PaperlayerView exact revision on arXiv

Source-linked AI summary

A Vision and Framework for the High Altitude Platform Station (HAPS) Networks of the Future

Gunes Kurt, Mohammad G. Khoshkholgh, Safwan Alfattani, Ahmed Ibrahim, Tasneem S. J. Darwish, Md Sahabul Alam, Halim Yanikomeroglu, Abbas Yongacoglu

arXiv:2007.15088v4cs.ITeess.SP

TL;DR

HAPS networks face growing metropolitan demand and unresolved requirements for waveforms, heterogeneous services, and resource management. This paper synthesizes the literature into a future-oriented framework, concluding that HAPS have substantial potential for next-generation deployments and diverse connectivity use cases.

Abstract

from arXiv · show

A High Altitude Platform Station (HAPS) is a network node that operates in the stratosphere at an of altitude around 20 km and is instrumental for providing communication services. Precipitated by technological innovations in the areas of autonomous avionics, array antennas, solar panel efficiency levels, and battery energy densities, and fueled by flourishing industry ecosystems, the HAPS has emerged as an indispensable component of next-generations of wireless networks. In this article, we provide a vision and framework for the HAPS networks of the future supported by a comprehensive and state-of-the-art literature review. We highlight the unrealized potential of HAPS systems and elaborate on their unique ability to serve metropolitan areas. The latest advancements and promising technologies in the HAPS energy and payload systems are discussed. The integration of the emerging Reconfigurable Smart Surface (RSS) technology in the communications payload of HAPS systems for providing a cost-effective deployment is proposed. A detailed overview of the radio resource management in HAPS systems is presented along with synergistic physical layer techniques, including Faster-Than-Nyquist (FTN) signaling. Numerous aspects of handoff management in HAPS systems are described. The notable contributions of Artificial Intelligence (AI) in HAPS, including machine learning in the design, topology management, handoff, and resource allocation aspects are emphasized. The extensive overview of the literature we provide is crucial for substantiating our vision that depicts the expected deployment opportunities and challenges in the next 10 years (next-generation networks), as well as in the subsequent 10 years (next-next-generation networks).

I. INTRODUCTION · A. Survey and Overview Articles on HAPS · B. Contributions and Outline

The article positions HAPS as an integral stratospheric layer of future 6G VHetNets, extending ubiquitous connectivity toward high-capacity services in urban, suburban, and remote areas. It surveys prior work while presenting a framework spanning HAPS mega-constellations, enabling technologies, networking methods, AI, and deployment challenges over the next 10–20 years.

I. INTRODUCTION

  • HAPS operates around 20 km in the stratosphere and can provide quasi-stationary, ultra-wide-coverage, high-capacity connectivity as part of a three-layer 6G VHetNet.The envisioned VHetNet integrates satellite, aerial, and terrestrial layers, with HAPS serving as an integral aerial component.Source: §I · p. 3
  • A HAPS mega-constellation could provide communication, computation offloading, and data analytics to millions of users and devices, including in dense urban areas.The framework includes intelligent long-distance satellite communication, distributed data-center functions, UAV edge intelligence, URLLC and eMBB support, and Internet, IoT, and distributed-machine-learning services.Source: §I · p. 3

A. Survey and Overview Articles on HAPS

  • Prior HAPS surveys covered architectures, applications, deployments, optical communications, channel models, hybrid satellite-terrestrial systems, and enabling techniques, but mainly targeted sparsely populated or underdeveloped areas.The reviewed literature also includes UAV communication surveys, whose low- to medium-altitude assumptions differ from HAPS coverage areas of 60 km to 400 km and multi-month or multi-year operation.Source: A. Survey and Overview Articles on HAPS · pp. 3–5
  • This article addresses the gap by positioning HAPS for 5G and beyond, emphasizing large-scale communications, intelligent relaying, computation offloading, distributed machine learning, and dense metropolitan coverage.It contrasts with, which focused mainly on 3G/4G communications, single-station remote-area and disaster applications, and less precise distinctions between UAVs and HAPS.Source: A. Survey and Overview Articles on HAPS · p. 5

B. Contributions and Outline

  • The article presents HAPS-mounted Super Macro Base Stations as cost-effective platforms that enhance coverage and capacity while supporting data acquisition, computing, caching, and processing.It also reviews HAPS energy subsystems, communications payload innovations, and network-architecture evolution.Source: B. Contributions and Outline · p. 5
  • The framework covers RSS-enabled payloads, radio-resource and interference management, waveform and multiple-access designs, and inter- and intra-HAPS mobility management.Discussed techniques include FTN, SEFDM, FBMC, NOMA, power and channel allocation, spectrum sharing, massive MIMO, handoff, network slicing, software-defined networking, and network-function virtualization.Source: B. Contributions and Outline · p. 5
  • AI and ML are examined for HAPS design, topology management, handoff, and resource allocation, alongside challenges and roadmaps for next-generation and next-next-generation deployments.The article distinguishes deployment horizons of the next 10 years and 10–20 years and organizes the discussion across use cases, regulation, subsystems, channels, resource management, handoff, network management, AI, and open issues.Source: B. Contributions and Outline · pp. 5–6

II. PROMISING USE-CASES FOR HAPS SYSTEMS IN NEXT-GENERATION NETWORKS … 1) HAPS-SMBS systems to support IoT services:

HAPS networks are envisioned as a complementary layer for next-generation wireless systems, extending beyond rural connectivity to support dense urban demand and diverse applications. HAPS-mounted super macro base stations (HAPS-SMBSs) can combine wide-area connectivity, capacity enhancement, computation, caching, and IoT support, while requiring new approaches for massive access and energy-constrained devices.

II. PROMISING USE-CASES FOR HAPS SYSTEMS IN NEXT-GENERATION NETWORKS

  • HAPS systems can complement terrestrial networks in urban and suburban areas by addressing concentrated, increasing demand without relying solely on network densification.The envisioned architecture uses one platform for multiple applications and is summarized against conventional HAPS systems in Table III.Source: §II · pp. 6, 8

A. HAPS-Mounted Super Macro Base Station (HAPS-SMBS)

  • HAPS-SMBSs address metropolitan coverage and capacity limitations that small cells and UAV-mounted base stations face because of rising demand, SWAP constraints, and UAV mobility.HAPS systems are quasi-stationary and offer larger footprints, greater computational power, and better line-of-sight links than UAVs.Source: A. HAPS-Mounted Super Macro Base… · p. 8
  • HAPS-SMBSs complement rather than replace terrestrial base stations by rapidly supplying capacity for high or unpredictable traffic and supporting coordinated massive-MIMO or distributed-MIMO operation.Multiple coordinated HAPS-SMBSs can provide flexible, highly precise beams for metropolitan areas.Source: A. HAPS-Mounted Super Macro Base… · p. 8
  • Beyond connectivity, HAPS-SMBSs can function as intelligent platforms for communication, computation, caching, data acquisition, and processing across diverse application domains.Their designs are intended to exploit machine-learning algorithms and support future services illustrated in Fig. 3.Source: A. HAPS-Mounted Super Macro Base… · p. 8, Figure 3 · p. 9

1) HAPS-SMBS systems to support IoT services:

  • HAPS-SMBSs can support IoT deployments requiring massive connectivity, reliability, and low latency by covering many low-rate devices, including those outside terrestrial coverage.They can collect IoT data and provide reliable, seamless, efficient, and cost-effective uplink connections.Source: 1) HAPS-SMBS systems to support IoT… · pp. 8, 10
  • Low-rate IoT devices can communicate directly with a HAPS using low transmission power, despite the roughly 20 km link, supporting devices that may operate for decades without battery replacement.Required transmission power depends on received SNR and transmission rate; low-rate operation reduces the required power.Source: 1) HAPS-SMBS systems to support IoT… · p. 10
  • HAPS-SMBS IoT access must handle mx-MTC with x >> 1, creating challenges in multiple access, collision reduction, retransmission control, reliability, energy consumption, and delay.Designers must balance fewer retransmissions against the reliability needs of mission-critical applications.Source: 1) HAPS-SMBS systems to support IoT… · p. 10

2) HAPS-SMBS systems for backhauling small and isolated BSs: … 5) HAPS-SMBS systems to fill coverage gaps:

The paper envisions HAPS-SMBS systems as versatile infrastructure for backhauling isolated small cells, supporting unpredictable demand, enabling aerial computing, and repairing terrestrial coverage gaps. Across these roles, HAPS advantages include large platforms, massive MIMO, high transmission power, LOS links, and persistent service, while weather, energy, mobility, and stale-CSI challenges require adaptive designs.

2) HAPS-SMBS systems for backhauling small and isolated BSs:

  • At 200 km, a HAPS-to-femto-BS link can provide nearly the average power gain of a macro-BS 1,000 m away, while quasi-stationarity reduces beam-management demands.For 50 km HAPS coverage, a femto-cell may receive power comparable to a macro-BS about 100 m away; occasional beam adjustment can offset delay concerns.Source: 2) HAPS-SMBS systems for backhauling… · p. 10
  • HAPS backhaul can offset mmWave rain and cloud losses through higher transmission power and large directional arrays, while limited scattering may improve received power by 10-20 dB.Fiber remains superior but may be costly for small-cell backhaul; mmWave with wide bandwidth, MIMO beamforming, and high-gain antennas is a cost-effective alternative.Source: 2) HAPS-SMBS systems for backhauling… · p. 10
  • Hybrid mmWave RF/FSO backhaul can improve weather robustness, but automatic switching, sophisticated resource allocation, and 3D beamforming may be needed because mmWave can reduce spectral efficiency.FSO performance can deteriorate substantially in cloudy, rainy, or foggy conditions.Source: 2) HAPS-SMBS systems for backhauling… · pp. 10–11

3) HAPS-SMBS systems to cover unplanned user events:

  • For flash crowds and other temporary events, HAPS-SMBS systems can shift capacity toward demand more effectively than SWAP-constrained UAV BSs through larger platforms, massive MIMO, and higher transmission power.Big-data prediction of event timing, traffic volume, and geographic demand is needed to allocate bandwidth, power, and computing resources.Source: 3) HAPS-SMBS systems to cover… · p. 11

4) HAPS-SMBS systems as aerial data centers:

  • HAPS-SMBS systems can function as aerial data centers, offering computational offloading, multiple service levels, backup computing, and LOS connectivity that reduces disconnection risk.AR applications are an example requiring high computational capability.Source: 4) HAPS-SMBS systems as aerial data… · p. 11
  • Flying data centers require sufficient solar power and payload capacity, while low atmospheric temperatures around -15°C to -50°C may reduce cooling energy and onboard processing can lower response delays.Design must quantify computational power demand, harvested solar energy, cooling needs, payload limits, and response-delay reduction.Source: 4) HAPS-SMBS systems as aerial data… · p. 11

5) HAPS-SMBS systems to fill coverage gaps:

  • HAPS-SMBS systems can permanently fill terrestrial coverage gaps caused by obstruction, using 3D massive-MIMO beamforming to create disjoint narrow beams for users.Long distances can make CSI feedback outdated, motivating beamforming and resource allocation based more on long-term or statistical CSI.Source: 5) HAPS-SMBS systems to fill coverage… · p. 11

6) HAPS-SMBS systems for supporting and managing aerial networks: … A. Aviation Regulations

The paper presents HAPS-SMBS systems as platforms for computing, connectivity, traffic management, and LEO handoff coordination across aerial and terrestrial networks. It also identifies aviation regulation as a major requirement for safe large-scale deployment, with rules spanning international and national authorities.

6) HAPS-SMBS systems for supporting and managing aerial networks:

  • HAPS-SMBS systems can extend UAV networks’ limited onboard computing through powerful processors, wide-area data collection, reduced terrestrial dependence, and lower interference during computational offloading.Machine learning can further enable intelligent UAV-network control with minimal reliance on terrestrial infrastructure, provided reliable, wide, low-latency connectivity and sufficient onboard processing and cooling are available.Source: 6) HAPS-SMBS systems for supporting… · pp. 11–12

7) HAPS-SMBS systems for supporting intelligent transportation systems:

  • HAPS-SMBS can support connected autonomous transportation by providing ubiquitous coverage and low-delay data offloading for vehicles, including in rural and remote areas.The system may relay sensor data through terrestrial gateways or process it onboard, requiring joint communication-computation planning and high QoS for reliable vehicle-to-HAPS links.Source: 7) HAPS-SMBS systems for supporting… · p. 12
  • A single HAPS-SMBS can cover large numbers of cargo drones in cities, supporting emerging 3D air highways and delivery services.Safe operation requires highly reliable, low-latency communications, potentially combining radio and vision, plus substantial offloading for path planning and navigation.Source: 7) HAPS-SMBS systems for supporting… · p. 12

8) HAPS-SMBS systems for handling LEO satellite handoffs and providing seamless connectivity:

  • HAPS-SMBS systems can manage frequent LEO-satellite handoffs through their wide upper footprint, providing seamless connectivity to aerial and terrestrial users without requiring specialized LEO terminals.User-to-HAPS links may use RF while HAPS-to-LEO links may use FSO, and the system can reduce terminal transmit-power, cost, and size through improved link budgets.Source: 8) HAPS-SMBS systems for handling LEO… · p. 12
  • Machine learning can predict LEO mobility and pre-establish handoffs, but models must adapt to changing satellite constellations and may incorporate tracking data to account for sudden orbital changes and collision risks.The proposed approach depends on learning satellite mobility patterns before the current connection is lost and remaining flexible under continual environmental changes.Source: 8) HAPS-SMBS systems for handling LEO… · pp. 12–13

III. REGULATORY ASPECTS

  • HAPS regulation involves ITU-R spectrum rules, ICAO safety oversight, and national civil-aviation authorities’ licensing and airspace controls, whose requirements vary across countries.HAPS operate near the boundary between controlled and uncontrolled airspace, while definitions, target altitudes, and safety studies remain important for secure operations.Source: §III · p. 13

A. Aviation Regulations

  • ICAO distinguishes HAPS as unmanned free balloons or unmanned aircraft, with substantially different regulatory treatment because balloons are excluded from real-time management.Although guidance is still developing, increasing propulsion and computational capabilities may enable real-time management of balloon platforms, as noted by.Source: A. Aviation Regulations · p. 13
  • Because large-scale HAPS deployments may involve international consortia, the paper calls for international rules governing licensing and operations to reduce near-space legal uncertainty.Liu and Tronchetti proposed designating the 18–100 km region as exclusive utilization space with corresponding rules.Source: A. Aviation Regulations · p. 13

B. Spectrum Regulations … 2) Terrestrial Part:

The paper outlines HAPS spectrum regulations and presents HAPS as a stratospheric communication system composed of onboard, non-terrestrial, and terrestrial network elements. It details the roles of spectrum allocations, control stations, gateways, and terrestrial users in enabling HAPS communications.

B. Spectrum Regulations

  • HAPS spectrum regulations designate multiple L-, S-, C-, K-, Ka-, and V-band allocations, while specifying bandwidth needs and interference-control requirements for HAPS links.Ground-to-HAPS links require 396–2969 MHz and HAPS-to-ground links require 324–1505 MHz; WRC-19 added the 31–31.3 GHz and 38–39.5 GHz bands, alongside existing 47.2–47.5 GHz and 47.9–48.2 GHz allocations.Source: B. Spectrum Regulations · pp. 13–14, Figure 4 · p. 14

IV. THE HAPS SYSTEM

  • The HAPS system combines a non-terrestrial part with a terrestrial part and includes onboard subsystems and network components supporting deployment and communications.The system overview covers energy and communications payload advancements, HAPS types, and major HAPS projects.Source: §IV · p. 14, A. System Components · p. 14

A. System Components

  • HAPS platforms operate in the stratosphere, whose low weather disturbance, limited clouds, and altitude above commercial air traffic support long-duration quasi-stationary deployment.Both aerostatic and aerodynamic platforms can be deployed in this environment.Source: A. System Components · p. 14

1) Non-terrestrial Part:

  • Onboard subsystems comprise flight control, energy management, and communications payload units that respectively stabilize and orient the platform, manage energy, and provide communications.Flight control also exchanges telemetry, tracking, and command information with the ground control station.Source: 1) Non-terrestrial Part · p. 14
  • The non-terrestrial network can connect HAPS nodes into constellations, integrate satellite layers, and link with LAPS platforms or UAV users,,,,.These links support networks involving UAV base stations or relays and diverse UAV user swarms.Source: 1) Non-terrestrial Part · p. 14

2) Terrestrial Part:

  • The terrestrial part contains control stations, communications gateways, and terrestrial networks that coordinate HAPS operations, connect to core networks, and serve terrestrial nodes and users.Control stations manage links and resources, monitor HAPS position and direction, and oversee takeoff and landing; gateways use wired backhaul and may be co-located with control stations.Source: 2) Terrestrial Part · pp. 14, 16
  • Communications gateways connect HAPS to core networks through wired backhaul, while terrestrial networks include base stations, mobile users, and IoT sensors.HAPS may communicate directly with terrestrial users or exchange data through the gateway; gateway and control-station equipment includes amplifiers, processing units, and directional antennas.Source: 2) Terrestrial Part · p. 16

B. Types of HAPS and Related Projects · C. Energy Management Subsystem · 1) HAPS Energy Sources:

HAPS are classified by crewed status and lifting principle, with unmanned platforms generally preferred for communications and platform choice governed by mission requirements. Energy management is essential for prolonged, cost-effective operation, with solar power favored because of stratospheric sunlight and large available panel area.

B. Types of HAPS and Related Projects

  • Unmanned HAPS are preferred for communications because prolonged operations are difficult for human pilots in the harsh stratospheric environment.Earlier manned platforms such as the B-57 Canberra, Lockheed F-104, and Proteus served meteorological, scientific, military, or telecommunications purposes.Source: B. Types of HAPS and Related Projects · p. 16
  • Aerostatic HAPS float using buoyancy, whereas aerodynamic HAPS generate lift through motion; aerostatic designs include balloons and airships.Airships can remain aloft for months or years but face drag and launch challenges, while aerodynamic aircraft have limited payloads, wind sensitivity, and large wingspans requiring beam-pointing adjustments,.Source: B. Types of HAPS and Related Projects · p. 16
  • Mission requirements determine the choice between aerostatic and aerodynamic HAPS, balancing deployment cost, coverage, payload, endurance, positioning, and flight duration.Aerodynamic platforms suit emergencies through lower deployment costs, easier takeoff and landing, and mobility, whereas aerostatic platforms suit longer-term missions requiring high payload and energy capacity.Source: B. Types of HAPS and Related Projects · p. 16
  • Hybrid HAPS combining aerostatic and aerodynamic advantages may support near-future applications, following the 2019 strategic relationship between Loon and HAPSMobile.Table IV summarizes popular HAPS projects, their classifications, and key features.Source: B. Types of HAPS and Related Projects · p. 16, Table IV · pp. 17–18

C. Energy Management Subsystem

  • Energy management affects HAPS flight duration and deployment costs, making it essential for feasible and cost-effective prolonged communications operations.The subsystem must manage both supplied and consumed energy during extended missions.Source: C. Energy Management Subsystem · p. 16

1) HAPS Energy Sources:

  • Conventional sources provide only about 48 hours of flight, while microwave and laser energy beams raise irradiation-safety concerns, making solar power the safer renewable alternative.Solar-array efficiency has reached 37.75% and powers exceeding 1.5 kW/kg; current lithium-ion batteries and fuel cells reach 250 Wh/kg and 1600 Wh/kg, respectively,,.Source: 1) HAPS Energy Sources · p. 18
  • Solar energy is the main HAPS power source because platforms operate above clouds and can support large solar panels, while batteries or hydrogen fuel cells cover nights and winter.A control unit manages recharging and operation between primary solar power and secondary sources –.Source: 1) HAPS Energy Sources · p. 18

2) HAPS Energy Consumption: … 1) Empirical-Statistical Models:

HAPS energy consumption depends on flight characteristics, payload architecture, and processing demands, motivating use-case-specific platform selection and energy-aware computation. Passive HAPS-RSS can reduce payload energy and deployment costs, while channel-model selection must account for stratospheric propagation differences from terrestrial environments.

2) HAPS Energy Consumption:

  • HAPS energy is divided between flight control and communications payloads, with aerodynamic platforms, larger or heavier designs, and payloads containing more active components generally requiring more energy.Flight energy includes stability, propulsion, and altitude- or direction-control demands, while payload energy depends on communications equipment and techniques.Source: 2) HAPS Energy Consumption · p. 19

1) Active Payload:

  • HAPS-BS payloads consume more energy than HAPS-RS payloads because they include greater processing capability, whereas HAPS-RS requires an intermediate station and increases round-trip delay.HAPS-SMBS capabilities for dense urban environments add payload weight and energy costs, motivating controlled computational tradeoffs and load sharing with terrestrial networks.Source: 1) Active Payload · p. 19

2) Passive payload- HAPS-RSS:

  • Passive RSS is proposed as a lightweight, low-energy payload for profit-driven deployments where advanced onboard communication, caching, and computing may not be economically justified.RSS can increase received SNR quadratically when the number of units is doubled and can substantially improve data rates.Source: 2) Passive payload- HAPS-RSS · p. 19
  • HAPS-RSS can support backhaul from rural areas to gateways and inter-HAPS links, with a ground control station configuring the onboard controller to direct reflected signals.Its passive nature also supports cost-effective multi-hop links for HAPS constellations.Source: 2) Passive payload- HAPS-RSS · pp. 19, 21, Figure 6 · p. 20
  • Passive HAPS-RSS reduces communication-payload energy to the RSS controller, with each reflector unit consuming 0.33 mW and reported systems achieving 40% higher energy efficiency than relay-assisted systems.The resulting weight and energy reductions can extend flight duration and lower maintenance and deployment costs.Source: 2) Passive payload- HAPS-RSS · p. 20
  • HAPS platform selection must match the use case: aerodynamic platforms suit rapid deployment and can provide up to 1,000 W, while airships support heavier payloads such as 450 kg at 8 kW for a 5-year mission.Long-duration IoT, ITS/CAV, and data-center applications require higher power and payload capacity than temporary large-event deployments.Source: 2) Passive payload- HAPS-RSS · p. 20

V. CHANNEL MODELS FOR HAPS SYSTEMS / 1) Empirical-Statistical Models:

  • HAPS channel modeling requires caution because stratospheric channels differ significantly from lower-altitude or terrestrial channels, where scatterers are more prevalent on the ground.Early models adapted land-mobile satellite channels, including a two-state Gilbert-Elliott structure based on DLR data, while later work modeled atmospheric effects.Source: §V · p. 21, 1) Empirical-Statistical Models · p. 21

2) Non-Geometric Stochastic Models: … VI. RADIO RESOURCE MANAGEMENT, INTERFERENCE MANAGEMENT, AND WAVEFORM DESIGN IN HAPS SYSTEMS

The paper surveys stochastic, optical, and performance models for HAPS channels, then frames radio-resource, interference, and waveform design around spectral efficiency, energy efficiency, QoS, and event-specific service requirements. It emphasizes evolving channel dynamics, atmospheric impairments, relay strategies, and application-aware resource management as central to future HAPS networks.

2) Non-Geometric Stochastic Models:

3) Geometry-Based Stochastic Models:

  • Geometry-based HAPS models show that elevation, array configuration, Doppler spread, scatterer distribution, and antenna placement materially affect correlation, capacity, and array design.The literature includes 3D geometry-based single-bounce models, Ricean fading extensions, sum-of-sinusoids simulations, wideband models, and relay-channel formulations.Source: 3) Geometry-Based Stochastic Models · p. 21

4) Non-Stationary Models:

  • Non-stationary HAPS models represent appearing and disappearing line-of-sight or multipath components with Markov processes and derive survival probabilities, correlation functions, and Doppler spectra.These models also cover multi-user environments, long-distance and small-scale time variation, and dual-polarized MIMO spatial and polarization correlation.Source: 4) Non-Stationary Models · p. 21

5) Air-to-Air Channels:

  • Air-to-air HAPS models incorporate horizontal and vertical node mobility with time-varying velocities, highlighting vertical movement while leaving scatterer modeling as an unresolved issue.The model derives time-frequency correlation and Doppler power spectrum for non-stationary air-to-air channels.Source: 5) Air-to-Air Channels · p. 22

B. Free-Space Optical Channels

  • FSO links are promising for HAPS backhaul because they are cost-effective, license-free, and high-bandwidth, while turbulence, wind, pressure fluctuations, and temperature irregularities affect received-signal stability.A numerical study demonstrated 9,000 km inter-HAPS distances with high reliability, while gamma-gamma and lognormal distributions were proposed to model channel characteristics,.Source: B. Free-Space Optical Channels · p. 22

C. System Performance Analysis

  • System analyses use Shannon capacity and coverage probability while showing that mobility, spatial correlation, antenna configuration, frequency, and relay conditions strongly shape HAPS performance.Multi-hop relaying mitigates optical impairments, but the worst relay channel can reduce capacity, outage, and BER; power allocation is proposed to limit these effects.Source: C. System Performance Analysis · p. 22

VI. RADIO RESOURCE MANAGEMENT, INTERFERENCE MANAGEMENT, AND WAVEFORM DESIGN IN HAPS SYSTEMS

  • HAPS radio-resource and interference management must jointly optimize spectral efficiency, energy efficiency, QoS, SINR, and placement across aerial, terrestrial, and satellite networks.Power, frequencies, beams, scheduling, and node placement affect outage probability, BER, delay, throughput, and spectral efficiency; Fig. 7 provides a taxonomy.Source: §VI · p. 23
  • For unplanned events, HAPS-SMBS resource management should maximize admitted users, minimize service dropping, satisfy heterogeneous rates fairly, and adapt priorities to event type.Emergency events may prioritize voice, festivals may prioritize hologram and augmented-reality services, and protests may prioritize video upstreaming; waveform and multiple-access design remain understudied.Source: §VI · p. 23

A. Power Control/Allocation and Interference Management in HAPS systems · 1) Mobility and Power Control: · 2) Power Control for Multicast Services:

HAPS power control centers on maintaining SINR and QoS under power and interference constraints, with schemes spanning admission control, mobility-aware allocation, access technologies, and multicast delivery. The literature also exploits HAPS-specific centralized coordination and evaluates terrestrial integration, renewable-energy access points, and channel selection.

A. Power Control/Allocation and Interference Management in HAPS systems

  • Power management is fundamental in HAPS systems because serving diverse QoS requirements depends on controlling receiver SINR through sophisticated allocation schemes.The reviewed power-control literature largely originated in early-2000s WCDMA systems.Source: A. Power Control/Allocation and… · p. 23
  • Power-control frameworks developed for WCDMA – can be adapted to multicarrier-CDMA and power-domain NOMA, whose SIC removes decoded interference.NOMA assigns more power to lower-gain UEs, whereas CDMA separates users through spreading-code differences.Source: A. Power Control/Allocation and… · p. 24
  • The CAC studies – maximize Grade of Service by minimizing weighted dropping and blocking probabilities while satisfying SINR and power constraints.These schemes admit users subject to service-quality and resource limits.Source: A. Power Control/Allocation and… · p. 24
  • HAPS collocation enables interference-condition exchange without signaling overhead, supporting centralized uplink and downlink CAC decisions based on cell, neighboring-cell, and platform power limits.Calls are blocked when admission would violate allowable base-station output powers or platform power and SIR thresholds.Source: A. Power Control/Allocation and… · p. 24

1) Mobility and Power Control:

  • Mobility-aware CAC remains sparsely studied: combines overflow and speed sensitivity across HAPS and terrestrial layers, while uses UE speed and direction to reduce handoff dropping.The hierarchical design avoids cross-layer interference by assigning HAPS and terrestrial coverage to different frequency bands.Source: 1) Mobility and Power Control · pp. 24–25, Figure 8 · p. 25
  • Renewable-energy on-the-move access points allocate available energy to maximize total utility for sequentially observed users.The scheme addresses service decisions under limited renewable energy.Source: 1) Mobility and Power Control · p. 25

2) Power Control for Multicast Services:

  • For multicast, selects terrestrial and/or HAPS channels for MBMS while preserving unicast QoS, and switches between point-to-point and point-to-multipoint channels using power and channel thresholds.The efficient multicast transport channel strongly affects overall system capacity, considering channel conditions, coverage radius, and application bit rates.Source: 2) Power Control for Multicast Services · p. 25

3) Joint Radio and Computational Power Management: … D. Antenna and Interference Management

The reviewed HAPS literature jointly addresses computational offloading, radio-resource allocation, spectrum sharing, placement, and antenna/interference control. These approaches adapt HAPS networks to changing demand, mobility, topology, and cross-system interference constraints.

3) Joint Radio and Computational Power Management:

  • Joint radio-computational management minimizes aerial MEC delay or weighted energy-time consumption through task offloading, transmission-power control, and adaptive service decisions, [163].[162] uses a multi-leader multi-follower Stackelberg game with an EPEC solution, while [163] uses SVM-based federated learning for proactive user association.Source: 3) Joint Radio and Computational Power… · pp. 25–26

B. Channel/Sub-Channel Allocation and Spectrum Sharing

  • Dynamic channel allocation exploits multi-user diversity, while spectrum sharing and inter-layer interference management support integrated multi-HAPS, terrestrial, and satellite networks.Reported methods address crosswind-induced handoffs, reinforcement-learning channel allocation, access-choice compensation, and proactive channel planning for changing user numbers and call volumes.Source: B. Channel/Sub-Channel Allocation and… · p. 26

1) Joint Power, Sub-Channel and Time-Slot Allocation:

  • Joint allocation of radio power, sub-channels, time slots, and frequency reuse optimizes multicast resource allocation, while priority-aware admission favors highest-priority users.The priority-aware formulation is solved with branch and cut using McCormick underestimators to relax a mixed binary quadratically constrained problem.Source: 1) Joint Power, Sub-Channel and… · p. 26

2) Spectrum Management and Sharing:

  • Spectrum-management studies examine cognitive radio, adaptive modulation, antenna beamwidth, frequency separation, and imperfect sensing to enable HAPS coexistence with terrestrial and satellite systems.A correction coefficient for non-ideal spectrum sensing was reported to improve HAPS-to-ground downlink capacity relative to ignoring estimation errors.Source: 2) Spectrum Management and Sharing · p. 26

C. HAPS Placement and Constellation Optimization

  • Self-organizing placement and constellation optimization adapt HAPS topology to user demand, atmospheric conditions, coverage, battery status, traffic, QoS, and connectivity requirements.Studies report benefits from aerial self-organization over fixed placement, cost-efficient broadband constellation design, and joint placement with power and fronthaul/backhaul association.Source: C. HAPS Placement and Constellation… · p. 27

D. Antenna and Interference Management

  • Highly directive, dynamically steerable multi-beam antennas are required to overcome path loss and manage intra-HAPS and terrestrial interference from overlapping main and side lobes.The review frames antenna and interference management through adaptive cell shaping, multipoint transmission, platform diversity, and massive MIMO.Source: D. Antenna and Interference Management · pp. 27–28

1) Adaptive Cell Shaping: … E. The Waveform: Signaling and Multiple Access

The section surveys HAPS antenna and waveform strategies for shaping coverage, managing interference, exploiting platform diversity, and improving spectral efficiency. It highlights adaptive beams, coordinated transmission, massive MIMO, and candidate waveforms while identifying unresolved design challenges for HAPS networks.

1) Adaptive Cell Shaping:

  • Adaptive beam design balances cell-edge received power against out-of-cell interference, with excessive directivity causing edge roll-off and insufficient directivity increasing inter-cell interference.Optimum directivity was formulated to maximize cell-edge power, while elliptical beams improved edge power under marginal RF link budgets.Source: 1) Adaptive Cell Shaping · pp. 28–29
  • Beam steering, irregular cell shaping, and cell-pointing techniques support flexible coverage, reduced interference, and potentially fewer handoffs, but steering complexity and handoff overhead remain important constraints.A four-actuator steering solution was simpler than independently gimbaling more than 100 antennas, while irregular cells achieved sidelobe levels as low as 40 dB.Source: 1) Adaptive Cell Shaping · p. 29

2) Exploiting Multipoint Transmission and Platform Diversity:

  • Multipoint transmission and platform diversity extend HAPS capacity and coverage through JT-CoMP, multiple HAPS nodes, constellations, and virtual MIMO, with approximately 4 km identified as optimal HAP spacing for a 5° beamwidth user antenna.JT-CoMP targets cell-edge user experience, while constellation evaluation considered angular separation, link-length ratio, sidelobe floor, and beamwidth.Source: 2) Exploiting Multipoint Transmission… · p. 29

3) Massive MIMO for HAPS Systems:

  • Massive MIMO improves HAPS diversity, SINR, and throughput through beamforming and spatial multiplexing, with distributed sub-arrays providing better diversity than co-located architectures.These capabilities support HAPS mega-constellations and coverage-gap filling, while beamforming methods address interference and power-rate trade-offs.Source: 3) Massive MIMO for HAPS Systems · pp. 29, 31
  • Massive-MIMO user grouping and beamforming achieved significant numerical gains and outperformed existing channel-correlation-matrix-based schemes.The approach exploits concentration of signal power in statistical eigen-modes and contributes to HAPS-SMBS coverage-gap deployment.Source: 3) Massive MIMO for HAPS Systems · p. 31

E. The Waveform: Signaling and Multiple Access

  • HAPS lacks a standardized waveform, so 6G waveform candidates may support access, inter-platform, and backhaul links despite limited reported HAPS-specific research.Existing wireless evolution includes OFDM/OFDMA and optional NOMA, but no specific HAPS waveform standard currently defines the radio interface.Source: E. The Waveform: Signaling and… · p. 31
  • OFDM’s PAPR, spectral-efficiency, and mmWave power-amplifier limitations motivate FBMC and single-carrier FTN signaling as candidate HAPS alternatives.FTN signaling can achieve higher spectral efficiencies than Nyquist signaling for the same target BER, while deep learning may reduce precoded-OFDM equalization complexity.Source: E. The Waveform: Signaling and… · p. 31

1) Filter Bank MultiCarrier (FBMC) Scheme: … VII. HANDOFF MANAGEMENT IN HAPS NETWORKS

The section surveys candidate waveforms and access schemes for HAPS, emphasizing spectral efficiency, implementation complexity, and massive connectivity, then explains why HAPS mobility requires specialized handoff management. It highlights adaptive, predictive, cooperative, and application-aware approaches to maintain QoS despite changing coverage and platform motion.

1) Filter Bank MultiCarrier (FBMC) Scheme:

  • FBMC flexibly shapes sub-carriers and resists time- and frequency-domain impairments, but long filters and high MIMO detection complexity limit low-latency and short-burst use.Filter selection can target channel conditions, such as rectangular filters for time dispersion and raised-cosine filters for frequency dispersion.Source: 1) Filter Bank MultiCarrier (FBMC)… · pp. 31–32

2) Faster-than-Nyquist Signaling:

  • FTN increases data rate within fixed bandwidth and energy by intentionally introducing controllable ISI, while requiring more complex receiver detection.For binary sinc pulses, τ ≥ 0.802 preserves minimum Euclidean distance with optimal MLSE, yielding around 25% higher data rate; a binary 30% excess-bandwidth root-raised-cosine example reaches 42% higher bit density at τ = 0.703.Source: 2) Faster-than-Nyquist Signaling · p. 32, Figure 10 · p. 33
  • For HAPS-SMBS NB-IoT uplinks, FTN can pack more bits/s/Hz without additional SNR or transmission errors, shifting cost to receiver complexity that onboard HAPS computing can accommodate.Combining FTN at IoT devices with mMIMO beamforming could further improve efficiency, but receivers must address both ISI and inter-device interference.Source: 2) Faster-than-Nyquist Signaling · p. 32

3) Spectrally Efficient Frequency Division Multiplexing (SEFDM):

  • SEFDM extends FTN into the frequency domain by packing sub-carriers more closely, improving spectral and energy efficiency for HAPS access links while introducing controlled ICI and open per-user spacing decisions.Its sub-carrier selection and spacing should depend on each UE’s application type and QoS requirements.Source: 3) Spectrally Efficient Frequency… · p. 32

4) Non-Orthogonal Multiple Access:

  • NOMA multiplexes users with different channel conditions in the power domain and uses superposition coding with SIC, improving capacity and user-fairness trade-offs over orthogonal access.SDMA exploits HAPS spatial beams, while MIMO-NOMA clusters multiple UEs per beam to support massive connectivity.Source: 4) Non-Orthogonal Multiple Access · pp. 32–34

VII. HANDOFF MANAGEMENT IN HAPS NETWORKS

  • HAPS handoff is harder than terrestrial handoff because both UEs and irregularly moving coverage footprints change, causing instability, frequent handoffs, and platform-triggered inter- or intra-HAPS transfers.HAPS position should remain within an ITU-recommended cylinder of radius 400 m and height approximately 700 m [7].Source: §VII · p. 34
  • Fixed-threshold handoff algorithms can cause ping-pong effects or delayed service interruption, motivating predictive, adaptive, cooperative, load-balancing, and traffic-aware schemes.Reported approaches use predicted RSSI, platform and terminal motion, channel gain, cooperative transmission, traffic load, and energy awareness to improve reliability and resource use [209, 213, 214, 215, 218, 219, 222].Source: §VII · pp. 34–35
  • Future HAPS handoff algorithms must match mobility speed and application latency needs, since rapidly moving transportation and aerial-network users may face unacceptable delays from conventional synchronization and access procedures.IoT and unplanned-event users may tolerate normal handoff delays, whereas time-sensitive applications require fast handoff and low packet loss; conventional 3/4-way access handshakes may violate 5G and beyond latency requirements.Source: §VII · pp. 35–36

VIII. HAPS NETWORK MANAGEMENT AND COMPUTATIONAL ROLE

Future HAPS networks require autonomous, collaborative management to coordinate swarms, support dynamic network slicing, and distribute control and computation across aerial infrastructure. HAPS can also serve as network-edge or MEC platforms, provided they use intelligent scheduling, resource allocation, offloading, and reliable inter-HAPS collaboration.

Autonomous coordination

  • Autonomous HAPS coordination can reduce human intervention while enabling collaborative aerial network management, network slicing, and privacy-preserving federated learning.Multiple HAPS deployments increase operational complexity and require reliable collaboration to exploit distributed computational resources.Source: §VIII · p. 36
  • Swarm intelligence converges faster and stabilizes coverage, whereas reinforcement learning achieves higher peak coverage but risks coverage dips and depends on suitable reward functions.Swarm intelligence may therefore be more efficient and reliable, while reinforcement learning is valuable for detecting or filling coverage gaps without tractable mathematical models.Source: §VIII · pp. 36–37

Network control and slicing

  • Softwarization through SDN, NFV, and network slicing can automate self-organizing control and dynamically allocate resources according to predicted user demands.Dynamic slicing must address changing traffic, multiple providers, and mobility through accurate demand prediction and resource allocation.Source: §VIII · pp. 37–38

Computational role

  • Distributed orchestrators, HAPS-based MEC clusters, and aerial data centers require intelligent scheduling and offloading decisions that account for energy, computation, processing loads, and network conditions.HAPS can provide a stable, powerful edge above aerial networks, but reliable inter-HAPS communication and advanced resource allocation remain necessary.Source: §VIII · p. 39

IX. THE ROLE OF AI IN HAPS SYSTEMS

AI is positioned as essential for automating, orchestrating, and optimizing future HAPS networks, while HAPS can provide aerial data-center and edge-computing resources for distributed, privacy-preserving machine learning. Progress depends on resource-efficient AI hardware, reliable inter-HAPS communications, and scalable algorithms for dynamic, large-scale environments.

Resource-constrained machine learning

  • Resource-efficient machine learning is advancing through GPU-capable microcontrollers, embedded classifiers, and compact recurrent-network algorithms designed for highly resource-scarce HAPS hardware.Examples include FastGRNN and FastRNN for implementing recurrent and gated recurrent neural networks on tiny devices.Source: §IX · pp. 39–40

Learning algorithms

  • Deep reinforcement learning can address handoffs, spectrum sharing, and scheduling in complex communications environments, but large state-action spaces and scalability remain challenges.Simplified approaches such as compressed deep neural-network learning have been proposed for resource-limited equipment.Source: §IX · pp. 39–40

Collaborative aerial intelligence

  • HAPS can support collaborative and federated learning by collecting wide-area data, hosting aerial edge-computing resources, and distributing machine-learning tasks across cooperating platforms.Federated learning offers privacy-preserving distributed learning, but reliable communications among HAPS are required.Source: §IX · p. 40

AI-enabled HAPS automation

  • AI can automate future multi-HAPS management by detecting changing user distributions, redirecting or forming beams, and predicting required capacity and coverage.Conventional management may become inefficient or impossible as networks include many HAPS of different types and characteristics.Source: §IX · p. 40

AI applications in HAPS

  • AI studies optimize HAPS constellation capacity and cost, cell-edge handoffs, beamforming interference, UAV task offloading, and multi-HAPS task planning.Reported approaches include an artificial immune algorithm [70], radial-basis-function neural networks, particle-swarm optimization, Stackelberg games, and genetic algorithms [249].Source: §IX · pp. 40–41

AI hardware and execution

  • A successful test of a commercial off-the-shelf neural-network accelerator on a HAPS demonstrates on-platform AI feasibility while reducing transmitted data and accelerating reactions to dynamic environments.Specialized hardware, including neuromorphic processors, can outperform the general-purpose CPUs otherwise needed for equivalent tasks.Source: §IX · p. 41

X. OPEN ISSUES … 2) System Issues:

The open issues span incremental next-generation research and a disruptive next-next-generation redesign of HAPS networks, alongside regulatory, energy, payload, and RSS trade-offs. Key needs include interference-aware spectrum management, sustainable multi-function platforms, and coordination between HAPS and satellite mega-constellations.

X. OPEN ISSUES

  • Next-generation HAPS research must adapt massive MIMO and mmWave techniques to restricted DoF, limited transmission energy, and unavailable channel statistics or models.These technologies are established terrestrially but require additional HAPS-specific investigation.Source: §X · p. 41
  • Next-next-generation HAPS networks require a disruptive redesign, particularly to determine whether HAPS and satellite mega-constellations should complement or compete with one another.Uncertain satellite technologies make early HAPS mega-constellation design highly challenging and speculative.Source: §X · p. 41
  • Table IX organizes the challenges by the paper sections that overwhelmingly influence them, while acknowledging that issues such as handoff and network management span multiple sections.Handoff management is mainly discussed in Section VII, with network management in Sections VIII and IX also contributing.Source: §X · p. 41

1) Regulatory Aspects:

  • Dedicated ITU spectrum is critical, while unlicensed ISM-band operation must be carefully planned to prevent interference with radio astronomy, as demonstrated by Project Loon tests in Oceania.Dynamic frequency allocation with cognitive-radio capabilities is identified as a promising mitigation.Source: 1) Regulatory Aspects · pp. 41–42

2) System Issues:

  • HAPS design must balance platform type, cost, performance, flight endurance, payload capacity, and energy consumption as stations add data-center, computation, or machine-learning functions.Further work is needed on energy management and service continuity, potentially including remote charging or nuclear energy.Source: 2) System Issues · p. 42
  • RSS reduces payload weight and energy consumption but competes with solar-panel area, creating a trade-off between reflected-signal directionality, spectral efficiency, and long-term solar absorption.A proposed configuration places solar panels on the platform’s upper surface and RSS functions below, while flat-versus-curved surface effects require study.Source: 2) System Issues · p. 42

3) PHY and Related Cross Layer Design: … 8) Networks Management of HAPS Systems:

The paper identifies open design challenges across HAPS PHY, resource management, channel modeling, MIMO, beam tracking, and network management, while highlighting AI-enabled and cross-layer approaches for future deployments. Key needs include HAPS-specific waveforms and models, low-complexity adaptive optimization, interference-aware architectures, and self-organizing control.

3) PHY and Related Cross Layer Design:

  • HAPS lacks a specified PHY waveform; candidate mm-Wave waveforms, pulse-shaping filters, and detectors must be evaluated for propagation, massive MIMO integration, complexity, and incomplete channel knowledge.AI/ML methods are proposed because they can operate without accurate stochastic models, extract features from complex observations, and converge faster than iterative model-based approaches.Source: 3) PHY and Related Cross Layer Design · p. 42
  • FTN and SEFDM require joint optimization of power, channel allocation, and squeezing or acceleration parameters, balancing spectral-efficiency gains against interference and detection-performance losses.For FSO links, BASK-based single-carrier FTN is proposed to improve spectral efficiency despite nonlinearities and could support Tbps HAPS-SMBS backhauling.Source: 3) PHY and Related Cross Layer Design · p. 42

4) Radio Resource Management (RRM):

  • AI/ML is positioned to reduce the computational burden of HAPS RRM, with a DNN reported to achieve orders-of-magnitude speedup over optimization-based power-allocation algorithms [254].Major gaps remain for integrated HAPS/LEO systems and for real-time multi-objective RRM supporting heterogeneous QoS requirements such as broadband throughput, URLLC delay, and packet loss.Source: 4) Radio Resource Management (RRM) · pp. 42–43
  • HAPS RRM must incorporate rechargeable solar-battery energy, multihop inter-platform routing, 3D drone mobility, association or double connectivity, media traffic, and overlooked metrics such as QoE and age of information.These requirements extend to mission-critical robotics, autonomous vehicles, AR/VR, cargo drones, video caching, and control-communication systems.Source: 4) Radio Resource Management (RRM) · p. 43

5) Channel Model and Performance Evaluation:

  • HAPS performance evaluation is hindered by incomplete channel models, limited single-HAPS studies, and neglected terrestrial-HAPS-satellite interactions, motivating more sophisticated analysis tools.Stochastic geometry can model large-scale HAPS deployments and severe LOS-driven inter-cell interference, while point processes and meta-distribution analysis can address station repulsion and information freshness.Source: 5) Channel Model and Performance… · pp. 43–44

6) Massive MIMO Communications:

  • Massive MIMO faces severe pilot contamination and limited spatial separability in highly LOS-dominant, low-scattering HAPS channels, requiring energy- and computation-aware user clustering and correlation-aware MIMO techniques.Stations hundreds of kilometers apart may contaminate pilots, while correlated signals can limit spatial-domain separation regardless of antenna count or processing power.Source: 6) Massive MIMO Communications · p. 44

7) Beam Tracking:

  • mmWave HAPS links need accurate, low-cost, fast beam tracking because rapidly maneuvering UAVs and temporary blockages can disrupt narrow-beam communication, especially for mission-critical applications.Machine learning for mobility prediction and computer vision for blockage detection are proposed to improve beam alignment and tracking.Source: 7) Beam Tracking · p. 44

8) Networks Management of HAPS Systems:

  • HAPS networks require intelligent self-organizing management that jointly adapts deployment, coverage, capacity, energy, computation, and changing platform responsibilities.Future multi-HAPS systems may use centralized manager-follower or distributed coordination to avoid interference, wasted resources, overlapping footprints, and beamforming conflicts.Source: 8) Networks Management of HAPS Systems · pp. 44–45

9) Handoff Management in HAPS Networks: … 1) Integration with Satellite Network:

Future HAPS networks require intelligent, cross-layer handoff management for large, dynamic multi-layer constellations, while their broader roles introduce computational, security, privacy, and satellite-integration challenges. These challenges motivate adaptive coordination, secure architectures, and robust multilevel satellite/HAPS designs.

9) Handoff Management in HAPS Networks:

  • Future HAPS mega-constellations will span multiple layers with hundreds of components, making conventional handoff management inefficient.Solutions must address complex future networks rather than only simple early-deployment scenarios.Source: 9) Handoff Management in HAPS Networks · p. 45
  • Handoff solutions must jointly manage Layer 2 and Layer 3 procedures for rapidly moving users and time-sensitive applications.Layer 2 includes scanning, channel selection, and cell association; Layer 3 includes IPv6 reconfiguration, mobility registration, and packet rerouting.Source: 9) Handoff Management in HAPS Networks · p. 45
  • Intelligent, self-adaptive inter- and intra-HAPS handoff should combine holistic beamforming with dynamic objectives while avoiding ping-pong effects.Beamforming should reduce handoff frequency, minimize transmitted power, or maximize capacity without creating undesirable oscillations.Source: 9) Handoff Management in HAPS Networks · p. 45
  • 5G handoffs face unacceptable three-way-handshake delays and weather-sensitive mmWave attenuation, motivating microwave-assisted double connectivity for handoff and beam management.Signal-strength-based algorithms may trigger unnecessary handoffs under variable mmWave conditions.Source: 9) Handoff Management in HAPS Networks · p. 45

10) Computational Roles:

  • HAPS can support aerial network management, network slicing, and computational offloading because their altitude, computing capacity, quasi-stationarity, and large footprints reduce delays and interruptions.Offloading from satellites or UAVs to HAPS can also free terrestrial links for terrestrial-aerial and terrestrial-satellite communications.Source: 10) Computational Roles · p. 45
  • Distributed HAPS computation requires strong inter-HAPS collaboration, federated learning for privacy, and intelligent scheduling based on energy, capability, and processing load.Coordination supports resource management, load balancing, UE mobility management, and decisions between HAPS MEC processing and terrestrial transfer.Source: 10) Computational Roles · p. 45

11) Privacy and Security Concerns:

  • HAPS security is difficult because a compromised node can expose extensive communications, while mmWave or FSO links remain vulnerable to misalignment, sidelobes, eavesdropping, and jamming.Proposed protections include information-theoretic and covert communications, quantum key distribution, and dedicated monitoring stations with radar, computer vision, and jamming capabilities.Source: 11) Privacy and Security Concerns · pp. 45–46

1) Integration with Satellite Network:

  • Vertical satellite/HAPS integration is deemed imperative for super connectivity but increases vulnerability and design complexity when satellite failures create coverage holes.Multilevel architectures must address compensation, regulation, extra HAPS functionality, and transparency between satellite and HAPS networks,.Source: 1) Integration with Satellite Network · p. 46

2) HAPS Mega-Constellation: … XI. CONCLUSIONS

The paper presents HAPS mega-constellations as a foundation for global connectivity, efficient reconfiguration, and edge intelligence in future networks. Its conclusion consolidates the proposed technologies, applications, and open challenges across next-generation and next-next-generation deployments.

2) HAPS Mega-Constellation:

  • HAPS mega-constellations can extend satellite coverage, reduce reliance on costly ground relays, and provide backbone infrastructure for fast cross-border Internet access.HAPS can boost, combine, and retransmit weak satellite signals, while clusters can support communications, relaying, routing, data centers, computation, and security functions.Source: 2) HAPS Mega-Constellation · p. 46

3) Efficient Network Reconfigurations:

  • HAPS network reconfiguration is difficult because stations are dynamic, heterogeneous, energy-constrained, and continuously coordinated with ground stations or satellite constellations.Aerostatic and aerodynamic platforms differ in mobility and functionality, making conventional structure-specific coordination, allocation, and networking approaches impractical.Source: 3) Efficient Network Reconfigurations · pp. 46–47
  • Meta-learning10 could enable rapid adaptation to new HAPS tasks and environments without repeatedly solving optimization problems for routing, coverage, energy, offloading, or latency.Training can cover coverage preservation, capacity enhancement, energy minimization, computation offloading, and latency reduction before adapting to emerging tasks.Source: 3) Efficient Network Reconfigurations · p. 47, §XI · p. 47

4) Support for Edge Intelligence:

  • HAPS networks can provide a universal intelligence blanket by collecting and training on large numbers of edge-device sub-models for distributed edge machine learning.Stations may aggregate hundreds of thousands of sub-models synchronously or asynchronously and incorporate their own collected data or intelligence.Source: 4) Support for Edge Intelligence · p. 47
  • Supporting edge intelligence requires secure, timely links and joint allocation of communication and control resources, while sharing resources between data communication and intelligence remains unresolved.Resource allocation, scheduling, and computational offloading must reflect the communication and computation requirements of large-scale distributed machine learning and mission-critical robotics.Source: 4) Support for Edge Intelligence · p. 47

XI. CONCLUSIONS

  • The paper identifies HAPS systems as important to ubiquitous 6G connectivity and presents mega-constellations and the HAPS-mounted Super Macro Base Station as prospective deployment paradigms.The SMBS is described as a promising, cost-effective response to future traffic demands and can support computing, caching, and processing for sensing, machine-type communications, UAVs, and IoT.Source: §XI · pp. 47–48
  • The proposed framework spans HAPS energy systems, RSS payloads, radio-resource and interference management, waveform and access design, mobility management, softwarized networking, and AI enablers.It includes inter-HAPS and intra-HAPS handoff algorithms and interactions with network slicing, software-defined networking, and network-function virtualization.Source: §XI · p. 48
  • Future work is divided between next-generation challenges requiring incremental technology development and next-next-generation challenges, including unresolved HAPS channel, energy, and detection limitations.Massive MIMO and mmWave communications are identified as next-generation examples, while insufficient channel knowledge, restricted transmission energy, and missing channel statistics remain open issues.Source: §XI · p. 48

Related papers

Loading 2007.15088v4…