , vol. 14, no. 3, pp. 28–41, Jul. 2019.[17] S. Ali, W. Saad, N. Rajatheva, K. Chang, D. Steinbach, B. Sliwa,C. Wietfeld, K. Mei, H. Shiri, H.-J. Zepernick et al. , “6G whitepaper on machine learning in wireless communication networks,” arXivpreprint arXiv:2004.13875 , 2020.[18] E. Peltonen, M. Bennis, M. Capobianco, M. Debbah, A. Ding, F. Gil-Casti˜neira, M. Jurmu, T. Karvonen, M. Kelanti, A. Kliks et al. , “6Gwhite paper on edge intelligence,” arXiv preprint arXiv:2004.14850 ,2020.[19] N. Rajatheva, I. Atzeni, E. Bjornson, A. Bourdoux, S. Buzzi, J.-B.Dore, S. Erkucuk, M. Fuentes, K. Guan, Y. Hu et al. , “White paperon broadband connectivity in 6G,” arXiv preprint arXiv:2004.14247 ,2020.[20] N. H. Mahmood, S. B¨ocker, A. Munari, F. Clazzer, I. Moerman,K. Mikhaylov, O. Lopez, O.-S. Park, E. Mercier, H. Bartz et al. , “Whitepaper on critical and massive machine type communication towards6G,” arXiv preprint arXiv:2004.14146 , 2020.[21] X. You, C.-X. Wang, J. Huang, X. Gao, Z. Zhang, M. Wang, Y. Huang,C. Zhang, Y. Jiang, J. Wang et al. , “Towards 6G wireless communica-tion networks: Vision, enabling technologies, and new paradigm shifts,”
Science China Information Sciences , vol. 64, no. 1, pp. 1–74, Jan. 2021.[22] T. Huang, W. Yang, J. Wu, J. Ma, X. Zhang, and D. Zhang, “A surveyon green 6G network: Architecture and technologies,”
IEEE Access ,vol. 7, pp. 175 758–175 768, Dec. 2019.[23] T. O. Olwal, K. Djouani, and A. M. Kurien, “A survey of resourcemanagement toward 5G radio access networks,”
IEEE CommunicationsSurveys & Tutorials , vol. 18, no. 3, pp. 1656–1686, Thirdquarter 2016. [24] A. Mohamed, O. Onireti, M. A. Imran, A. Imran, and R. Tafazolli,“Control-data separation architecture for cellular radio access networks:A survey and outlook,” IEEE Communications Surveys & Tutorials ,vol. 18, no. 1, pp. 446–465, Firstquarter 2015.[25] M. Peng, Y. Sun, X. Li, Z. Mao, and C. Wang, “Recent advancesin cloud radio access networks: System architectures, key techniques,and open issues,”
IEEE Communications Surveys & Tutorials , vol. 18,no. 3, pp. 2282–2308, Thirdquarter 2016.[26] I. A. Alimi, A. L. Teixeira, and P. P. Monteiro, “Toward an efficient C-RAN optical fronthaul for the future networks: A tutorial on technolo-gies, requirements, challenges, and solutions,”
IEEE CommunicationsSurveys & Tutorials , vol. 20, no. 1, pp. 708–769, Firstquarter 2018.[27] M. A. Habibi, M. Nasimi, B. Han, and H. D. Schotten, “A comprehen-sive survey of RAN architectures toward 5G mobile communicationsystem,”
IEEE Access , vol. 7, pp. 70 371–70 421, May 2019.[28] Q.-V. Pham, F. Fang, V. N. Ha, M. J. Piran, M. Le, L. B. Le, W.-J.Hwang, and Z. Ding, “A survey of multi-access edge computing in5G and beyond: Fundamentals, technology integration, and state-of-the-art,”
IEEE Access , vol. 8, pp. 116 974–117 017, 2020.[29] M. Peng, S. Yan, K. Zhang, and C. Wang, “Fog-computing-based radioaccess networks: Issues and challenges,”
IEEE Network , vol. 30, no. 4,pp. 46–53, Jul.-Aug. 2016.[30] A. Reznik, L. M. C. Murillo, Y. Fang, W. Featherstone, M. Filippou,F. Fontes, F. Giust, Q. Huang, A. Li, C. Turyagyenda et al. , “CloudRAN and MEC: A perfect pairing,”
ETSI MEC , no. 23, p. 25, Feb.2018.[31] Q.-D. Vu, L.-N. Tran, M. Juntti, and E.-K. Hong, “Energy-efficientbandwidth and power allocation for multi-homing networks,”
IEEETransactions on Signal Processing , vol. 63, no. 7, pp. 1684–1699, Apr.2015.[32] X. Wang, L. Kong, F. Kong, F. Qiu, M. Xia, S. Arnon, and G. Chen,“Millimeter wave communication: A comprehensive survey,”
IEEECommunications Surveys & Tutorials , vol. 20, no. 3, pp. 1616–1653,Thirdquarter 2018.[33] C. Han, Y. Wu, Z. Chen, and X. Wang, “Terahertz communications(TeraCom): Challenges and impact on 6G wireless systems,” arXivpreprint arXiv:1912.06040 , 2019.[34] S. Lagen, L. Giupponi, S. Goyal, N. Patriciello, B. Bojovi´c, A. Demir,and M. Beluri, “New radio beam-based access to unlicensed spectrum:Design challenges and solutions,”
IEEE Communications Surveys &Tutorials , vol. 22, no. 1, pp. 8–37, Firstquarter 2020.[35] X. Ling, J. Wang, T. Bouchoucha, B. C. Levy, and Z. Ding, “Blockchainradio access network (B-RAN): Towards decentralized secure radioaccess paradigm,”
IEEE Access , vol. 7, pp. 9714–9723, 2019.[36] M. Yao, M. Sohul, V. Marojevic, and J. H. Reed, “Artificial intelligencedefined 5G radio access networks,”
IEEE Communications Magazine ,vol. 57, no. 3, pp. 14–20, Mar. 2019.[37] O. Kodheli, E. Lagunas, N. Maturo, S. K. Sharma, B. Shankar, J. F. M.Montoya, J. C. M. Duncan, D. Spano, S. Chatzinotas, S. Kisseleff et al. ,“Satellite communications in the new space era: A survey and futurechallenges,”
IEEE Communications Surveys & Tutorials , 2021.[38] N. Saeed, A. Elzanaty, H. Almorad, H. Dahrouj, T. Y. Al-Naffouri,and M.-S. Alouini, “CubeSat communications: Recent advances andfuture challenges,”
IEEE Communications Surveys & Tutorials , 2020,in press.[39] P. Wang, J. Zhang, X. Zhang, Z. Yan, B. G. Evans, and W. Wang,“Convergence of satellite and terrestrial networks: A comprehensivesurvey,”
IEEE Access , vol. 8, pp. 5550–5588, 2019.[40] 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 secu-rity challenges,”
IEEE Communications Surveys & Tutorials , vol. 21,no. 4, pp. 3417–3442, Fourthquarter 2019.[41] X. Cao, P. Yang, M. Alzenad, X. Xi, D. Wu, and H. Yanikomeroglu,“Airborne communication networks: A survey,”
IEEE Journal onSelected Areas in Communications , vol. 36, no. 9, pp. 1907–1926, Sep.2018.[42] D. He, S. Chan, and M. Guizani, “Drone-assisted public safety net-works: The security aspect,”
IEEE Communications Magazine , vol. 55,no. 8, pp. 218–223, Apr. 2017.[43] W. Shi, H. Zhou, J. Li, W. Xu, N. Zhang, and X. Shen, “Drone assistedvehicular networks: Architecture, challenges and opportunities,”
IEEENetwork , vol. 32, no. 3, pp. 130–137, Jan. 2018.[44] N. Zhao, W. Lu, M. Sheng, Y. Chen, J. Tang, F. R. Yu, and K.-K.Wong, “UAV-assisted emergency networks in disasters,”
IEEE WirelessCommunications , vol. 26, no. 1, pp. 45–51, Feb. 2019. [45] S. A. Shah, T. Khattab, M. Z. Shakir, and M. O. Hasna, “Associationof networked flying platforms with small cells for network centric 5G+C-RAN,” in
IEEE 28th Annual International Symposium on Personal,Indoor, and Mobile Radio Communications (PIMRC) , Montreal, Que-bec, Canada, 2017, pp. 1–7.[46] X. Huang, J. A. Zhang, R. P. Liu, Y. J. Guo, and L. Hanzo, “Airplane-aided integrated networking for 6G wireless: Will it work?”
IEEEVehicular Technology Magazine , vol. 14, no. 3, pp. 84–91, Jul. 2019.[47] H. Wang, H. Zhao, W. Wu, J. Xiong, D. Ma, and J. Wei, “Deploymentalgorithms of flying base stations: 5G and beyond with UAVs,”
IEEEInternet of Things Journal , vol. 6, no. 6, pp. 10 009–10 027, Aug. 2019.[48] J. Lee and V. Friderikos, “Path optimization for flying base stationsin multi-cell networks,” in , Seoul, South Korea, 2020, pp. 1–6.[49] A. Chriki, H. Touati, H. Snoussi, and F. Kamoun, “FANET: Communi-cation, mobility models and security issues,”
Computer Networks , vol.163, p. 106877, Nov. 2019.[50] D. S. Lakew, U. Sa’ad, N.-N. Dao, W. Na, and S. Cho, “Routing in fly-ing ad hoc networks: A comprehensive survey,”
IEEE CommunicationsSurveys & Tutorials , vol. 22, no. 2, pp. 1071–1120, Mar. 2020.[51] Z. Lin, M. Lin, J.-B. Wang, T. de Cola, and J. Wang, “Joint beamform-ing and power allocation for satellite-terrestrial integrated networkswith non-orthogonal multiple access,”
IEEE Journal of Selected Topicsin Signal Processing , vol. 13, no. 3, pp. 657–670, Feb. 2019.[52] A. Guidotti, B. Evans, and M. Di Renzo, “Integrated satellite-terrestrialnetworks in future wireless systems,”
International Journal of SatelliteCommunications and Networking , vol. 37, no. 2, pp. 73–75, Mar. 2019.[53] X. Zhu, C. Jiang, L. Kuang, N. Ge, S. Guo, and J. Lu, “Cooperativetransmission in integrated terrestrial-satellite networks,”
IEEE network ,vol. 33, no. 3, pp. 204–210, Jan. 2019.[54] J. Li, K. Xue, D. S. Wei, J. Liu, and Y. Zhang, “Energy efficiencyand traffic offloading optimization in integrated satellite/terrestrial radioaccess networks,”
IEEE Transactions on Wireless Communications ,vol. 19, no. 4, pp. 2367–2381, Jan. 2020.[55] L. Song, Z. Han, and B. Di, “Aerial access networks for 6G: From UAV,HAP, to satellite communication networks,” in
IEEE InternationalConference on Communications (ICC) , Virtual Program, 2020, pp. 1–1.[56] N.-N. Dao, Q.-V. Pham, D.-T. Do, and S. Dustdar, “The sky is the edge–Toward mobile coverage from the sky,”
IEEE Internet Computing ,2021.[57] United States. Department of the Army, “’Eyes of the army’: U.S.army roadmap for unmanned systems, 2010–2035,” Apr. 2010.[58] M. A. Kishk, A. Bader, and M.-S. Alouini, “On the 3-D placementof airborne base stations using tethered UAVs,”
IEEE Transactions onCommunications , vol. 68, no. 8, pp. 5202–5215, Aug. 2020.[59] M. Kishk, A. Bader, and M.-S. Alouini, “Aerial base station deploymentin 6G cellular networks using tethered drones: The mobility andendurance tradeoff,”
IEEE Vehicular Technology Magazine , vol. 15,no. 4, pp. 103–111, Dec. 2020.[60] A. Alzidaneen, A. Alsharoa, and M.-S. Alouini, “Resource and place-ment optimization for multiple UAVs using backhaul tethered bal-loons,”
IEEE Wireless Communications Letters , vol. 9, no. 4, pp. 543–547, Apr. 2020.[61] J. Pelton, “High altitude platform systems (HAPS) and unmanned aerialvehicles (UAV) as an alternative to small satellites,”
Handbook of SmallSatellites: Technology, Design, Manufacture, Applications, Economicsand Regulation , pp. 1–16, 2020.[62] S. C. Arum, D. Grace, and P. D. Mitchell, “A review of wirelesscommunication using high-altitude platforms for extended coverageand capacity,”
Computer Communications , vol. 157, no. 1, pp. 232–256, May 2020.[63] Y. Su, Y. Liu, Y. Zhou, J. Yuan, H. Cao, and J. Shi, “Broadband LEOsatellite communications: Architectures and key technologies,”
IEEEWireless Communications , vol. 26, no. 2, pp. 55–61, Apr. 2019.[64] F. Davoli, C. Kourogiorgas, M. Marchese, A. Panagopoulos, andF. Patrone, “Small satellites and CubeSats: Survey of structures,architectures, and protocols,”
International Journal of Satellite Com-munications and Networking , vol. 37, no. 4, pp. 343–359, Jul. 2019.[65] 3GPP TS 23.501 V16.5.0 Release 16, “5G; System architecture for the5G system (5GS),” Jul. 2020.[66] ETSI TR 103 611 V1.1.1, “Satellite Earth stations and systems (SES);Seamless integration of satellite and/or HAPS (high altitude platformstation) systems into 5G and related architecture options,” Jun. 2020.[67] F. Rinaldi, H.-L. Maattanen, J. Torsner, S. Pizzi, S. Andreev, A. Iera,Y. Koucheryavy, and G. Araniti, “Non-terrestrial networks in 5g &beyond: A survey,”
IEEE Access , vol. 8, pp. 165 178–165 200, Sep.2020. [68] 3GPP TS 38.300 V16.2.0 Release 16, “5G; NR; NR and NG-RANoverall description; stage-2,” Jul. 2020.[69] W. Khawaja, I. Guvenc, D. W. Matolak, U.-C. Fiebig, and N. Schneck-enburger, “A survey of air-to-ground propagation channel modeling forunmanned aerial vehicles,” IEEE Communications Surveys & Tutorials ,vol. 21, no. 3, pp. 2361–2391, Thirdquarter 2019.[70] A. D. Panagopoulos, P.-D. M. Arapoglou, and P. G. Cottis, “Satellitecommunications at Ku, Ka, and V bands: Propagation impairments andmitigation techniques,”
IEEE Communications Surveys & Tutorials ,vol. 6, no. 3, pp. 2–14, Thirdquarter 2004.[71] F. Dovis, R. Fantini, M. Mondin, and P. Savi, “Small-scale fading forhigh-altitude platform (HAP) propagation channels,”
IEEE Journal onSelected areas in Communications , vol. 20, no. 3, pp. 641–647, Apr.2002.[72] S. Karapantazis and F. Pavlidou, “Broadband communications viahigh-altitude platforms: a survey,”
IEEE Communications Surveys &Tutorials , vol. 7, no. 1, pp. 2–31, Firstquarter 2005.[73] Q. Feng, J. McGeehan, E. K. Tameh, and A. R. Nix, “Path lossmodels for air-to-ground radio channels in urban environments,” in , vol. 6, Melbourne,Vic., Australia, 2006, pp. 2901–2905.[74] A. Al-Hourani, S. Kandeepan, and A. Jamalipour, “Modeling air-to-ground path loss for low altitude platforms in urban environments,”in ,Austin, TX, USA, 2014, pp. 2898–2904.[75] I. Bor-Yaliniz, S. S. Szyszkowicz, and H. Yanikomeroglu,“Environment-aware drone-base-station placements in modernmetropolitans,”
IEEE Wireless Communications Letters , vol. 7, no. 3,pp. 372–375, Jun. 2017.[76] A. Al-Hourani and K. Gomez, “Modeling cellular-to-UAV path-lossfor suburban environments,”
IEEE Wireless Communications Letters ,vol. 7, no. 1, pp. 82–85, 2018.[77] Z. Cui, C. Briso, K. Guan, D. W. Matolak, C. Calvo-Ram´ırez, B. Ai,and Z. Zhong, “Low-altitude UAV air-ground propagation channelmeasurement and analysis in a suburban environment at 3.9 GHz,”
IETMicrowaves, Antennas & Propagation , vol. 13, no. 9, pp. 1503–1508,Apr. 2019.[78] D. W. Matolak and R. Sun, “Air–ground channel characterizationfor unmanned aircraft systems—part I: Methods, measurements, andmodels for over-water settings,”
IEEE Transactions on Vehicular Tech-nology , vol. 66, no. 1, pp. 26–44, Jan. 2017.[79] R. Sun and D. W. Matolak, “Air–ground channel characterization forunmanned aircraft systems part II: Hilly and mountainous settings,”
IEEE Transactions on Vehicular Technology , vol. 66, no. 3, pp. 1913–1925, Mar. 2016.[80] D. W. Matolak and R. Sun, “Air–ground channel characterization forunmanned aircraft systems—part III: The suburban and near-urbanenvironments,”
IEEE Transactions on Vehicular Technology , vol. 66,no. 8, pp. 6607–6618, Aug. 2017.[81] R. Sun, D. W. Matolak, and W. Rayess, “Air-ground channel character-ization for unmanned aircraft systems—part IV: Airframe shadowing,”
IEEE Transactions on Vehicular Technology , vol. 66, no. 9, pp. 7643–7652, Sep. 2017.[82] D. Rieth, C. Heller, and G. Ascheid, “Aircraft to ground-station C-band channel—small airport scenario,”
IEEE Transactions on VehicularTechnology , vol. 68, no. 5, pp. 4306–4315, May 2019.[83] R. M. Gutierrez, H. Yu, Y. Rong, and D. W. Bliss, “Comparisonof UAS-to-ground small-scale fading in residential and mountainousdesert terrains,”
IEEE Transactions on Vehicular Technology , vol. 68,no. 10, pp. 9348–9358, Oct. 2019.[84] X. Cheng and Y. Li, “A 3-D geometry-based stochastic model for UAV-MIMO wideband nonstationary channels,”
IEEE Internet of ThingsJournal , vol. 6, no. 2, pp. 1654–1662, Apr. 2019.[85] H. Jiang, Z. Zhang, C.-X. Wang, J. Zhang, J. Dang, L. Wu, andH. Zhang, “A novel 3D UAV channel model for A2G communica-tion environments using AoD and AoA estimation algorithms,”
IEEETransactions on Communications , 2020, in press.[86] Z. Ma, B. Ai, R. He, G. Wang, Y. Niu, and Z. Zhong, “A wideband non-stationary air-to-air channel model for UAV communications,”
IEEETransactions on Vehicular Technology , vol. 69, no. 2, pp. 1214–1226,Feb. 2020.[87] J. K. Stolaroff, C. Samaras, E. R. O’Neill, A. Lubers, A. S. Mitchell,and D. Ceperley, “Energy use and life cycle greenhouse gas emissionsof drones for commercial package delivery,”
Nature communications ,vol. 9, no. 1, pp. 1–13, Feb. 2018. [88] C.-M. Tseng, C.-K. Chau, K. Elbassioni, and M. Khonji, “Autonomousrecharging and flight mission planning for battery-operated autonomousdrones,” arXiv preprint arXiv:1703.10049 , 2017.[89] A. Thibbotuwawa, P. Nielsen, B. Zbigniew, and G. Bocewicz, “Energyconsumption in unmanned aerial vehicles: a review of energy consump-tion models and their relation to the UAV routing,” in
InternationalConference on Information Systems Architecture and Technology , Nysa,Poland, 2018, pp. 173–184.[90] K. Dorling, J. Heinrichs, G. G. Messier, and S. Magierowski, “Vehiclerouting problems for drone delivery,”
IEEE Transactions on Systems,Man, and Cybernetics: Systems , vol. 47, no. 1, pp. 70–85, Jan. 2017.[91] G. J. Leishman,
Principles of helicopter aerodynamics . Cambridgeuniversity press, 2006.[92] J. L. Marins, T. M. Cabreira, K. S. Kappel, and P. R. Ferreira, “Aclosed-form energy model for multi-rotors based on the dynamic ofthe movement,” in
VIII Brazilian Symposium on Computing SystemsEngineering (SBESC) , Salvador, Brazil, 2018, pp. 256–261.[93] N. Michel, A. K. Sinha, Z. Kong, and X. Lin, “Multiphysical modelingof energy dynamics for multirotor unmanned aerial vehicles,” in
Inter-national Conference on Unmanned Aircraft Systems (ICUAS) , Atlanta,GA, USA, 2019, pp. 738–747.[94] R. Austin,
Unmanned aircraft systems: UAVs design, development anddeployment . John Wiley & Sons, 2011, vol. 54.[95] Y. Zeng and R. Zhang, “Energy-efficient UAV communication with tra-jectory optimization,”
IEEE Transactions on Wireless Communications ,vol. 16, no. 6, pp. 3747–3760, Jun. 2017.[96] Y. Zeng, J. Xu, and R. Zhang, “Energy minimization for wirelesscommunication with rotary-wing UAV,”
IEEE Transactions on WirelessCommunications , vol. 18, no. 4, pp. 2329–2345, Apr. 2019.[97] M. Horani and M. O. Hasna, “Latency analysis of UAV based com-munication networks,” in
International Conference on Information andCommunication Technology Convergence (ICTC) , Jeju, South Korea,2018, pp. 385–390.[98] K.-M. Liao, G.-Y. Chen, Y.-J. Chen, and Y.-F. Chen, “End-to-enddelay analysis in mmWave UAV-assisted wireless caching networks,”in , Seoul, Korea, 2020, pp. 1–6.[99] X. Wu, Q. Li, Y. Lu, H. V. Poor, V. C. Leung, and P. Ching, “Latency-minimized design of secure transmissions in UAV-aided communica-tions,” in
IEEE International Conference on Acoustics, Speech andSignal Processing (ICASSP) , Barcelona, Spain, 2020, pp. 8753–8757.[100] C. She, C. Liu, T. Q. Quek, C. Yang, and Y. Li, “Ultra-reliable and low-latency communications in unmanned aerial vehicle communicationsystems,”
IEEE Transactions on Communications , vol. 67, no. 5, pp.3768–3781, May 2019.[101] H. Ren, C. Pan, K. Wang, Y. Deng, M. Elkashlan, and A. Nallanathan,“Achievable data rate for URLLC-enabled UAV systems with 3-Dchannel model,”
IEEE Wireless Communications Letters , vol. 8, no. 6,pp. 1587–1590, Dec. 2019.[102] A. Ranjha and G. Kaddoum, “Quasi-optimization of distance andblocklength in URLLC aided multi-hop UAV relay links,”
IEEEWireless Communications Letters , vol. 9, no. 3, pp. 306–310, Mar.2020.[103] P. Bellavista, D. Belli, S. Chessa, and L. Foschini, “A social-drivenedge computing architecture for mobile crowd sensing management,”
IEEE Communications Magazine , vol. 57, no. 4, pp. 68–73, Apr. 2019.[104] P. Bellavista, S. Chessa, L. Foschini, L. Gioia, and M. Girolami,“Human-enabled edge computing: Exploiting the crowd as a dynamicextension of mobile edge computing,”
IEEE Communications Maga-zine , vol. 56, no. 1, pp. 145–155, Jan. 2018.[105] Q.-V. Pham, L. B. Le, S.-H. Chung, and W.-J. Hwang, “Mobile edgecomputing with wireless backhaul: Joint task offloading and resourceallocation,”
IEEE Access , vol. 7, pp. 16 444–16 459, 2019.[106] Q.-V. Pham, H. T. Nguyen, Z. Han, and W.-J. Hwang, “Coalitionalgames for computation offloading in NOMA-enabled multi-access edgecomputing,”
IEEE Transactions on Vehicular Technology , vol. 69, no. 2,pp. 1982–1993, Feb. 2020.[107] X. Hu, K.-K. Wong, K. Yang, and Z. Zheng, “UAV-assisted relayingand edge computing: Scheduling and trajectory optimization,”
IEEETransactions on Wireless Communications , vol. 18, no. 10, pp. 4738–4752, Oct. 2019.[108] R. Han, Y. Wen, L. Bai, J. Liu, and J. Choi, “Rate splitting on mobileedge computing for UAV-aided IoT systems,”
IEEE Transactions onCognitive Communications and Networking , 2020, in press.[109] L. Zhang and N. Ansari, “Latency-aware IoT service provisioning inUAV-aided mobile edge computing networks,”
IEEE Internet of ThingsJournal , 2020, in press. [110] Z. Zhang, W. Zhang, and F.-H. Tseng, “Satellite mobile edge comput-ing: Improving QoS of high-speed satellite-terrestrial networks usingedge computing techniques,” IEEE network , vol. 33, no. 1, pp. 70–76,Jan./Feb. 2019.[111] M. Mozaffari, W. Saad, M. Bennis, and M. Debbah, “Mobile unmannedaerial vehicles (UAVs) for energy-efficient internet of things commu-nications,”
IEEE Transactions on Wireless Communications , vol. 16,no. 11, pp. 7574–7589, Nov. 2017.[112] V. V. Chetlur and H. S. Dhillon, “Downlink coverage analysis fora finite 3-D wireless network of unmanned aerial vehicles,”
IEEETransactions on Communications , vol. 65, no. 10, pp. 4543–4558, Oct.2017.[113] J. Lyu, Y. Zeng, R. Zhang, and T. J. Lim, “Placement optimization ofUAV-mounted mobile base stations,”
IEEE Communications Letters ,vol. 21, no. 3, pp. 604–607, Mar. 2017.[114] J. Chen and D. Gesbert, “Efficient local map search algorithms for theplacement of flying relays,”
IEEE Transactions on Wireless Communi-cations , vol. 19, no. 2, pp. 1305–1319, Feb. 2020.[115] P. K. Sharma and D. I. Kim, “Random 3D mobile UAV networks:Mobility modeling and coverage probability,”
IEEE Transactions onWireless Communications , vol. 18, no. 5, pp. 2527–2538, May 2019.[116] P. K. Sharma, D. Deepthi, and D. I. Kim, “Outage probability of 3-Dmobile UAV relaying for hybrid satellite-terrestrial networks,”
IEEECommunications Letters , vol. 24, no. 2, pp. 418–422, Feb. 2020.[117] P. K. Sharma and D. I. Kim, “Secure 3D mobile UAV relaying forhybrid satellite-terrestrial networks,”
IEEE Transactions on WirelessCommunications , vol. 19, no. 4, pp. 2770–2784, Apr. 2020.[118] Z. Qu, G. Zhang, H. Cao, and J. Xie, “LEO satellite constellation forInternet of things,”
IEEE Access , vol. 5, pp. 18 391–18 401, 2017.[119] A. Papa, T. de Cola, P. Vizarreta, M. He, C. Mas-Machuca, andW. Kellerer, “Design and evaluation of reconfigurable SDN LEO con-stellations,”
IEEE Transactions on Network and Service Management ,May 2020.[120] I. Del Portillo, B. G. Cameron, and E. F. Crawley, “A technicalcomparison of three low earth orbit satellite constellation systems toprovide global broadband,”
Acta Astronautica , vol. 159, pp. 123–135,Jun. 2019.[121] C. You and R. Zhang, “3D trajectory optimization in Rician fadingfor UAV-enabled data harvesting,”
IEEE Transactions on WirelessCommunications , vol. 18, no. 6, pp. 3192–3207, Jun. 2019.[122] H. Hu, K. Xiong, G. Qu, Q. Ni, P. Fan, and K. B. Letaief, “AoI-minimaltrajectory planning and data collection in UAV-assisted wireless pow-ered IoT networks,”
IEEE Internet of Things Journal , 2020, in press.[123] X. Liu, Y. Liu, Y. Chen, and L. Hanzo, “Trajectory design and powercontrol for multi-UAV assisted wireless networks: A machine learningapproach,”
IEEE Transactions on Vehicular Technology , vol. 68, no. 8,pp. 7957–7969, Aug. 2019.[124] Q. Liu, L. Shi, L. Sun, J. Li, M. Ding, and F. Shu, “Path planningfor UAV-mounted mobile edge computing with deep reinforcementlearning,”
IEEE Transactions on Vehicular Technology , vol. 69, no. 5,pp. 5723–5728, May 2020.[125] L. Xu, M. Chen, M. Chen, Z. Yang, C. Chaccour, W. Saad, and C. S.Hong, “Joint location and power optimization for THz-enabled UAVcommunications,” arXiv preprint arXiv:2011.01065 , 2020.[126] W. Youn, H. Choi, A. Cho, S. Kim, and M. B. Rhudy, “AerodynamicModel-Aided estimation of Attitude, 3D Wind, Airspeed, AOA, andSSA for High-Altitude Long-Endurance UAV,”
IEEE Transactions onAerospace and Electronic Systems , Apr. 2020.[127] I. Kantor, A. N. Srivastava, D. M. Pasko, H. Batla, and G. Ubhi,“Unmanned aerial vehicle network-based recharging,” Aug. 9 2016,US Patent 9,412,279.[128] V. Sharma, K. Srinivasan, H.-C. Chao, K.-L. Hua, and W.-H. Cheng,“Intelligent deployment of UAVs in 5G heterogeneous communicationenvironment for improved coverage,”
Journal of Network and Com-puter Applications , vol. 85, pp. 94–105, May 2017.[129] P. Dunphy and F. A. Petitcolas, “A first look at identity managementschemes on the blockchain,”
IEEE Security & Privacy , vol. 16, no. 4,pp. 20–29, Aug. 2018.[130] R. Rosa and C. E. Rothenberg, “Blockchain-based decentralized ap-plications for multiple administrative domain networking,”
IEEE Com-munications Standards Magazine , vol. 2, no. 3, pp. 29–37, Sep. 2018.[131] T. Alladi, V. Chamola, N. Sahu, and M. Guizani, “Applications ofblockchain in unmanned aerial vehicles: A review,”
Vehicular Commu-nications , p. 100249, Jan. 2020.[132] V. Hassija, V. Chamola, D. N. G. Krishna, and M. Guizani, “Adistributed framework for energy trading between UAVs and charging stations for critical applications,”
IEEE Transactions on VehicularTechnology , vol. 69, no. 5, pp. 5391–5402, Mar. 2020.[133] S. Popov, “The tangle,” cit. on
IEEE Transactions on Industrial Informatics , vol. 15,no. 6, pp. 3680–3689, Mar. 2019.[135] E. Dicembrini, M. Scanavino, F. Dabbene, and G. Guglieri, “Modellingand simulation of a tethered UAS,” in . IEEE, Oct. 2020, pp. 1801–1808.[136] O. M. Bushnaq, M. A. Kishk, A. C¸ elik, M.-S. Alouini, and T. Y.Al-Naffouri, “Optimal deployment of tethered drones for maximumcellular coverage in user clusters,”
IEEE Transactions on WirelessCommunications
IEEE Access , vol. 5, pp. 21 264–21 285,2017.[146] X. Lu, P. Wang, D. Niyato, D. I. Kim, and Z. Han, “Wireless charg-ing technologies: Fundamentals, standards, and network applications,”
IEEE Communications Surveys & Tutorials , vol. 18, no. 2, pp. 1413–1452, Secondquarter 2016.[147] H. H. Wu, A. Gilchrist, K. D. Sealy, and D. Bronson, “A highefficiency 5 kW inductive charger for EVs using dual side control,”
IEEE Transactions on Industrial Informatics , vol. 8, no. 3, pp. 585–595, Aug. 2012.[148] T. Campi, S. Cruciani, F. Maradei, and M. Feliziani, “Wireless chargingsystem integrated in a small unmanned aerial vehicle (UAV) withhigh tolerance to planar coil misalignment,” in . Sapporo, Japan: IEEE, 2019, pp. 601–604.[149] S. Obayashi, Y. Kanekiyo, K. Nishizawa, and H. Kusada, “85-kHzband 450-W Inductive Power Transfer for Unmanned Aerial VehicleWireless Charging Port,” in , London, United Kingdom, 2019, pp. 80–84.[150] C. Wang and Z. Ma, “Design of wireless power transfer device forUAV,” in , Harbin, China, 2016, pp. 2449–2454.[151] L. Qiong, W. Huayun, W. Wenbin, M. Tianqi, and W. Yongyue,“Optimal Design of Magnetic Resonance Wireless Charging Coil forUnmanned Aerial Vehicles,” in . Xiamen, China,China: IEEE, 2018, pp. 469–472.[152] S. Sekander, H. Tabassum, and E. Hossain, “Statistical performancemodeling of solar and wind-powered UAV communications,”
IEEETransactions on Mobile Computing , pp. 1–1, Apr. 2020.[153] J. Zhang, M. Lou, L. Xiang, and L. Hu, “Power cognition: Enablingintelligent energy harvesting and resource allocation for solar-poweredUAVs,”
Future Generation Computer Systems , vol. 110, pp. 658–664,Sep. 2020.[154] S. Jashnani, T. Nada, M. Ishfaq, A. Khamker, and P. Shaholia, “Sizingand preliminary hardware testing of solar powered UAV,”
The EgyptianJournal of Remote Sensing and Space Science , vol. 16, no. 2, pp. 189–198, Dec. 2013. [155] S. Morton, R. D’Sa, and N. Papanikolopoulos, “Solar powered UAV:Design and experiments,” in , Hamburg, Germany, 2015,pp. 2460–2466.[156] P. Oettershagen, A. Melzer, T. Mantel, K. Rudin, T. Stastny,B. Wawrzacz, T. Hinzmann, K. Alexis, and R. Siegwart, “Perpetualflight with a small solar-powered UAV: Flight results, performanceanalysis and model validation,” in ,Big Sky, MT, USA, 2016, pp. 1–8.[157] V. S. Dwivedi, J. Patrikar, A. Addamane, and A. Ghosh, “MARAAL:A low altitude long endurance solar powered UAV for surveillanceand mapping applications,” in . Miedzyzdroje,Poland: IEEE, 2018, pp. 449–454.[158] H. Wang and J. Shen, “Analysis of the characteristics of solar cell arraybased on MATLAB/Simulink in solar unmanned aerial vehicle,” IEEEAccess , vol. 6, pp. 21 195–21 201, Apr. 2018.[159] P. Rajendran and H. Smith, “Experimental study of solar module &maximum power point tracking system under controlled temperatureconditions,”
Internation Journal on Advanced Science EngineeringInformation Technology , vol. 8, no. 4, pp. 1147–1153, 2018.[160] S. Hosseini, R. Dai, and M. Mesbahi, “Optimal path planning andpower allocation for a long endurance solar-powered UAV,” in . Washington, DC, USA: IEEE, 2013,pp. 2588–2593.[161] Y. Sun, D. W. K. Ng, D. Xu, L. Dai, and R. Schober, “Resourceallocation for solar powered UAV communication systems,” in , Kalamata, Greece, 2018, pp. 1–5.[162] Y. Sun, D. Xu, D. W. K. Ng, L. Dai, and R. Schober, “Optimal 3D-trajectory design and resource allocation for solar-powered UAV com-munication systems,”
IEEE Transactions on Communications , vol. 67,no. 6, pp. 4281–4298, Feb. 2019.[163] P. Wang, J. L. Rosales, and V. A. Chiriac, “Multi-rotor aerial drone withthermal energy harvesting,” Feb. 28 2019, US Patent App. 15/689,994.[164] V. Bonnin, E. B´enard, J.-M. Moschetta, and C. Toomer, “Energy-harvesting mechanisms for UAV flight by dynamic soaring,”
Interna-tional Journal of Micro Air Vehicles , vol. 7, no. 3, pp. 213–229, Sep.2015.[165] S. R. Anton and D. J. Inman, “Vibration energy harvesting forunmanned aerial vehicles,” in
Active and Passive Smart Structures andIntegrated Systems 2008 , vol. 6928. San Diego, California, UnitedStates: International Society for Optics and Photonics, 2008, p. 692824.[166] A. Koszewnik and D. Oldziej, “Performance assessment of an energyharvesting system located on a copter,”
The European Physical JournalSpecial Topics , vol. 228, no. 7, pp. 1677–1692, Aug. 2019.[167] S. R. Anton and D. J. Inman, “Performance modeling of unmannedaerial vehicles with on-board energy harvesting,” in
Active and PassiveSmart Structures and Integrated Systems 2011 , vol. 7977. Diego, Cal-ifornia, United States: International Society for Optics and Photonics,2011, p. 79771H.[168] D. Belo, D. C. Ribeiro, P. Pinho, and N. B. Carvalho, “A selective,tracking, and power adaptive far-field wireless power transfer system,”
IEEE Transactions on Microwave Theory and Techniques , vol. 67,no. 9, pp. 3856–3866, Sep. 2019.[169] V. Kuhn, C. Lahuec, F. Seguin, and C. Person, “A multi-band stackedRF energy harvester with RF-to-DC efficiency up to 84%,”
IEEETransactions on Microwave Theory and Techniques , vol. 63, no. 5,pp. 1768–1778, May 2015.[170] S. Dunbar, F. Wenzl, C. Hack, R. Hafeza, H. Esfeer, F. Defay,S. Prothin, D. Bajon, and Z. Popovic, “Wireless far-field chargingof a micro-UAV,” in , Boulder, CO, USA, 2015, pp. 1–4.[171] A. Muharam, T. M. Mostafa, and R. Hattori, “Design of powerreceiving side in wireless charging system for UAV application,” in . Jakarta, Indonesia: IEEE, 2017, pp. 133–139.[172] T. Long, M. Ozger, O. Cetinkaya, and O. B. Akan, “Energy neutralinternet of drones,”
IEEE Communications Magazine , vol. 56, no. 1,pp. 22–28, Jan. 2018.[173] W. Chen, S. Zhao, Q. Shi, and R. Zhang, “Resonant Beam Charging-Powered UAV-Assisted Sensing Data Collection,”
IEEE Transactionson Vehicular Technology , vol. 69, no. 1, pp. 1086–1090, Oct. 2019.[174] M.-A. Lahmeri, M. A. Kishk, and M.-S. Alouini, “Stochastic geometry-based analysis of airborne base stations with laser-powered UAVs,”
IEEE Communications Letters , vol. 24, no. 1, pp. 173–177, Feb. 2020.[175] J. Ouyang, Y. Che, J. Xu, and K. Wu, “Throughput maximizationfor laser-powered UAV wireless communication systems,” in . Kansas City, MO, USA: IEEE, 2018, pp. 1–6.[176] Q. Chen, D. Zhang, D. Zhu, Q. Shi, J. Gu, and Y. Ai, “Design andexperiment for realization of laser wireless power transmission forsmall unmanned aerial vehicles,” in
AOPC 2015: Advances in LaserTechnology and Applications , vol. 9671. Beijing, China: InternationalSociety for Optics and Photonics, 2015, p. 96710N.[177] S. Marano and P. Willett, “Resource allocation in energy-harvestingsensor networks,”
IEEE Transactions on Signal and Information Pro-cessing over Networks , vol. 4, no. 3, pp. 585–598, Jan. 2018.[178] J. Xu, Y. Zeng, and R. Zhang, “UAV-enabled wireless power transfer:Trajectory design and energy optimization,”
IEEE Transactions onWireless Communications , vol. 17, no. 8, pp. 5092–5106, May 2018.[179] C. Su, F. Ye, L.-C. Wang, L. Wang, Y. Tian, and Z. Han, “UAV-assistedwireless charging for energy-constrained IoT devices using dynamicmatching,”
IEEE Internet of Things Journal , Jan. 2020.[180] H. Wang, J. Wang, G. Ding, L. Wang, T. A. Tsiftsis, and P. K. Sharma,“Resource allocation for energy harvesting-powered D2D communica-tion underlaying UAV-assisted networks,”
IEEE Transactions on GreenCommunications and Networking , vol. 2, no. 1, pp. 14–24, Oct. 2018.[181] T. Chen, M. Matinmikko, X. Chen, X. Zhou, and P. Ahokangas,“Software defined mobile networks: concept, survey, and researchdirections,”
IEEE Communications Magazine , vol. 53, no. 11, pp. 126–133, Nov. 2015.[182] L. Mamushiane, A. Lysko, and S. Dlamini, “A comparative evaluationof the performance of popular SDN controllers,” in . Dubai, United Arab Emirates: IEEE, 2018, pp. 54–59.[183] O. S. Oubbati, M. Atiquzzaman, T. A. Ahanger, and A. Ibrahim,“Softwarization of UAV networks: A survey of applications and futuretrends,”
IEEE Access , vol. 8, pp. 98 073–98 125, May 2020.[184] Z. Zhao, P. Cumino, A. Souza, D. Rosario, T. Braun, E. Cerqueira, andM. Gerla, “Software-defined unmanned aerial vehicles networking forvideo dissemination services,”
Ad Hoc Networks , vol. 83, pp. 68–77,Feb. 2019.[185] X. Zhang, H. Wang, and H. Zhao, “An SDN framework for UAVbackbone network towards knowledge centric networking,” in
IEEEConference on Computer Communications Workshops (INFOCOMWKSHPS) , Honolulu, HI, USA, Apr. 2018, pp. 456–461.[186] F. Xiong, A. Li, H. Wang, and L. Tang, “An SDN-MQTT based com-munication system for battlefield UAV swarms,”
IEEE CommunicationsMagazine , vol. 57, no. 8, pp. 41–47, Aug. 2019.[187] A. Hermosilla, A. M. Zarca, J. B. Bernabe, J. Ortiz, and A. Skarmeta,“Security orchestration and enforcement in NFV/SDN-aware UAVdeployments,”
IEEE Access , vol. 8, pp. 131 779–131 795, Jul. 2020.[188] B. Barritt and V. Cerf, “Loon SDN: Applicability to NASA’snext-generation space communications architecture,” in , Big Sky, MT, USA, Mar. 2018, pp. 1–9.[189] A. Papa, T. De Cola, P. Vizarreta, M. He, C. M. Machuca, andW. Kellerer, “Dynamic SDN controller placement in a LEO constel-lation satellite network,” in
IEEE Global Communications Conference(GLOBECOM) , Abu Dhabi, United Arab Emirates, United Arab Emi-rates, Dec. 2018, pp. 206–212.[190] Z. Jia, M. Sheng, J. Li, Y. Zhu, W. Bai, and Z. Han, “Virtual networkfunctions orchestration in software defined LEO small satellite net-works,” in
IEEE International Conference on Communications (ICC) ,Dublin, Ireland, Jun. 2020, pp. 1–6.[191] A. Abbas, S. U. Khan, and A. Y. Zomaya,
Fog Computing: Theory andPractice . John Wiley & Sons, 2020.[192] N. Mohamed, J. Al-Jaroodi, I. Jawhar, H. Noura, and S. Mahmoud,“UAVFog: A UAV-based fog computing for Internet of Things,” in , SanFrancisco, CA, USA, 2017, pp. 1–8.[193] M.-A. Messous, H. Sedjelmaci, N. Houari, and S.-M. Senouci, “Com-putation offloading game for an UAV network in mobile edge com-puting,” in , Paris, France, 2017, pp. 1–6.[194] M. F. Pinto, A. L. Marcato, A. G. Melo, L. M. Hon´orio, and C. Ur-diales, “A framework for analyzing fog-cloud computing cooperationapplied to information processing of UAVs,”
Wireless Communicationsand Mobile Computing , vol. 2019, 2019.[195] B. Liu, W. Zhang, W. Chen, H. Huang, and S. Guo, “Online Compu-tation Offloading and Traffic Routing for UAV Swarms in Edge-CloudComputing,”
IEEE Transactions on Vehicular Technology , May 2020. [196] X. Diao, J. Zheng, Y. Cai, Y. Wu, and A. Anpalagan, “Fair dataallocation and trajectory optimization for UAV-assisted mobile edgecomputing,” IEEE Communications Letters , vol. 23, no. 12, pp. 2357–2361, Dec. 2019.[197] B. Yang, X. Cao, C. Yuen, and L. Qian, “Offloading optimization inedge computing for deep learning enabled target tracking by internet-of-UAVs,”
IEEE Internet of Things Journal , Aug. 2020.[198] M. Ilchenko, T. Narytnik, V. Prisyazhny, S. Kapshtyk, andS. Matvienko, “The solution of the problem of the delay determinationin the information transmission and processing in the LEO satelliteinternet of things system,” in . Kyiv, Ukraine, Ukraine: IEEE, Oct. 2019,pp. 419–425.[199] C. Qiu, H. Yao, F. R. Yu, F. Xu, and C. Zhao, “Deep Q-learning aidednetworking, caching, and computing resources allocation in software-defined satellite-terrestrial networks,”
IEEE Transactions on VehicularTechnology , vol. 68, no. 6, pp. 5871–5883, Jun. 2019.[200] L. Qian, J. Zhu, and S. Zhang, “Survey of Wireless Big Data,”
Journalof Communications and Information Networks , vol. 2, no. 1, pp. 1–18,Mar. 2017.[201] C. Zhang, P. Patras, and H. Haddadi, “Deep learning in mobile andwireless networking: A survey,”
IEEE Communications Surveys &Tutorials , vol. 21, no. 3, pp. 2224–2287, Mar. 2019.[202] J. Han, J. Pei, and M. Kamber,
Data mining: concepts and techniques .Elsevier, 2011.[203] P. S. Bithas, E. T. Michailidis, N. Nomikos, D. Vouyioukas, and A. G.Kanatas, “A survey on machine-learning techniques for UAV-basedcommunications,”
Sensors , vol. 19, no. 23, p. 5170, Nov. 2019.[204] A. Carrio, C. Sampedro, A. Rodriguez-Ramos, and P. Campoy, “Areview of deep learning methods and applications for unmanned aerialvehicles,”
Journal of Sensors , vol. 2017, Aug. 2017.[205] N. C. Luong, D. T. Hoang, S. Gong, D. Niyato, P. Wang, Y.-C.Liang, and D. I. Kim, “Applications of deep reinforcement learningin communications and networking: A survey,”
IEEE CommunicationsSurveys & Tutorials , vol. 21, no. 4, pp. 3133–3174, Fourthquarter 2019.[206] M. Mukherjee, V. Kumar, A. Lat, M. Guo, R. Matam, and Y. Lv,“Distributed Deep Learning-based Task Offloading for UAV-enabledMobile Edge Computing,” in
IEEE INFOCOM 2020-IEEE Confer-ence on Computer Communications Workshops (INFOCOM WKSHPS) ,Toronto, ON, Canada, 2020, pp. 1208–1212.[207] M. Mozaffari, W. Saad, M. Bennis, Y.-H. Nam, and M. Debbah, “Atutorial on UAVs for wireless networks: Applications, challenges, andopen problems,”
IEEE Communications Surveys & Tutorials , vol. 21,no. 3, pp. 2334–2360, Thirdquarter 2019.[208] R. Shakeri, M. A. Al-Garadi, A. Badawy, A. Mohamed, T. Khattab,A. K. Al-Ali, K. A. Harras, and M. Guizani, “Design challenges ofmulti-UAV systems in cyber-physical applications: A comprehensivesurvey and future directions,”
IEEE Communications Surveys & Tuto-rials , vol. 21, no. 4, pp. 3340–3385, Fourthquarter 2019.[209] P. V. R. Ferreira, R. Paffenroth, A. M. Wyglinski, T. M. Hackett, S. G.Bilen, R. C. Reinhart, and D. J. Mortensen, “Reinforcement learning forsatellite communications: from LEO to deep space operations,”
IEEECommunications Magazine , vol. 57, no. 5, pp. 70–75, May 2019.[210] M. Werth, J. Lucas, T. Kyono, I. McQuaid, and J. Fletcher, “SILO:A machine learning dataset of synthetic ground-based observations ofLEO satellites,” in . Big Sky, MT,USA: IEEE, Mar. 2020, pp. 1–8.[211] H. Chen, V. Chandrasekar, R. Cifelli, and P. Xie, “A machine learningsystem for precipitation estimation using satellite and ground radarnetwork observations,”
IEEE Transactions on Geoscience and RemoteSensing , vol. 58, no. 2, pp. 982–994, Feb. 2020.[212] N. Kato, Z. M. Fadlullah, F. Tang, B. Mao, S. Tani, A. Okamura, andJ. Liu, “Optimizing space-air-ground integrated networks by artificialintelligence,”
IEEE Wireless Communications , vol. 26, no. 4, pp. 140–147, Aug. 2019.[213] S. Parkvall, E. Dahlman, A. Furuskar, and M. Frenne, “NR: Thenew 5G radio access technology,”
IEEE Communications StandardsMagazine , vol. 1, no. 4, pp. 24–30, Dec. 2017.[214] P. Zhou, X. Fang, Y. Fang, R. He, Y. Long, and G. Huang, “Beammanagement and self-healing for mmWave UAV mesh networks,”
IEEETransactions on Vehicular Technology , vol. 68, no. 2, pp. 1718–1732,Feb. 2018.[215] J. Zhao and W. Jia, “Efficient channel tracking strategy for mmWaveUAV communications,”
Electronics Letters , vol. 54, no. 21, pp. 1218–1220, Oct. 2018. [216] A. Mudonhi, C. Sacchi, and F. Granelli, “SDN-based multimediacontent delivery in 5G MmWave hybrid satellite-terrestrial networks,”in . Bologna, Italy:IEEE, 2018, pp. 1–7.[217] X. Artiga, A. P´erez-Neira, J. Baranda, E. Lagunas, S. Chatzinotas,R. Zetik, P. Gorski, K. Ntougias, D. Perez, and G. Ziaragkas, “Sharedaccess satellite-terrestrial reconfigurable backhaul network enabled bysmart antennas at mmwave band,”
IEEE Network , vol. 32, no. 5, pp.46–53, Sep./Oct. 2018.[218] K. Tekbıyık, G. K. Kurt, A. R. Ekti, A. G¨orc¸in, and H. Yanikomeroglu,“Reconfigurable intelligent surface empowered terahertz communica-tion for LEO satellite networks,” arXiv preprint arXiv:2007.04281 , Jul.2020.[219] Z. Wang, R. Malaney, and J. Green, “Inter-satellite quantum key distri-bution at terahertz frequencies,” in
ICC 2019-2019 IEEE InternationalConference on Communications (ICC) . Shanghai, China: IEEE, 2019,pp. 1–7.[220] R. Mendrzik, D. Cabric, and G. Bauch, “Error bounds for TerahertzMIMO positioning of swarm UAVs for distributed sensing,” in . Kansas City, MO, USA: IEEE, 2018, pp. 1–6.[221] Z. Chu, W. Hao, P. Xiao, and J. Shi, “UAV Assisted SpectrumSharing Ultra-Reliable and Low-Latency Communications,” in . Waikoloa,HI, USA: IEEE, Dec. 2019, pp. 1–6.[222] A. Han, T. Lv, and X. Zhang, “UAV beamwidth design for Ultra-Reliable and Low-Latency Communications with NOMA,” in . Shanghai, China: IEEE, 2019, pp. 1–6.[223] W. Cho and J. P. Choi, “Cross layer optimization of wireless controllinks in the software-defined LEO satellite network,”
IEEE Access ,vol. 7, pp. 113 534–113 547, Aug 2019.[224] I. Leyva-Mayorga, B. Soret, M. R¨oper, D. W¨ubben, B. Matthiesen,A. Dekorsy, and P. Popovski, “LEO small-satellite constellations for5G and beyond-5G communications,”
Ieee Access , vol. 8, pp. 184 955–184 964, Oct. 2020.[225] F. Jameel, M. A. A. Haider, A. A. Butt et al. , “Massive MIMO:A survey of recent advances, research issues and future directions,”in . Islamabad, Pakistan: IEEE, 2017, pp. 1–6.[226] A. F. Molisch, V. V. Ratnam, S. Han, Z. Li, S. L. H. Nguyen, L. Li,and K. Haneda, “Hybrid beamforming for massive MIMO: A survey,”
IEEE Communications Magazine , vol. 55, no. 9, pp. 134–141, Sep.2017.[227] P. Chandhar, D. Danev, and E. G. Larsson, “Massive MIMO forcommunications with drone swarms,”
IEEE Transactions on WirelessCommunications , vol. 17, no. 3, pp. 1604–1629, Mar. 2017.[228] G. Geraci, A. Garcia-Rodriguez, L. G. Giordano, D. L´opez-P´erez,and E. Bj¨ornson, “Supporting UAV cellular communications throughmassive MIMO,” in . Kansas City, MO, USA: IEEE,2018, pp. 1–6.[229] C. D’Andrea, A. Garcia-Rodriguez, G. Geraci, L. G. Giordano, andS. Buzzi, “Cell-free massive MIMO for UAV communications,” in . Shanghai, China: IEEE, 2019, pp. 1–6.[230] L. You, K. Li, J. Wang, X. Gao, X. Xia, and B. Otterstenx, “LEOsatellite communications with Massive MIMO,” in
ICC 2020 - 2020IEEE International Conference on Communications (ICC) , Dublin,Ireland, 2020, pp. 1–6.[231] K.-X. Li, L. You, J. Wang, X. Gao, G.Tsinos, S. Chatzinotas, andB. Ottersten, “Downlink transmit design in massive MIMO LEO satel-lite communications,” arXiv preprint arXiv:2008.05343 , Aug. 2020.[232] B. Makki, K. Chitti, A. Behravan, and M. Alouini, “A survey ofNOMA: Current status and open research challenges,”
IEEE OpenJournal of the Communications Society , vol. 1, pp. 179–189, Jan. 2020.[233] F. Fang, Y. Xu, Q.-V. Pham, and Z. Ding, “Energy-efficient design ofIRS-NOMA networks,”
IEEE Transactions on Vehicular Technology ,vol. 69, no. 11, pp. 14 088–14 092, Nov. 2020.[234] J. Chu, X. Chen, C. Zhong, and Z. Zhang, “Robust design for NOMA-based Multi-Beam LEO satellite Internet of Things,”
IEEE Internet ofThings Journal , 2020, in press.[235] Z. Gao, A. Liu, and X. Liang, “The performance analysis of downlinkNOMA in LEO satellite communication system,”
IEEE Access , vol. 8,pp. 93 723–93 732, May 2020. [236] A. A. Nasir, H. D. Tuan, T. Q. Duong, and H. V. Poor, “UAV-enabledcommunication using NOMA,” IEEE Transactions on Communica-tions , vol. 67, no. 7, pp. 5126–5138, Jul. 2019.[237] P. Botsinis, D. Alanis, C. Xu, Z. Babar, D. Chandra, S. X. Ng, andL. Hanzo, “Air-to-ground NOMA systems for the “Internet-Above-the-Clouds”,”
IEEE Access , vol. 6, pp. 47 442–47 460, Aug. 2018.[238] T. Hou, Y. Liu, Z. Song, X. Sun, and Y. Chen, “Multiple antennaaided NOMA in UAV networks: A stochastic geometry approach,”
IEEE Transactions on Communications , vol. 67, no. 2, pp. 1031–1044,Feb. 2019.[239] Q.-V. Pham, T. Huynh-The, M. Alazab, J. Zhao, and W.-J. Hwang,“Sum-rate maximization for UAV-assisted visible light communicationsusing NOMA: Swarm intelligence meets machine learning,”
IEEEInternet of Things Journal , vol. 7, no. 10, pp. 10 375–10 387, Oct.2020.[240] S. Gong, X. Lu, D. T. Hoang, D. Niyato, L. Shu, D. I. Kim, andY. Liang, “Towards smart wireless communications via intelligentreflecting surfaces: A contemporary survey,”
IEEE CommunicationsSurveys Tutorials , pp. 1–33, Jun. 2020.[241] B. Matthiesen, E. Bj¨ornson, E. De Carvalho, and P. Popovski, “Intelli-gent reflecting surface operation under predictable receiver mobility: Acontinuous time propagation model,” arXiv preprint arXiv:2006.06991 ,Aug. 2020.[242] M. Hua, L. Yang, Q. Wu, C. Pan, C. Li, and A. L. Swindlehurst,“UAV-assisted intelligent reflecting surface symbiotic radio system,” arXiv preprint arXiv:2007.14029 , Aug. 2020.[243] L. Ge, P. Dong, H. Zhang, J.-B. Wang, and X. You, “Joint beamformingand trajectory optimization for intelligent reflecting surfaces-assistedUAV communications,”
IEEE Access , vol. 8, pp. 78 702–78 712, Apr.2020.[244] Q. Zhang, W. Saad, and M. Bennis, “Reflections in the sky: Millimeterwave communication with UAV-carried intelligent reflectors,” in . IEEE, Dec.2019, pp. 1–6.[245] P. Bupe, R. Haddad, and F. Rios-Gutierrez, “Relief and emergencycommunication network based on an autonomous decentralized UAVclustering network,” in
SoutheastCon 2015 . Fort Lauderdale, FL,USA: IEEE, 2015, pp. 1–8.[246] C. Gomez and H. Purdie, “UAV-based photogrammetry and geocom-puting for hazards and disaster risk monitoring–a review,”
Geoenviron-mental Disasters , vol. 3, no. 1, p. 23, 2016.[247] W. Ejaz, A. Ahmed, A. Mushtaq, and M. Ibnkahla, “Energy-efficienttask scheduling and physiological assessment in disaster managementusing UAV-assisted networks,”
Computer Communications , Apr. 2020.[248] E. Baranwal, P. Seth, H. Pande, S. Raghavendra, and S. Kushwaha,“Application of unmanned aerial vehicle (UAV) for damage assessmentof a cultural heritage monument,” in
International Conference onUnmanned Aerial System in Geomatics . Roorkee, India: Springer,2019, pp. 123–131.[249] S. N. K. B. Amit, S. Shiraishi, T. Inoshita, and Y. Aoki, “Analysisof satellite images for disaster detection,” in . Beijing,China: IEEE, 2016, pp. 5189–5192.[250] J. Doshi, S. Basu, and G. Pang, “From satellite imagery to disasterinsights,” arXiv preprint arXiv:1812.07033 , Dec. 2018.[251] G. Deepak, A. Ladas, Y. A. Sambo, H. Pervaiz, C. Politis, and M. A.Imran, “An overview of post-disaster emergency communication sys-tems in the future networks,”
IEEE Wireless Communications , vol. 26,no. 6, pp. 132–139, Dec. 2019.[252] S. Hayat, E. Yanmaz, T. X. Brown, and C. Bettstetter, “Multi-objectiveUAV path planning for search and rescue,” in . Singapore: IEEE,2017, pp. 5569–5574.[253] E. T. Alotaibi, S. S. Alqefari, and A. Koubaa, “LSAR: Multi-UAVcollaboration for search and rescue missions,”
IEEE Access , vol. 7, pp.55 817–55 832, Apr. 2019.[254] K. Nakadai, M. Kumon, H. G. Okuno, K. Hoshiba, M. Wakabayashi,K. Washizaki, T. Ishiki, D. Gabriel, Y. Bando, T. Morito et al. ,“Development of microphone-array-embedded UAV for search andrescue task,” in . Vancouver, BC, Canada: IEEE, 2017,pp. 5985–5990.[255] L. Bracciale, P. Loreti, M. Luglio, C. Roseti, and F. Zampognaro,“SMARTSOS: Systems for maritime advanced rescue through satelliteand optimized sensor networks,” in . IEEE, 2016, pp. 1–5. [256] M. Rossi and D. Brunelli, “Autonomous gas detection and mappingwith unmanned aerial vehicles,”
IEEE Transactions on Instrumentationand Measurement , vol. 65, no. 4, pp. 765–775, Apr. 2016.[257] Z. Zaheer, A. Usmani, E. Khan, and M. A. Qadeer, “Aerial surveillancesystem using UAV,” in , Hyderabad,India, 2016, pp. 1–7.[258] A. Saha, A. Kumar, and A. K. Sahu, “FPV drone with GPS used forsurveillance in remote areas,” in , Kolkata, India, 2017, pp. 62–67.[259] R. Dai, S. Fotedar, M. Radmanesh, and M. Kumar, “Quality-aware UAVcoverage and path planning in geometrically complex environments,”
Ad Hoc Networks , vol. 73, pp. 95–105, May 2018.[260] S. R. Pokhrel, J. Jin, and H. L. Vu, “Mobility-aware multipath commu-nication for unmanned aerial surveillance systems,”
IEEE Transactionson Vehicular Technology , vol. 68, no. 6, pp. 6088–6098, Jun. 2019.[261] S. Garg, A. Singh, S. Batra, N. Kumar, and L. T. Yang, “UAV-empowered edge computing environment for cyber-threat detection insmart vehicles,”
IEEE Network , vol. 32, no. 3, pp. 42–51, Jun. 2018.[262] M. Bacco, A. Berton, E. Ferro, C. Gennaro, A. Gotta, S. Matteoli,F. Paonessa, M. Ruggeri, G. Virone, and A. Zanella, “Smart farm-ing: Opportunities, challenges and technology enablers,” in ,Tuscany, Italy, 2018, pp. 1–6.[263] M. N. Islam Sarker, M. Wu, B. Chanthamith, S. Yusufzada, D. Li, andJ. Zhang, “Big data driven smart agriculture: Pathway for sustainabledevelopment,” in , Chengdu, China, China, 2019,pp. 60–65.[264] A. D. Boursianis, M. S. Papadopoulou, P. Diamantoulakis, A. Liopa-Tsakalidi, P. Barouchas, G. Salahas, G. Karagiannidis, S. Wan, andS. K. Goudos, “Internet of things (IoT) and agricultural unmannedaerial vehicles (UAVs) in smart farming: A comprehensive review,”
Internet of Things , p. 100187, Mar. 2020.[265] P. K. R. Maddikunta, S. Hakak, M. Alazab, S. Bhattacharya, T. R.Gadekallu, W. Z. Khan, and Q.-V. Pham, “Unmanned aerial vehiclesin smart agriculture: Applications, requirements, and challenges,”
IEEESensors Journal , 2021.[266] N. T. Tam, H. T. Dat, P. M. Tam, V. T. Trinh, N. T. Hung, Q.-T. Huynh,and J. Jo, “Monitoring agriculture areas with satellite images and deeplearning,”
Applied Soft Computing , p. 106565, Oct. 2020.[267] D. Murugan, A. Garg, T. Ahmed, and D. Singh, “Fusion of drone andsatellite data for precision agriculture monitoring,” in .Roorkee, India: IEEE, 2016, pp. 910–914.[268] D. Murugan, A. Garg, and D. Singh, “Development of an adaptiveapproach for precision agriculture monitoring with drone and satellitedata,”
IEEE Journal of Selected Topics in Applied Earth Observationsand Remote Sensing , vol. 10, no. 12, pp. 5322–5328, Dec. 2017.[269] M. Z. Chowdhury, M. Shahjalal, S. Ahmed, and Y. M. Jang, “6Gwireless communication systems: Applications, requirements, technolo-gies, challenges, and research directions,”
IEEE Open Journal of theCommunications Society , vol. 1, pp. 957–975, Jul. 2020.[270] C. Grasso and G. Schembra, “Design of a UAV-based videosurveillancesystem with tactile internet constraints in a 5G ecosystem,” in ,Montreal, QC, Canada, 2018, pp. 449–455.[271] Y. Jin, Z. Qian, and W. Yang, “UAV cluster-based video surveillancesystem optimization in heterogeneous communication of smart cities,”
IEEE Access , vol. 8, pp. 55 654–55 664, Mar. 2020.[272] S. Ullah, K.-I. Kim, K. H. Kim, M. Imran, P. Khan, E. Tovar, andF. Ali, “UAV-enabled healthcare architecture: Issues and challenges,”
Future Generation Computer Systems , vol. 97, pp. 425–432, Jan. 2019.[273] C. Tang, C. Zhu, X. Wei, J. J. P. C. Rodrigues, M. Guizani, andW. Jia, “UAV placement optimization for internet of medical things,” in , Limassol, Cyprus, Cyprus, 2020, pp. 752–757.[274] R. Gupta, A. Shukla, P. Mehta, P. Bhattacharya, S. Tanwar, S. Tyagi,and N. Kumar, “VAHAK: A blockchain-based outdoor delivery schemeusing UAV for healthcare 4.0 services,” in
IEEE INFOCOM 2020 -IEEE Conference on Computer Communications Workshops (INFO-COM WKSHPS) , Aug., 2020, pp. 255–260.[275] M. I. Malik, I. Mcateer, P. Hannay, and Z. Baig, “Preparing for securewireless medical environment in 2050: A vision,”
IEEE Access , vol. 6,pp. 25 666–25 674, May 2018. [276] A. Anema, N. D. Preston, M. Platz, and C. Unnithan, “Shapingthe future of global health: A review of canadian space technologyapplications in healthcare,” in Space Capacity Building in the XXICentury . Springer, Apr. 2020, pp. 193–205.[277] H. Menouar, I. Guvenc, K. Akkaya, A. S. Uluagac, A. Kadri, andA. Tuncer, “UAV-enabled intelligent transportation systems for thesmart city: Applications and challenges,”
IEEE Communications Mag-azine , vol. 55, no. 3, pp. 22–28, Mar. 2017.[278] C. Kyrkou, S. Timotheou, P. Kolios, T. Theocharides, and C. G.Panayiotou, “Optimized vision-directed deployment of UAVs for rapidtraffic monitoring,” in , Las Vegas, NV, USA, 2018, pp. 1–6.[279] R. A. Nazib and S. Moh, “Routing Protocols for Unmanned AerialVehicle-Aided Vehicular Ad Hoc Networks: A survey,”
IEEE Access ,vol. 8, pp. 77 535–77 560, Apr. 2020.[280] M. Samir, S. Sharafeddine, C. Assi, T. M. Nguyen, and A. Ghrayeb,“Trajectory planning and resource allocation of multiple UAVs for datadelivery in vehicular networks,”
IEEE Networking Letters , vol. 1, no. 3,pp. 107–110, Sep. 2019.[281] H. Wu, F. Lyu, C. Zhou, J. Chen, L. Wang, and X. Shen, “Optimal UAVcaching and trajectory in aerial-assisted vehicular networks: A learning-based approach,”
IEEE Journal on Selected Areas in Communications ,vol. 38, no. 12, pp. 2783–2797, Dec. 2020.[282] S. A. Hadiwardoyo, C. T. Calafate, J.-C. Cano, K. Krinkin, D. Klion-skiy, E. Hern´andez-Orallo, and P. Manzoni, “Three dimensionalUAV Positioning for Dynamic UAV-to-Car Communications,”
Sensors ,vol. 20, no. 2, p. 356, Jan. 2020.[283] O. Bouachir, M. Aloqaily, I. A. Ridhawi, O. Alfandi, and H. B. Salameh, “UAV-Assisted Vehicular Communication for DenselyCrowded Environments,” in
NOMS 2020 - 2020 IEEE/IFIP NetworkOperations and Management Symposium , Budapest, Hungary, 2020,pp. 1–4.[284] F. Dovis, L. Ruotsalainen, R. Toledo-Moreo, Z. Z. M. Kassas, andV. Gikas, “Recent advancement on the use of global navigation satel-lite system-based positioning for intelligent transport systems [guesteditorial],”
IEEE Intelligent Transportation Systems Magazine , vol. 12,no. 3, pp. 6–9, Jul. 2020.[285] V. Sharma, M. Bennis, and R. Kumar, “UAV-assisted heterogeneousnetworks for capacity enhancement,”
IEEE Communications Letters ,vol. 20, no. 6, pp. 1207–1210, Jun. 2016.[286] M. Mozaffari, A. Taleb Zadeh Kasgari, W. Saad, M. Bennis, andM. Debbah, “Beyond 5G with UAVs: Foundations of a 3D wirelesscellular network,”
IEEE Transactions on Wireless Communications ,vol. 18, no. 1, pp. 357–372, Jan. 2019.[287] M. De Sanctis, E. Cianca, G. Araniti, I. Bisio, and R. Prasad, “Satellitecommunications supporting internet of remote things,”
IEEE Internetof Things Journal , vol. 3, no. 1, pp. 113–123, Feb. 2016.[288] S. Zhang, C. Xiang, and S. Xu, “6G: Connecting everything by 1000times price reduction,”
IEEE Open Journal of Vehicular Technology ,vol. 1, pp. 107–115, Mar. 2020.[289] K. B. Letaief, W. Chen, Y. Shi, J. Zhang, and Y.-J. A. Zhang, “Theroadmap to 6G: AI empowered wireless networks,”
IEEE Communi-cations Magazine , vol. 57, no. 8, pp. 84–90, Aug. 2019.[290] P. Yang, Y. Xiao, M. Xiao, and S. Li, “6G wireless communications:Vision and potential techniques,”