PaperlayerView exact revision on arXiv

Source-linked AI summary

A Survey on Millimeter-Wave Beamforming Enabled UAV Communications and Networking

Zhenyu Xiao, Lipeng Zhu, Yanming Liu, Pengfei Yi, Rui Zhang, Xiang-Gen Xia, Robert Schober

arXiv:2104.09204v2eess.SP

TL;DR

UAV data demands are pushing beyond congested sub-6 GHz capacity, while mmWave links introduce propagation, mobility, coverage, and deployment challenges. This paper surveys mmWave beamforming enabled UAV communications and networking across antennas, channels, cellular-connected UAVs, and ad hoc networks. It synthesizes existing technologies, solutions, open problems, and research directions.

  • Problem

    UAV communication capacity must support growing mission data demands, but mmWave systems face blockage, rapidly varying channels, coverage limits, and directional-beam alignment challenges.

  • Method

    The paper conducts a comprehensive survey of antenna structures, channel modeling, UAV-connected mmWave cellular networks, and mmWave-UAV ad hoc networks.

  • Results

    The survey identifies technical potentials, key technologies, application-specific solutions, and open problems for mmWave beamforming enabled UAV communications and networking.

  • Takeaways & Limitations

    MmWave beamforming enabled UAV communications has broad application potential, but practical systems require solutions for mobility, coverage, channel variation, and network resource coordination.

  • Takeaways & Limitations

    Existing mmWave-UAV studies focus mainly on theoretical performance analysis and commonly use statistical channel models.

Abstract

from arXiv · show

Unmanned aerial vehicles (UAVs) have found widespread commercial, civilian, and military applications. Wireless communication has always been one of the core technologies for UAV. However, the communication capacity is becoming a bottleneck for UAV to support more challenging application scenarios. The heavily-occupied sub-6 GHz frequency band is not sufficient to meet the ultra high-data-traffic requirements. The utilization of the millimeter-wave (mmWave) frequency bands is a promising direction for UAV communications, where large antenna arrays can be packed in a small area on the UAV to perform three-dimensional (3D) beamforming. On the other hand, UAVs serving as aerial access points or relays can significantly enhance the coverage and quality of service of the terrestrial mmWave cellular networks. In this paper, we provide a comprehensive survey on mmWave beamforming enabled UAV communications and networking. The technical potential of and challenges for mmWave-UAV communications are presented first. Then, we provide an overview on relevant mmWave antenna structures and channel modeling. Subsequently, the technologies and solutions for UAV-connected mmWave cellular networks and mmWave-UAV ad hoc networks are reviewed, respectively. Finally, we present open issues and promising directions for future research in mmWave beamforming enabled UAV communications and networking.

I. INTRODUCTION

UAV communication capacity is strained by growing data demands, motivating mmWave links with directional beamforming. The survey reviews mmWave-UAV potentials, challenges, applications, prior work, and two networking paradigms.

  • Growing sensor resolution and mission data demands make higher-rate UAV data links increasingly necessary.
  • Applications: UAVs can provide LoS-prone, flexibly deployed aerial communication support for remote areas, emergencies, and terrestrial mmWave networks.
  • Potentials: MmWave offers broad bandwidth and short wavelengths that support large antenna arrays, high beam gains, and compact UAV integration.More than 150 GHz of bandwidth is available in favorable mmWave bands, and over 600 half-wavelength-spaced antennas can fit in 1 square decimeter at 38 GHz.
  • Challenges: MmWave-UAV systems face blockage, high path loss, mobility, Doppler, jitter, aircraft shadowing, limited coverage, and stringent UAV SWAP constraints.
  • Scope and contributions: The survey covers antenna structures, channel modeling, UAV-connected mmWave cellular networks, and mmWave-UAV ad hoc mesh networks.
  • Existing works and contributions: Compared with earlier surveys, this work provides broader coverage of mmWave-UAV technologies, state-of-the-art progress, and open problems.

II. ANTENNA STRUCTURE

The section surveys directional, aperture, integrated, conformal, and array antennas for mmWave-UAV platforms. It explains array geometry and beamforming principles while emphasizing gain, directivity, and UAV deployment constraints.

  • MmWave antenna design must offset higher path loss and atmospheric attenuation while satisfying UAV size, weight, and power constraints.
  • The survey also introduces conformal arrays, passive reflective beamforming with IRSs/RISs, and antenna deployment considerations for UAV payloads.
  • Aperture antennas: Aperture antennas provide high gain and directivity but generally require substantial deployment space, limiting suitability for small UAVs.Reflector antennas can achieve communication rates up to 50 Mbps in the cited Global Hawk example.
  • Integrated antennas: Integrated antennas include AoC and AiP structures: AoC saves space and cost, whereas AiP supports better performance through heterogeneous materials and processes.
  • Antenna arrays: ULA, URA, and UCA geometries support array beamforming, whose steering vectors depend on element positions, propagation direction, and carrier wavelength.
  • Antenna arrays: Array radiation is determined by the inner product of antenna weights and the steering vector, with weights controlling excitation amplitude and phase.

C. Conformal Array

Conformal arrays fit UAV surfaces while supporting wide-angle coverage, additional antenna elements, and aerodynamic advantages. Their mmWave development targets wideband, high-gain communication and full-space connectivity.

  • C. Conformal Array: Conformal arrays integrate with UAV surfaces without extra nacelles, reducing aerodynamic impact, drag, and fuel consumption.Their lightweight, compact design suits UAV space, payload, and energy constraints.
  • C. Conformal Array: Their fuselage-conforming geometry provides more surface area for additional antenna elements and larger mmWave beam gains.Conformal arrays also offer more geometry-design degrees of freedom than regular arrays with half-space coverage.
  • C. Conformal Array: CCA-enabled UAVs can form multiple beams across full space to connect simultaneously with neighboring UAVs and ground base stations.This network concept uses cylindrical conformal arrays in the mmWave frequency band.
  • C. Conformal Array: Conformal-array design still requires application-specific choices concerning materials, geometry, analysis, and synthesis.Flexible materials are identified as practical alternatives for space-limited UAV platforms.
  • C. Conformal Array: A mmWave conformal array achieves 9 dBi gain across the Ka-band and a peak gain of 11.35 dBi at 35 GHz.The cited Ka-band spans 26.5–40 GHz.

D. Beamforming Architectures

Beamforming architectures trade flexibility, hardware complexity, energy consumption, and beamforming performance. Hybrid and analog approaches are particularly relevant to practical mmWave-UAV systems, while IRS/RIS can reconfigure blocked propagation paths.

  • D. Beamforming Architectures: Beamforming electronically controls antenna-element phase and amplitude weights to produce different radiation patterns.The far-field radiation depends on both the array steering vector and antenna weight vector.
  • D. Beamforming Architectures: Digital beamforming gives each antenna element an independent RF chain, enabling flexible precoding, multi-stream transmission, and directional signal separation.Its many RF chains create high hardware complexity, cost, and power consumption, especially for large arrays.
  • D. Beamforming Architectures: Analog beamforming uses one RF chain with phase shifters or switches, reducing hardware requirements but providing fewer degrees of freedom.With phase-shifter implementation, only each element’s signal phase can be adjusted.
  • D. Beamforming Architectures: Hybrid beamforming reduces RF-chain count and energy consumption while retaining multi-stream transmission, but its architectures trade hardware complexity against beamforming gain.Fully connected designs provide full gain with NRF × N RF paths, whereas partially connected designs require N RF paths but yield lower gain.
  • D. Beamforming Architectures: Analog and hybrid architectures are preferred over fully digital beamforming for mmWave-UAV systems because of their cost and energy efficiency.Fully digital beamforming may be unsuitable for practical large-antenna implementations because of complexity, cost, and power consumption.
  • E. Intelligent Reflecting Surface: IRS/RIS addresses mmWave blockage by reconfiguring propagation through adjustable reflected paths and passive beamforming.Its reflecting elements independently control reflected-signal amplitude and phase; the surveyed scenarios include aerial BSs, UAV relays, and UAV user equipment.
  • F. Carrying Capability of UAVs: UAV SWAP limits constrain antenna-array deployment, with rotary-wing UAV payloads generally ranging from 0.3 to 2 kg.Professional carrier UAVs can lift 20–200 kg, illustrating substantial platform-dependent variation.
  • F. Carrying Capability of UAVs: MmWave’s short wavelength and advances in antenna and UAV manufacturing enable many antenna elements and high-gain antennas in limited UAV space.The cited developments include lightweight patch and conformal arrays and increased payload capacities.

A. Propagation Characteristics

mmWave-UAV channels combine short-wavelength propagation losses and sparse multipath with UAV-specific mobility, blockage, airframe shadowing, and position fluctuations. Existing models therefore distinguish deterministic and stochastic approaches while incorporating spatial and temporal channel characteristics.

  • Propagation characteristics: MmWave propagation features short wavelength, large bandwidth, high penetration loss, and strong atmospheric attenuation, while UAV mobility produces non-stationary, dynamically changing channels.Atmospheric effects include oxygen, water-vapor, and rain attenuation; penetration loss generally increases with carrier frequency.
  • Propagation characteristics: Free-space path loss is a central transmission-range bottleneck, with the free-space model applicable to unobstructed ideal A2A and some high-altitude rural LoS scenarios.The model uses carrier wavelength, transmitter–receiver distance, and a 1 m reference loss parameter.
  • Propagation characteristics: MmWave-UAV channels have limited effective multipath components because narrow Fresnel zones hinder diffraction and scattering depends strongly on the operating environment.Relevant multipath components mainly arise from reflections by the earth surface, buildings, and human bodies.
  • Propagation characteristics: UAV motion causes Doppler shifts, carrier-frequency offset, inter-carrier interference, and reduced coherence time; sufficient OFDM subcarrier spacing can alleviate the interference.Doppler depends on carrier frequency, mobile velocity, and angular dispersion.
  • Propagation characteristics: Airframe shadowing and hovering fluctuations can block or perturb LoS paths because of UAV structure, antenna placement, flight status, engine vibrations, and wind turbulence.A measured example at 28 GHz found an average Doppler spread of approximately -20 Hz to +20 Hz over 1.1–7.2 m transceiver distances.
  • Channel modeling: Existing mmWave channel models comprise large-scale fading, small-scale fading, and spatial-temporal characteristics, and are classified mainly as deterministic or stochastic models.Deterministic models include ray tracing and map-based methods; stochastic models include GSCMs and tapped delay line models.

1) A2A MmWave Channel Modeling:

A2A mmWave channels are typically time-varying, highly likely to contain a dominant LoS path, and sparse in multipath components. Modeling studies address coherence time, Doppler interference, hovering vibrations, blockage, spatial-temporal evolution, and learned channel generation.

  • A2A channel characteristics: A2A mmWave channels generally have high LoS probability, few scatterers, few MPCs, and a LoS path whose power gain exceeds NLoS paths.Basic analytical models often retain only the LoS path and assume quasi-static behavior within an appropriate time slot.
  • A2A channel characteristics: Prior work found channel coherence time can be shorter than the communication time slot under high velocities, high frequencies, and narrow beams, motivating models that include practical propagation factors.The cited static model incorporates atmospheric absorption, precipitation, and small-scale fading from UAV-position fluctuations.
  • A2A channel modeling: Sufficiently large OFDM subcarrier spacing makes Doppler-spread impact negligible in mmWave-UAV mesh networks.The cited evaluation studied inter-carrier interference using switch-based analog beam patterns.
  • A2A channel modeling: Higher directional gain increases vulnerability to orientation fluctuations in hovering multi-rotor UAVs.A segment ULA gain model provides closed-form SNR distribution expressions and evaluates outage probability against vibration angle and antenna-element count.
  • Related A2G modeling: A2G modeling accounts for random LoS/NLoS conditions, blockage, and ground multipath, while geometry-based approaches model scatterer locations and non-stationary cluster evolution.A 3D regular-shaped GSCM with spherical wavefronts uses birth-death processes for spatial-temporal clusters.
  • Learning-based modeling: A learned distributed channel-generation approach produces samples containing locations, channel gains, and AoA/AoD information, with higher modeling accuracy than non-cooperative strategies.Multiple UAVs can share generated samples over sub-6 GHz OFDMA links in a distributed hop-by-hop manner.

C. Summary and Discussion

The survey identifies mobility, blockage, airframe effects, and antenna directionality as central practical challenges, while comparing tractable antenna-pattern approximations for performance analysis. It also highlights the need for convergence and time efficiency in cooperative learning methods.

  • Summary and discussion: MmWave-UAV systems face distinctive obstacles, large Doppler shifts, aircraft shadowing, and UAV fluctuations that complicate channel estimation and performance analysis.Cooperative learning may improve multi-UAV channel estimation, but convergence and time efficiency remain requirements.
  • Antenna radiation pattern: Directional antenna patterns make mmWave-UAV analysis difficult because beam gain depends on elevation and azimuth angles and is generally not available in closed form.Approximate patterns trade analytical tractability against accuracy.
  • Antenna radiation pattern: The flat pattern is analytically tractable but can be inaccurate, particularly for interference-dominated systems.It assumes constant gains within the mainlobe and sidelobe regions.
  • Antenna radiation pattern: The sinc pattern more closely approximates both mainlobe and sidelobe behavior, whereas cosine and 3GPP patterns approximate the mainlobe well.The sinc approximation also retains the Fejér-kernel null space with a more tractable affine denominator.
  • Antenna radiation pattern: For N = 32 and θtilt = 90°, the comparison uses an HPBW of θ3dB = 3.17° to assess the approximation strategies against the antenna pattern.The same strategies generalize to 3D UPA patterns through orthogonal ULA steering-vector decomposition.

B. Performance Metric

UAV-assisted mmWave network analysis evaluates conventional link metrics alongside 3D coverage, local traffic capacity, energy efficiency, and secrecy throughput. UAV position and directional antenna patterns affect both desired signals and interference, while operating conditions determine whether noise or interference dominates.

  • Performance metrics: Performance metrics include SINR, spectrum efficiency, energy efficiency, outage probability, throughput, delay, coverage probability, coverage density, local traffic capacity, and secrecy throughput.UAV AP evaluation additionally requires metrics reflecting altitude, mobility, and three-dimensional service regions.
  • SINR and coverage: UAV positions and antenna patterns influence both desired-signal power and interference, making performance analysis and optimization more challenging.The SINR model includes path loss, shadowing, fading, multipath, delay, antenna gains, transmit power, noise, and intra- and inter-cluster interference.
  • Operating regimes: As bandwidth increases, noise can dominate; with high building density, blocked LoS and reduced signal power can make the system noise-limited, whereas small bandwidth and high UAV density favor interference limitation.These regimes support SNR or SIR approximations to SINR, respectively.
  • Coverage metrics: Coverage probability measures whether a user’s SINR exceeds a threshold, while 3D coverage density aggregates covered users over the UAV-served volume and trajectory.The covered-user set is defined by the SINR threshold, and the coverage volume is measured in cubic meters.
  • Energy and traffic metrics: Energy efficiency counts successfully transmitted bits per joule and includes both navigation-related and communication-related power consumption for UAV APs.Navigation power includes hovering and propulsion; communication power includes transmit, RF, and baseband consumption.
  • Security metrics: Increasing UAV transmit power does not always improve secrecy throughput because both legitimate-user and eavesdropper rates increase; positioning and beamforming can enhance it.The metric is the effective average confidential-message transmission rate and accounts for colluding or non-colluding eavesdroppers.

C. Summary and Discussion

The survey compares antenna-pattern models and frames mmWave-UAV cellular networking around coverage, access, relaying, mobility, and UAV-specific beamforming constraints.

  • Cosine and 3GPP antenna patterns balance tractability and accuracy for theoretical mmWave-UAV performance analysis.
  • UAV-connected mmWave cellular networks include aerial access points or relays that enhance terrestrial cellular service, and aerial UEs connected to ground networks.
  • The survey reviews beam coverage extension, multiple access, relaying, and mobility enhancement as four application scenarios for UAV-connected mmWave cellular networks.
  • Unlike ground BSs using mainly horizontal 2D beams, UAV APs require coverage strategies suited to multiple ground UEs in three-dimensional space.
  • UAV mmWave beamforming remains constrained by hardware bottlenecks and SWAP limits, while directional antennas offer a more mature alternative.
  • Single-directional beams are adequate for few concentrated UEs, but broader or multiple-beam strategies become necessary as UE number or dispersion increases.

2) Multi-Beam Coverage:

Multi-beam coverage for UAV APs uses hybrid, lens, phased-array, and optimization-based structures, while flexible beams address overlap and incomplete coverage.

  • Hybrid and lens structures: Hybrid analog-digital beamforming connects a small number of RF chains to many antennas through phase shifters or switches.
  • Hybrid and lens structures: Lens arrays simplify multi-beam generation but suffer power leakage, imperfect beam matching, and limited analog-domain interference mitigation.
  • Hybrid and lens structures: Phase-shifter and lens-array approaches generally generate separate beams through different RF chains, limiting analog-beam count and causing beam-gain loss under SWAP constraints.
  • Optimization-based coverage: Analog multi-beam design formulates array gain as Gi = |aH_i w|^2 and optimizes beamforming vectors toward desired directions.
  • Optimization-based coverage: The resulting problem is non-convex because of its constraints, motivating linear and modulus relaxations before optimization.
  • Optimization-based coverage: Heuristic strategies can approach multi-beam coverage optima but may require high computational complexity.
  • Flexible beams: Flexible beams are proposed when directional-beam overlap prevents complete coverage of widely distributed UEs.
  • Flexible beams: Riemannian-manifold optimization handles constant-modulus hardware constraints by minimizing beam-pattern error on a constrained manifold.

4) ML-Based Beam Coverage:

Machine learning is considered for beam coverage, access, backhaul, and trajectory decisions, but training overhead and changing environments remain central limitations.

  • ML-based beam coverage may narrow the gap between hardware-constrained sub-optimal beamforming and the optimum, while also supporting UAV position decisions from partial information.
  • Existing ML beamforming studies mainly target analog beam steering or forming, including reinforcement, deep reinforcement, cross-entropy, and meta-learning methods.
  • Rapidly changing UAV environments can make offline-trained models unsatisfactory, and joint learning of beamforming and trajectory control remains open.
  • Multiple access: SDMA exploits sparse channels and directional beams for spatial multiplexing, but correlated channels and beam leakage create interference requiring channel-aware scheduling.
  • Multiple access: NOMA distinguishes signals by power levels and SIC, and reported mmWave-NOMA results outperform mmWave-OMA in spectrum and energy efficiency.
  • Multiple access: CoDMA uses flexible mapping constellations and rateless coding to adapt data rates to rapidly changing UAV and UE channel conditions.
  • Backhaul: MmWave UAV backhaul supports high-rate connectivity through large continuous bandwidth, while hybrid beamforming and stochastic-geometry models address backhaul design.

4) Summary and Discussion:

The survey identifies access, backhaul, relaying, and aerial-UE connectivity technologies for mmWave-UAV networks, emphasizing interference, mobility, and wireless-backhaul constraints.

  • Summary and discussion: SDMA, NOMA, and CoDMA are reviewed as promising multiple-access technologies, with scheduling, trajectory, deployment, and beamforming requiring joint design.
  • Backhaul: MmWave-UAV integrated access and backhaul is promising, but bandwidth partitioning and beamforming or trajectory optimization remain important open problems.
  • Relaying: UAV relays support ground, D2D, long-range-backhaul, and emergency links, using amplify-and-forward, decode-and-forward, or compress-and-forward strategies.
  • IRS-assisted relaying: IRS-assisted UAV relays use configurable reflections and passive beamforming to improve sparse, blockage-sensitive mmWave channels without transmit power or induced noise at the IRS.
  • Duplex modes: Relays operate in half-duplex, out-band full-duplex, or in-band full-duplex modes, with half-duplex forwarding requiring two alternating phases.
  • Relaying: Full-duplex UAV relay studies jointly optimize positioning, beamforming, and power control, showing analog beamforming can suppress self-interference.
  • Aerial UEs: At 28 GHz, reported 5G mmWave aerial-UE connectivity achieved up to 1 Gbps with latency not exceeding 1 ms near the mission area.
  • Aerial UEs: Fast aerial-UE motion makes beam training, channel estimation, tracking, and handover difficult, motivating transmit- and receive-beam tracking procedures.

1) Beam Tracking:

Beam tracking and handover address mobility-induced beam misalignment in mmWave UAV links. Position information, channel prediction, BS cooperation, and coordinated scheduling are reviewed, while aerial-terminal handover remains insufficiently studied.

  • Beam Tracking: IEEE 802.11ad initiates beam refinement when link quality falls below a threshold, using transmit or receive training units.IEEE 802.11ay extends these beam-tracking procedures.
  • Beam Tracking: Position-aided beam tracking uses UAV location and trajectory information, becoming more efficient when an LoS path exists.Kalman-filter fusion can track channel angles, but attitude errors may cause beam misalignment.
  • Beam Tracking: ML-based trajectory prediction and beam-coherence-time estimation can reduce training overhead when position information is available or unavailable.Joint beam training and angular-velocity estimation is used to calculate instantaneous beam coherence time.
  • Beam Handover: Inter-cell handover at cell edges is costly because directional 3D beam scanning incurs high time overhead; BS cooperation can reduce the switching search space.Sharing serving-beam pointing allows the candidate BS to estimate the possible AoD/AoA range.
  • Beam Handover: Inter-beam coordinated scheduling allocates resources on serving and candidate beams, but aerial-UE beam handover requires more specialized research.Existing strategies were primarily designed for terrestrial UEs.
  • Summary and Discussion: MmWave-UAV communication tests report high reliability, low delay, and large capacity, while mobility remains a central beam-management challenge.A moving UAV near a cell edge requires timely inter-cell handover; beam scanning can introduce substantial overhead and delay.

1) Network Topology:

MmWave-enabled FANETs require adaptable architectures and link-management methods because UAV motion and directional beams continually change topology and connectivity. SDN, neighbor discovery, and routing are surveyed as complementary responses, but routing remains difficult in dynamic networks.

  • SDN-Based Network: SDN improves network-state visibility, routing selection, and configuration by separating control and data planes and enabling programmatic control.Centralized controllers can schedule global resources and traffic, while limited RF chains constrain simultaneously accessed nodes.
  • Link Establishment and Maintenance: MmWave FANETs face link outages and performance degradation because directional beams and 3D motion complicate link establishment and maintenance.Neighbor discovery is foundational, but frequent discovery consumes energy and creates overhead.
  • Link Establishment and Maintenance: Directional neighbor discovery suffers spatial rendezvous, deafness, hidden-terminal, delay, and overhead problems, making exhaustive 3D scanning impractical.Deterministic, probabilistic, pseudo-deterministic, and learning-based schemes offer different delay and reliability tradeoffs.
  • Routing: FANET routing is classified as topology-based, geographic, hybrid, or bio-inspired, but its applicability to mmWave communications needs further study.After neighbor discovery, routing can be modeled as a time-dependent multi-commodity flow problem coupled with resource allocation.

3) Resource Allocation:

Resource allocation in mmWave-enabled FANETs must handle directional interference, fluctuating links, beam alignment, frame design, and rapidly growing optimization complexity. The surveyed approaches combine scheduling, prediction, self-healing, and learning, but scalability remains a major constraint.

  • Resource Allocation: Directional transmission, dynamic links, beam management, and time-consuming alignment make mmWave FANET MAC-resource allocation more challenging than in sub-6 GHz networks.Resource allocation must mitigate interference while improving throughput.
  • Resource Allocation: Half-duplex frame design must coordinate multipath and multihop traffic, whereas full duplex can double spectrum efficiency but requires interference-aware scheduling.The paper highlights frame design as a central time-domain challenge.
  • Resource Allocation: Fast beam tracking, resource reconfiguration, and collision-aware beam multiplexing are needed because neighbor nodes and link conditions change during operation.Frequency-division-assisted SDMA can avoid interference but reduces spectrum efficiency and transmission bandwidth.
  • Resource Allocation: Resource-optimization complexity grows exponentially with FANET scale, challenging timely network management.This scalability issue is separate from the computational demands of coupled routing and resource allocation.
  • Resource Allocation: Heuristic traffic distribution and transmission scheduling can improve delay and throughput while satisfying minimum traffic demands across multipath multihop flows.The surveyed scheme minimizes the number of time slots used for transmission.
  • Resource Allocation: A self-healing request/response mechanism improves mesh-network robustness and reduces directional-link establishment overhead.The associated method also reselects the UAV group leader to preserve link quality with the ground BS.
  • Resource Allocation: Position-prediction-based directional MAC protocols use mobility information across communication-control and data-transmission phases.The surveyed resource-allocation literature also includes optimization and ML approaches, but accurate CSI and scale remain concerns.

4) Summary and Discussion:

mmWave-enabled FANETs combine sub-6 GHz control with directional mmWave data transmission to balance management reliability and high-rate communication. Key unresolved issues include 3D neighbor discovery, coupled routing and resource allocation, and interference in integrated networks.

  • Neighbor Discovery: Neighbor discovery supports FANET self-organization, routing, and resource scheduling, but 3D directional scanning can create intolerable latency.Location knowledge or mobility prediction may assist discovery, although ML requires data and existing strategies are largely 2D.
  • Routing and Resource Allocation: Routing and resource allocation are coupled decisions, and the applicability of existing FANET routing strategies to directional mmWave links remains unresolved.Data-path selection becomes a network-flow problem after routing discovery.
  • Integrated Bands: Sub-6 GHz control channels can carry mobility, routing, and resource-management information, while mmWave links support high-rate, delay-tolerant data.Sub-6 GHz location information can also assist mmWave beam management and data-link establishment.
  • Integrated Bands: A cooperative discovery scheme using 2.4 GHz assistance and 60 GHz data transmission reduced average discovery time by 69 %-78 % versus conventional directional discovery.
  • Integrated Bands: Multi-band FANET resource allocation still faces interference-control challenges, including channel assignment and the need for low-complexity adaptive methods.Graph coloring and game-theoretic approaches have been studied, including a scheme with a Nash equilibrium and distributed convergence.
  • Integrated Bands: The integrated-band approach must account for serious interference from ground cellular networks operating in sub-6 GHz bands.

D. Security

mmWave-UAV FANET security addresses eavesdropping, jamming, and untrusted nodes through beam directionality, UAV positioning, physical-layer methods, and network-layer cooperation. Security design remains connected to channel modeling, mobility, placement, and imperfect CSI challenges.

  • Security Metrics: Security evaluation uses secrecy capacity, secrecy outage probability, area secure link number, network-wide secrecy throughput, and network-wide secrecy energy efficiency.
  • Anti-eavesdropping Techniques: Physical-layer defenses include artificial noise and cooperative jamming, but jamming power must balance eavesdropper disruption against interference to legitimate transmissions.
  • Anti-eavesdropping Techniques: Multi-hop relaying can bypass eavesdroppers, while cooperative UAV jamming and CoMP can respectively degrade wiretap channels and focus beam energy toward legitimate receivers.
  • Security Threats and Opportunities: MmWave directionality reduces network vulnerability, while UAV 3D positioning provides additional flexibility against jamming areas.
  • Open Security Context: MmWave-UAV security research must account for immature A2A and A2G channel measurement and modeling under 3D UAV motion.
  • Open Security Context: Joint 3D placement and beamforming can target throughput, coverage, latency, and security, but perfect CSI is difficult under dynamic channels and estimation overhead.

4) Communication Under Imperfect CSI:

The survey identifies imperfect channel knowledge, mobility, hardware constraints, and coupled network decisions as central barriers to practical mmWave-UAV systems. It reviews enabling approaches and concludes with broader integration opportunities and research directions.

  • Communication Under Imperfect CSI: Beamforming-based links require fast beam tracking and inter- and intra-cell handovers because directional transmission is sensitive to movement.A2A high-speed communications still require methods that reduce tracking, detection, and beam-training signaling overhead.
  • Communication Under Imperfect CSI: MmWave-UAV ad hoc networks couple physical- and MAC-layer resource allocation with network-layer routing under rapidly changing topologies.Time slot, carrier frequency, beam, and power must be scheduled according to communication tasks.
  • Communication Under Imperfect CSI: AI is proposed for complicated, decentralized, and autonomous mmWave-UAV systems because data-driven methods can be model-free, adaptive, scalable, and distributed.
  • Communication Under Imperfect CSI: IRS/RIS and UAV mobility can help alleviate mmWave blockage, but practical strategies must balance computational complexity against system performance.
  • Communication Under Imperfect CSI: Lightweight, low-power, low-cost, integrated, precise, and controllable onboard hardware remains a key implementation requirement.
  • Survey Scope: The survey covers antenna structures, channel modeling, cellular-network technologies, FANET architecture, discovery, allocation, routing, band integration, security, and open research directions.
Loading 2104.09204v2…