High-Altitude Platforms in the Low-Altitude Economy: Bridging Communication, Computing, and Regulation

High-Altitude Platforms in the Lo w-Altitude Economy: Bridging Communication, Computing, and Re gulation Bang Huang, Member , IEEE , Baha Eddine Y oucef Belmekki, and Mohamed-Slim Alouini, F ellow , IEEE Abstract —The Low-Altitude Economy (LAE) is rapidly emerg- ing as a new technological and industrial frontier , with unmanned aerial vehicles (U A Vs), electric vertical tak eoff and landing (eV - TOL) aircraft, and aerial swarms increasingly deployed in logis- tics, infrastructure inspection, security , and emergency response. Howev er , the large-scale development of the LAE demands a reli- able aerial f oundation that ensur es not only r eal-time connectivity and computational support, but also navigation integrity and safe airspace management for safety-critical operations. High-Altitude Platforms (HAPs), positioned at around 20 km, provide a unique balance between wide-area coverage and low-latency responsive- ness. Compared with low earth orbit (LEO) satellites, HAPs are closer to end users and thus capable of delivering millisecond- level connectivity , fine-grained regulatory o versight, and po werful onboard computing and caching resources. Beyond connectivity and computation, HAPs-assisted sensing and regulation further enable navigation integrity and airspace trust, which are essential for safety-critical U A V and eVTOL operations in the LAE. This article proposes a five-stage ev olutionary roadmap for HAPs in the LAE: from ser ving as aerial infrastructure bases, to becoming super back-ends for U A V , to acting as fr ontline support f or ground users, further enabling swarm-scale U A V coordination, and ultimately advancing toward edge–air–cloud closed-loop autonomy . In parallel, HAPs complement LEO satellites and cloud infrastructures to form a global–regional–local three-tier architectur e. Looking forward, HAPs are expected to evolve from simple platforms into intelligent hubs, emerging as pivotal nodes for air traffic management, intelligent logistics, and emergency response. By doing so, they will accelerate the transition of the LAE toward large-scale deployment, autonomy , and sustainable gro wth. I . I N T R O D U C T I O N The Lo w-Altitude Economy (LAE) is emer ging as a promis- ing growth engine following the digital economy . Operating primarily within airspace below 1km and driv en by drones, electric vertical takeof f and landing (eVTOL) [1], and other low-altitude aerial vehicles, LAE spans diverse application domains, including logistics and deli very , emergenc y response, smart cities, precision agriculture, and tourism [2]. The global drones market was valued at USD 73 billion in 2024 and is projected to grow at a compound annual growth rate (CA GR) of 14.3 % from 2024 to 2030, highlighting the strong momentum of the sector 1 . It not only accelerates the commercialization of airspace resources but also fosters a Bang Huang and Mohamed-Slim Alouini are with King Abdul- lah University of Science and T echnology (KA UST), CEMSE divi- sion, Thuwal 23955-6900, Saudi Arabia (e-mail: bang.huang@kaust.edu.sa; slim.alouini@kaust.edu.sa). Baha Eddine Y oucef Belmekki is with Heriot-W att Univ ersity in Edinbur gh, United Kingdom (e-mail: B.Belmekki@hw .ac.uk). 1 https://www .grandviewresearch.com/industry-analysis/drone-market- report complete industrial value chain, from platform manufacturing and operational services to data platforms. Unsurprisingly , LAE has attracted strong attention from both industry and gov ernment. Aerospace leaders such as Airbus 2 and Boeing 3 are in vesting in low-altitude aviation and traffic management, while many countries are positioning LAE as a strategic sector beyond the digital economy and smart manufacturing 4 . W ith enabling policies and continuous technological advances, LAE is poised to become a new infrastructure ecosystem that tightly integrates communication, sensing, computation, and airspace services. Despite this momentum, the lar ge-scale deplo yment of LAE still faces critical bottlenecks. First, terrestrial cellular networks, designed primarily for ground users, fall short in providing wide-area and three-dimensional coverage for aerial vehicles below 1km, especially in mountainous regions, offshore en vironments, and post-disaster scenarios where com- munication blind spots are common. Second, edge computing capabilities remain limited, as most services rely on remote cloud or ground data centers, which cannot satisfy the low- latency task offloading required by fast-moving drone swarms. Third, airspace regulation is still immature. That means the lack of unified unmanned traffic management (UTM) 5 and real-time monitoring makes it difficult to detect trajectory conflicts, boundary violations, or potential collisions in time, thereby raising safety concerns. These challenges collectiv ely hinder LAE’ s transition from pilot deployments to large- scale, routine operations, underscoring the need for a new type of aerial infrastructure that can simultaneously provide communication, computing, and supervisory functions. More critically , such fragmentation prevents the establishment of a unified airspace supervision and safety assurance layer , which is indispensable for scaling the LAE beyond pilot deployments tow ard routine and large-scale operations. While lo w earth orbit (LEO) satellites ha ve been regarded as a key enabler of LAE [3], their limitations are becoming increasingly e vident. LEO constellations excel at global con- nectivity and navigation, but their latency , though lower than geostationary earth orbit (GEO) satellites, remains inadequate for mission-critical tasks demanding millisecond-lev el respon- siv eness. Moreover , their emphasis on “ubiquitous” co verage 2 https://www .airbus.com/en/innov ation/future-aircraft-operations/airbus- unmanned-traffic-management 3 https://www .boeing.com/innovation/inno vation- quarterly/2025/02/iq2025q1-evtole volution 4 https://www .arabianbusiness.com/industries/transport/saudi-arabia-set-to- launch-autonomous-air-taxi-flights-this-year 5 https://eda.europa.eu/docs/default-source/documents/sc4-final-report-v1- 0.pdf 2 comes at the expense of fine-grained, localized services in complex en vironments such as urban canyons, mountainous valle ys, or disaster zones. Finally , LEO lacks intrinsic capabil- ities for lo w-altitude traffic monitoring and regulation, limiting its ability to support real-time trajectory prediction, collision av oidance, and safe swarm coordination. In this context, high-altitude platforms (HAPs) emerge not merely as additional access points, but as bridging nodes that integrate communication, computing, and airspace regulation into a coherent aerial control plane [4]. Operating in the stratosphere at around 20 km, HAPs strike a unique balance between wide-area coverage and low-latenc y responsiveness. They can deli ver stable communication and sensing to both drones and ground users while acting as aerial “super nodes” that integrate edge computing, caching, and task scheduling. Furthermore, HAPs are naturally positioned to perform su- pervisory roles, monitoring unmanned aerial vehicle (UA V) trajectories, predicting risks, and issuing safety alerts, thereby complementing the capabilities of LEO satellites and terrestrial networks. Crucially , HAPs are unaf fected by ground geogra- phy and can maintain continuous coverage in mountainous, offshore, or post-disaster en vironments, ov ercoming the blind spots of terrestrial networks. More details can be seen in Fig. 1. T ogether with UA Vs, LEO satellites, and ground/cloud infrastructures, HAPs can form a hierarchical global–regional– local architecture, laying the foundation for the scaled, safe, and intelligent dev elopment of the LAE. Recognizing that the integration of HAPs into the LAE can effecti vely mitigate current deficiencies in communication cov erage, computational offloading, and regulatory o versight, this article articulates an e volutionary trajectory for HAPs- enabled LAE. The en visioned role of HAPs can be delineated into fiv e progressiv e stages: • F oundational infrastructure platform : HAPs serve as “aerial bases” that provide ubiquitous connectivity , edge computing, caching, and supervisory control, establishing the fundamental infrastructure required to support large- scale LAE operations. • U A V -centric service enabler : Moving beyond basic in- frastructure, HAPs ev olve into dedicated service nodes for U A Vs, offering reliable communication, high-precision sensing, navigation integrity support, and mission of- floading. This stage emphasizes direct enablement of U A V performance and autonomy , marking the shift from passiv e platform to acti ve enabler . • Adaptive aerial–air–gr ound collaboration : In complex or disrupted environments, HAPs cooperate with UA Vs to provide localized cov erage extension, en vironmental awareness, and resilient networking. This stage highlights context-a ware and en vironment-driv en collaboration, ad- dressing scenarios where terrestrial and satellite systems are insufficient. • Swarm-scale orchestration hub : As U A V operations scale, HAPs transform into orchestration centers for heterogeneous UA V swarms, enabling task allocation, cooperativ e sensing, collision avoidance, and large-scale trajectory optimization. Here, the focus shifts from single- U A V services to system-lev el coordination and swarm intelligence. • Cross-lay er intelligent autonomy : Ultimately , HAPs in- tegrate with ground networks, cloud infrastructures, U A V collectiv es, and satellite constellations to form a space– air–ground–cloud closed-loop ecosystem. This holistic ar - chitecture enables real-time autonomy , adaptiv e optimiza- tion, and AI-driv en orchestration across layers, thereby steering the LAE to ward self-sustaining, intelligent, and resilient growth. Through this staged progression, HAPs are expected to transcend their initial role as aerial relays and become a piv otal cross-layer , autonomy-dri ven infrastructure, underpinning the large-scale, intelligent, and sustainable e volution of the LAE. The remainder of this article further elaborates on these stages, highlighting the key enabling technologies, open research challenges, and prospecti ve directions that will shape the future of HAPs-assisted LAE. I I . W H Y H A P S , N O T ( O N L Y ) L E O ? When en visioning the LAE, LEO satellites are often the first technology to be considered [3]. Indeed, dense LEO constella- tions pro vide global coverage, ensuring baseline connectivity across remote oceans, deserts, and international air corridors. Using a transportation analogy , LEO resembles a transnational expressway: highly effecti ve for long-distance connectivity , but less adept at managing localized congestion or dynamic traffic control at specific intersections. What the LAE funda- mentally requires, ho wever , is a regional-scale traf fic controller capable of fine-grained orchestration and safety management, this is the unique role of the HAPs. More specifically , the respectiv e roles and dif ferentiated impacts of LEO satellites and HAPs on the LAE can be contrasted as follo ws: • Latency and control loops. Orbiting hundreds of kilo- meters abov e Earth, LEO satellites achiev e lo wer latency than GEO systems, yet their tens-of-milliseconds delay remains insufficient for sub-second closed-loop opera- tions such as U A V swarm collision av oidance, post- disaster emergency response, or urban airspace manage- ment. Additional impairments, including frequent han- dov ers and Doppler shifts, further compromise stability . In contrast, HAPs stationed at approximately 20 km alti- tude offer millisecond-scale round-trip delays, approach- ing terrestrial performance and enabling rapid com- mand–sense–feedback loops with enhanced resilience. • Coverage and targeted services. Whereas LEO excels in ubiquitous macro coverage, HAPs specialize in localized, fine-grained services. A single HAPs can sustain “pri vate- network-grade” support across 200–500 km regions, mak- ing it ideal for urban clusters, mountainous v alleys, maritime corridors, major events, and post-disaster zones. Moreov er , HAPs can be rapidly deployed within hours to provide on-demand emergency cov erage, an agility that satellites inherently lack. • Edge intelligence and energy efficiency . Lo w-altitude missions are typically data-intensiv e and computation- heavy , dominated by high-resolution imaging and real- time video streams. Offloading all tasks to the cloud 3 Fig. 1: Overvie w of LAE applications and supporting networks. The applications of the LAE can be categorized into three core domains: communication, sensing, and services. Communication focuses on addressing issues such as insuf ficient cov erage, excessi ve latency , and discontinuous connectivity with pro viding information support. Sensing targets en vironmental monitoring and situational awareness, with an emphasis on data acquisition. Services highlight scheduling, regulation, and coordination, underscoring system-le vel integration and efficient operation. T o enable these applications, a comprehensiv e air–space–ground network is leveraged, whose advantages and limitations are summarized. or satellites is both delay- and energy-prohibiti ve, par- ticularly for U A V with limited onboard resources. With greater energy reserves and payload capacity , HAPs can host AI accelerators and edge servers, effecti vely acting as “computing clouds in the sky . ” By supporting task offloading, feature e xtraction, and in-situ data fusion, HAPs reduce UA V energy consumption while enhancing end-to-end efficienc y and robustness. • Airspace regulation and safety assurance. Unlike LEO satellites, which primarily function as global broadcast- ers, HAPs occupy a regulatory sweet spot, above civil aviation corridors yet belo w satellite orbits. From this vantage point, they can continuously monitor U A V tra- jectories, fuse heterogeneous sensing data, and predict boundary violations or collision risks. Coupled with ground-based UTM systems, HAPs serv e as “intelligent traffic lights in the sky , ” enforcing order and enhancing safety across congested low-altitude airspace. Crucially , this does not diminish the role of LEO satellites. Instead, LEO and HAPs are complementary: LEO provides the global backbone, deliv ering seamless, transnational connectiv- ity; HAPs deliv er regional orchestration, edge intelligence, and airspace governance, ensuring localized efficienc y and safety; U A V and ground systems perform last-mile sensing, ex ecution, and feedback. A representativ e paradigm is a three-layer global–regional–local architecture. For example, in disaster scenarios, LEO satellites handle macro-le vel coordination and cross-domain aggregation, HAPs over the af fected region provide real-time scheduling and edge computing, while U A V conduct street-level inspection, e vidence collection, and material deliv ery . Only through multi-layered cooperation, with each component leveraging its strengths, can the LAE achiev e ubiquitous coverage, effecti ve regulation, and rapid responsiv eness. W ithin this integrated ecosystem, HAPs emerge as the most application-proximate and indispensable enablers, transforming the global highways of LEO into regional-scale real-time governance and intelligent operations. I I I . H A P S A S S I S TAN C E I N L A E W ith the rapid expansion of the LAE, U A V are expected to perform increasingly di verse and mission-critical tasks in environments that are often dynamic, unstructured, and resource-constrained. Howe ver , the large-scale deployment of U A V is still constrained by bottlenecks in connectivity , r egula- tion , computational capability , and swarm-level coordination . HAPs, positioned in the stratosphere at approximately 20 km, offer a unique v antage point and resource pool to address these 4 challenges. Specifically , the assistance provided by HAPs in LAE can be categorized into four complementary dimensions: communication, sensing and regulation, computation of fload- ing, and cooperativ e intelligence, as shown is Fig.2. Fig. 2: HAPs as aerial super nodes in the LAE, enabling four key dimensions: Communication, Sensing & Regulation, Computation, and Cooperativ e Intelligence. A. Communication Assistance: Stable and W ide-Ar ea Aerial Connectivity In the LAE, UA V operations span diverse and complex en vironments, from emergency logistics over mountainous terrain to inspection tasks in dense urban corridors. Howe ver , traditional terrestrial cellular netw orks are constrained by limited coverage radius and vertical beam patterns, leading to frequent handovers and service interruptions. These factors create critical bottlenecks for mission continuity and real- time command reliability . HAPs offer a macro-scale com- munication anchor that complements and enhances ground infrastructure. Functioning as aerial macro base stations, HAPs extend coverage across hundreds of kilometers and signif- icantly reduce handover frequency . Moreover , they can be integrated with terrestrial networks to form dual-path pro- tected communication links, where the HAPs link seamlessly takes over when the ground link e xperiences blockage or interference. For U A V swarms, HAPs also enable centralized spectrum coordination and routing management, effecti vely mitigating intra-swarm interference and supporting scalable, cooperativ e connecti vity . As illustrated in Fig. 3, Ground BSs ensure low delay and strong signal to noise ratio (SNR) only within a short service radius, while LEO satellites of fer global access but at the cost of higher propagation delay and signal attenuation. HAPs achiev e a balanced performance, providing both low-latency and wide-area cov erage, along with more stable SNR ov er long distances. This makes HAPs particularly well-suited as the communication backbone for large-scale, mobile, and safety- critical U A V/eVTOL swarm operations in the LAE. B. Sensing, Re gulation, and Navigation Integrity Assistance: Airspace Safety and Inte grated Sensing–Communication The large-scale expansion of the LAE requires not only ubiquitous communications but also effecti ve regulation and safety assurance across densely populated and highly dynamic low-altitude airspace. Lev eraging its high-altitude v antage point, the HAPs can continuously monitor UA V trajectories and recognize abnormal behaviors across wide areas. By in- corporating onboard AI models, the HAPs can predict potential collision risks and issue graded early warnings, effecti vely functioning as an intelligent radar for airspace management. Beyond monitoring, the HAPs can employ integrated sens- ing–communication (ISA C) wav eforms [5] to simultaneously support UA V navigation, velocity estimation, and positioning while maintaining data transmission, thus reducing the need for dedicated sensing payloads. In complex environments such as canyons or dense urban areas, the HAPs can cooperate with U A Vs or reconfigurable intelligent surfaces (RIS) [6] to mitigate blind spots and improv e coverage continuity . In this sense, the HAPs serves not only as the communication backbone but also as the air traffic coordinator and safety guardian of the LAE. In addition to providing wide-area situational awareness, persistent HAPs-based sensing also enables navigation integrity support for UA V and eVT OL operations. By maintaining high-probability line-of-sight links and continuous trajectory supervision, HAPs can complement the Global Navigation Satellite System (GNSS) in challenging en vironments such as urban canyons or obstructed regions, thereby enhancing trajectory reliability and airspace trust. It is worth noting that such navigation assistance is real- ized at the system and regulation le vel, rather than through dedicated onboard positioning algorithms. Instead of pursuing standalone localization accuracy , HAPs contribute to naviga- tion integrity by jointly leveraging sensing, communication, and regulatory supervision to ensure trustworthy and conflict- free aerial operations. T o provide a more tangible vie w of this concept, Fig. 4 presents representati ve simulation results from our recent HAPs–U A V integrated sensing and communication (ISARAC) study [7]. Figure 4b sho ws the optimized U A V trajectory under a joint sensing–communication configuration. In this setup, the HAPs transmits an ISAC wav eform that simultaneously enables do wnlink communication and synthetic aperture il- lumination, while the U A V acts as a passi ve receiv er col- lecting echoes for bistatic SAR processing. The optimized flight path provides suf ficient aperture diversity for imaging without compromising the quality of the downlink channel. Figure 4c illustrates the overall system performance trend, highlighting how integrated resource allocation affects both throughput and sensing reliability under shared power and en- ergy constraints. Although the exact numerical values depend on parameter settings, the observed trends demonstrate that the cooperati ve operation between HAPs and U A Vs achiev es 5 Gr ound B S HAP LEO (a) SNR - Gr ound B S SNR - HAP SNR - LEO Capacity - Gr ound B S Capacity - HAP Capacity - LEO (b) Fig. 3: Comparison of communication characteristics across dif ferent infrastructures for eVT OL connectivity . (a) One-way propagation delay versus horizontal separation distance, (b) Corresponding SNR and achiev able capacity . balanced resource utilization, maintaining robust connectivity while enabling high-fidelity sensing. T ogether , these insights qualitativ ely validate the feasibility and efficiency of HAPs- assisted ISA C architectures in enhancing situational aware- ness, airspace safety , and wide-area connectivity within the LAE [7]. C. Computation and Offloading Assistance: Aerial Cloud for U A V Computing Many U A V applications, such as high-definition video streaming, target detection, and remote imaging, require inten- siv e computation. Howe ver , U A Vs are inherently constrained by limited onboard processing capability and battery capac- ity . This creates a fundamental tension between the sensing demands of the mission and the platform’ s av ailable computa- tional resources. T o address this limitation, HAPs can function as an “aerial cloud data center , ” providing computation of- floading services over a wide coverage area. In environments where terrestrial connectivity is weak or unav ailable (e.g., mountainous regions or post-disaster zones), the U A V can adopt a “heavy sensing, light reporting” strategy: raw data is captured onboard but offloaded to the HAPs for feature extraction, inference, and recognition, while only essential results are fed back. This approach ensures task continuity while av oiding excessi ve onboard energy consumption. In addition, HAPs can jointly optimize computa- tion–caching–communication scheduling [8]. During low-traf fic periods, data or AI model pre-processing can be conducted proacti vely at the HAPs, whereas real-time computation tasks are prioritized during peak mission windows. This coordinated architecture ensures the U A V maintains reliable computational support ev en under tight energy budgets or heavy sensing loads. As sho wn in Fig. 6, relying solely on local computation sev erely restricts the U A V’ s computable task region. W ithout HAPs support (Fig. 5a), only tasks with relati vely small data sizes can be completed within the latency bound T max , while a large portion of tasks become infeasible (red region), reflecting strict limits in onboard computing and battery resources. When HAPs of floading is enabled (Fig. 6a), a substantial portion of these previously infeasible tasks becomes executable (green region). The feasible region expands simultaneously tow ard larger task sizes and greater UA V–HAPs separation distances, highlighting the role of HAPs as an aerial cloud extending computation capability across wide-area airspace. The white dashed contours denote the end-to-end offloading delay , sho w- ing that offloading remains within the latenc y constraint over a broad range. Consequently , HAPs offloading is well-suited for time-sensiti ve U A V missions such as disaster response, en vironmental monitoring, and search-and-rescue. In essence, HAPs transforms the U A V from a resource-limited sensing node into a computation-augmented intelligent platform, en- abling more complex, data-intensi ve, and long-duration aerial missions. D. Cooperative Intelligence Assistance: Swarm Intelligence and Closed-Loop Autonomy As UA V deployments scale up, single-drone operations are ev olving into collaborativ e swarm missions. In such scenarios, HAPs functions as an aerial coordination tower , responsible for assigning inspection regions, balancing mission loads, and coordinating intra-swarm routing and spectrum access so as to prev ent link congestion and task conflict. Beyond task allocation and control signaling, HAPs also facilitate multi-modal sensing fusion, integrating heterogeneous mea- surements such as radar echoes, mmW ave returns, and optical imagery collected by U A V swarms. Performing fusion and consistency filtering at the HAPs ensures robust perception, particularly in complex or dynamic en vironments. At a higher architectural lev el, UA Vs, HAPs, and terrestrial/cloud infras- tructures form a three-tier edge–air–cloud autonomy hierarchy , where U A Vs perform fast front-end detection, HAPs ex e- cute regional reasoning and mission-level decision-making, and cloud servers handle global optimization and long-term strategy planning. This edge–air–cloud closed-loop autonomy framew ork enables U A V swarms to complete the full cycle of detection–sensing–planning–control within second-level la- tency , thereby achieving true swarm intelligence. As sho wn in Fig. 6, the total network throughput depends critically on how co-channel interference is managed. In the 6  0 R X Y Z V O A R  Imaging area Ground (a) (b) 20 25 30 35 40 6000 6500 7000 7500 8000 8500 9000 Sum rate(bit/s/Hz) HAPs-UAV ISARAC HAPs only Communication (c) Fig. 4: Representativ e simulation results showing (a) the system model of a cooperati ve HAPs–U A V bistatic ISARA C scenario, (b) the optimized UA V trajectory that enhances synthetic aperture di versity while maintaining a reliable downlink for ISA C illumination from the HAPs, and (c) the ISA C performance trend illustrating balanced throughput and sensing robustness under shared ener gy constraints. L ocal O ffl oad Dr op (a) 0.1 0.1 0.2 0.2 0.3 0.3 0.5 0.5 0.8 L ocal O ffl oad Dr op (b) Fig. 5: Impact of HAPs-enabled computation offloading on U A V task executability . (a) Without HAPs. (b) With HAPs offloading support . No coordination baseline, U A Vs randomly select channels, causing sev ere mutual interference as swarm size increases. The Po wer-only strategy reduces interference by lo wering transmit po wer to meet SNR targets, b ut this results in ov erly conserv ati ve spectrum usage, of fering only marginal throughput gains. In contrast, the proposed HAPs spectrum coordination strategy performs joint link-quality estimation, channel assignment, and bandwidth allocation at the HAPs lev el. By exploiting U A V spatial distribution and traf fic load, the HAPs minimizes co-channel interference and reallocates bandwidth proportionally to achiev able link rates. As a result, the netw ork maintains scalable throughput gro wth e ven at large swarm sizes, significantly outperforming both baselines. E. Summary These four modules directly address the most urgent de- mands of the LAE, namely reliable communication, secure regulation, computational empo werment, and swarm intelli- gence. Importantly , they are not isolated but mutually reinforc- #UA Vs (a) Fig. 6: Sum throughput performance under different interference management strategies. ing: communication ensures stable data exchange, sensing and regulation provide safety guarantees, computation offloading 7 enhances efficienc y and endurance, while cooperati ve intel- ligence enables scalable swarm-le vel autonomy . Collecti vely , these capabilities enable HAPs to function as aerial control anchors that jointly support connecti vity , intelligence, trusted navigation and airspace governance for large-scale and safety- sensitiv e lo w-altitude operations. As a result, the integration of HAPs establishes an intelligent aerial substrate that does not merely supplement terrestrial networks, but fundamentally reshapes the operational resilience, scalability , and intelligence of the LAE. I V . E V O L U T I O NA RY R OA D M A P O F H A P S I N T H E L A E The dev elopment of the LAE is not a sudden leap, but rather a process of gradual accumulation and stepwise transforma- tion. W ithin this ev olving ecosystem, U A V can be conceptu- alized as playing two complementary roles. On the one hand, they operate as service providers, deliv ering communication, sensing, and other functionalities to ground users and IoT devices. On the other hand, they also act as service consumers, depending on ground base stations, HAPs, or LEO satellites for navig ation, connectivity , and control. Existing research has primarily focused on the former role [9], whereas the latter remains largely une xplored [10], thereby highlighting a critical research gap. Against this backdrop, HAPs are expected to undergo a paradigm shift: moving beyond their traditional role as in- frastructure providers to emerge as intelligent decision-makers that orchestrate U A V operations and ensure safe, reliable, and adaptiv e LAE services. This transformation can be system- atically delineated into fiv e progressiv e stages, outlining an ev olutionary roadmap that spans from basic platform functions to full autonomy . A. Stag e I: Infrastructur e Layer (Aerial Base) Driv en by the lack of unified coverage and fragmented terrestrial infrastructures, the initial stage positions HAPs as stable aerial bases. They serve as wide-area access points for UA V and ground users, supporting computation offloading and caching distribution, while establishing unified spectrum regulation and airspace monitoring capabilities. At this stage, HAPs function as the “6G mega-to wers in the sky”, consol- idating fragmented communication and computing resources into an integrated aerial network. B. Stag e II: Super Backstage for UA V As U A V missions become increasingly heterogeneous and computation-intensiv e, HAPs upgrade their role into a su- per backstage. Beyond providing stable uplink and down- link connectivity , HAPs employ ISA C wa veforms to enable simultaneous navigation and sensing. UA V can also offload computation-intensiv e imaging and detection tasks to HAPs for rapid processing. In this stage, HAPs ev olve beyond being mere “base stations” and act as super servers for UA V . C. Stag e III: F r ontline Support for Gr ound Users Motiv ated by the urgent demand for resilient connectivity in disaster and obstructed environments, HAPs begin to directly serve ground users. They can be rapidly deployed to restore communications in disaster areas. In valleys or dense urban en vironments, the y ensure reliable signal delivery . At this stage, HAPs transition from backstage support to frontline support , becoming indispensable for the functioning of ground society . D. Stag e IV : Swarm-Scale Coordination W ith the exponential gro wth of UA V fleets and the limita- tions of individual scheduling, HAPs emerge as aerial coor- dination towers for large swarms. They centrally orchestrate spectrum and routing to prevent congestion and interference. Lev eraging global situational awareness, HAPs dynamically allocate mission areas, ensuring efficient and orderly U A V collaboration. Here, HAPs transcend the role of communi- cation nodes to become the neural centers of UA V swarm intelligence. E. Stag e V : Full Autonomy with Edge–Air–Cloud Syner gy Driv en by the need for real-time intelligence and seamless multi-layer inte gration, HAPs integrate with ground edge nodes and cloud platforms to form a three-tier computational synergy architecture. U A V perform real-time detection, HAPs conduct re gional reasoning and mission-lev el decision-making, while the cloud executes global optimization. Through this closed loop, the entire system completes the sense–decide–act cycle within second-level latency , achieving true high-level autonomy . This signifies that HAPs are no longer merely aerial support platforms , but the intelligent brain of the LAE ecosystem. The ev olutionary roadmap of HAPs in the LAE illustrates a gradual yet transformativ e trajectory: from filling infrastruc- ture gaps and enhancing U A V missions, to enabling resilient ground services, orchestrating swarm-scale operations, and ultimately achieving full autonomy through edge–air–cloud synergy . Each stage reflects not only a response to pressing technical challenges but also a step tow ard a broader vision of intelligent and sustainable low-altitude ecosystems. By reshaping how U A V and ground systems interact, HAPs are set to become the intelligent backbone of the LAE, guiding both technological innov ation and policy directions toward a safer , more resilient, and economically vibrant future. V . F U T U R E R E S E A R C H D I R E C T I O N S Although HAPs have already demonstrated unique potential in the LAE, this remains a rapidly ev olving research area. The future development of HAPs will not be limited to acting as “communication relays” or “aerial bases, ” but will advance to ward higher levels of intelligence, coordination, and sustainability . The following four directions outline the core research pathways for the next phase. 8 1) AI-Driven HAPs Intelligent Scheduling Future HAPs will no longer rely on static deployment but will possess capabilities for autonomous trajectory prediction and dynamic scheduling. By lev eraging deep reinforcement learning (DRL) and large-model-based scenario forecasting, HAPs can adapti vely adjust altitude, position, and beam dis- tribution according to UA V activity density , communication demand distrib ution, and en vironmental dynamics, thereby realizing on-demand co verage. This intelligent scheduling transforms HAPs into truly adaptiv e network nodes, rather than passiv e base stations. 2) HAPs–LEO Cooperative NTN The cooperation between HAPs and LEO satellites repre- sents a key trend for future Non-T errestrial Networks (NTNs). LEO satellites provide global coverage, while HAPs deliv er fine-grained regional services. Their integration enables the construction of a holistic space–air network that combines global and regional adv antages. In this architecture, LEO satellites handle intercontinental transmission and large-scale broadcasting, while HAPs provide localized, high-throughput, low-latenc y management. For applications such as cross- border logistics, maritime surveillance, and post-disaster emer- gency response, this complementary NTN paradigm will be essential for achieving truly seamless connectivity . 3) Inte grated Airspace–Spectrum–Computation Management W ith the exponential growth of UA V deployments, future challenges will extend beyond communication links to encom- pass the integrated management of airspace, spectrum, and computation. HAPs are expected to become piv otal nodes for UTM: they can allocate airspace to avoid trajectory conflicts, while dynamically coordinating spectrum and computational resources to ensure efficient swarm-le vel task ex ecution un- der resource constraints. Such cross-dimensional orchestration positions HAPs as the grand coordinators of the LAE. 4) Gr een and Sustainable HAPs For HAPs to support large-scale applications, issues of energy and cost must be addressed. Future research will focus on solar-po wered designs, lightweight balloons, and cost- effecti ve aerostats to achiev e long-endurance station-keeping and reusable deployment. Combined with intelligent energy management and eco-friendly materials, HAPs will evolv e tow ard low-carbon and sustainable platforms. This direction not only aligns with global trends in green communications but also significantly lowers the threshold for scaling the LAE. In summary , the future ev olution of HAPs will be charac- terized by four intertwined trajectories: intelligence (AI-driv en autonomy), globalization (HAPs–LEO cooperation), integra- tion (joint airspace–spectrum–computation management), and sustainability (green energy and lo w-cost platforms). These re- search directions will determine whether HAPs can transition from experimental testbeds into the core infrastructure that sustains the LAE at scale. V I . C O N C L U S I O N The LAE is emer ging as the next technological and indus- trial frontier . T o truly unlock its potential, howe ver , a stable and intelligent aerial infrastructure is indispensable. In this process, HAPs play a dual role, serving both as the aerial base and as the coordination tower for airspace governance. Compared with LEO satellites, HAPs are much closer to users and thus capable of providing millisecond-level real- time communication. Leveraging their stratospheric vantage point, HAPs also assume the functions of airspace supervision and risk prediction. Equipped with onboard computing and caching resources, they further enable intelligent processing and task offloading for U A Vs and ground users. Through the integration of communication, computing, and regulation, HAPs further enable navigation integrity and airspace trust as emergent system-level properties, paving the way toward safe, scalable, and intelligent deployments of the LAE. At the application le vel, the synergy between HAPs and U A Vs unleashes tremendous potential. HAPs can not only provide communication assurance and sensing offloading for U A V swarms but also be rapidly deployed in disaster-stricken areas to restore connecti vity , perform sensing imaging, and deliv er frontline support for rescue and logistics. Beyond this, HAPs complement LEO satellites, terrestrial cellular systems, and cloud computing resources to form a three- tier global–regional–local architecture: LEO satellites ensure global connectivity , HAPs enable regional fine-grained man- agement, and U A Vs ex ecute localized detection and opera- tions. Looking ahead, HAPs are poised to become piv otal nodes that connect air traffic management, intelligent logistics, and emergenc y response. They will not only enhance the scala- bility of LAE operations but also dri ve the ecosystem toward high autonomy and sustainable growth. From functioning as an aerial base to ev olving into an intelligent brain, HAPs are reshaping the foundational logic of the LAE and accelerating its transition from exploration to large-scale, autonomous deployment. R E F E R E N C E S [1] A. A. Zaid, B. E. Y . Belmekki, and M.-S. Alouini, “eVTOL commu- nications and networking in U AM: Requirements, key enablers, and challenges, ” IEEE Communications Magazine , vol. 61, no. 8, pp. 154– 160, 2023. [2] M. Song, Y . Lin, J. W ang, G. Sun, C. Dong, N. Ma, D. Niyato, and P . Zhang, “T rustworthy intelligent networks for low-altitude economy , ” IEEE Communications Magazine , vol. 63, no. 7, pp. 72–79, 2025. [3] S. He, J. W ang, Y .-C. Liang, G. Sun, and D. Niyato, “Satellite- assisted lo w-altitude economy networking: Concepts, applications, and opportunities, ” arXiv pr eprint arXiv:2505.04098 , 2025. [4] Z. Lou, B. E. Y . Belmekki, and M.-S. Alouini, “HAPS in the non- terrestrial network nexus: Prospective architectures and performance insights, ” IEEE W ir eless Communications , vol. 30, no. 6, pp. 52–58, 2023. [5] J. T ang, Y . Y u, C. Pan, H. Ren, D. W ang, J. W ang, and X. Y ou, “Co- operativ e ISA C-empowered low-altitude economy , ” IEEE T ransactions on W ireless Communications , vol. 24, no. 5, pp. 3837–3853, 2025. [6] M. Ahmed, A. A. Soofi, F . Khan, S. Raza, W . U. Khan, L. Su, F . Xu, and Z. Han, “T oward a sustainable low-altitude economy: A survey of energy-ef ficient RIS-UA V networks, ” IEEE Internet of Things Journal , pp. 1–1, 2025. 9 [7] B. Huang and M.-S. Alouini, “Joint 3d trajectory and power allocation for HAPs–U A V Bistatic ISARAC systems, ” IEEE Internet of Things Journal , 2025,Submitted. [8] Z. Jia, Q. W u, C. Dong, C. Y uen, and Z. Han, “Hierarchical aerial computing for internet of things via cooperation of haps and uavs, ” IEEE Internet of Things Journal , vol. 10, no. 7, pp. 5676–5688, 2023. [9] P . Kanani, M. J. Omidi, M. Modarres-Hashemi, and H. Y anikomeroglu, “Optimizing network performance and resource allocation in haps-uav integrated sensing and communication systems for 6g, ” IEEE Tr ansac- tions on W ireless Communications , pp. 1–1, 2025. [10] Z. Y an, H. Zhou, J. Pei, and H. T abassum, “Hierarchical and collab- orativ e llm-based control for multi-UA V motion and communication in integrated terrestrial and non-terrestrial networks, ” arXiv preprint arXiv:2506.06532 , 2025.