What Is Aerial Computing?

Aerial computing is a distributed computing paradigm in which processing, storage, and networking resources are carried by airborne and space-based platforms rather than by fixed ground infrastructure. It combines aerial radio access networks with edge computing, so that an unmanned aerial vehicle, a high-altitude platform station, or a low Earth orbit satellite acts simultaneously as a communication relay and as a compute node for the devices below it. The paradigm emerged from work on unmanned aerial vehicle communications during the 2010s and was consolidated as a distinct research area in surveys such as the treatment of aerial computing as a computing paradigm, which organizes the field into low-altitude, high-altitude, and satellite computing layers alongside conventional terrestrial cloud and edge tiers.

The motivation is coverage rather than raw throughput. Terrestrial edge servers are anchored to sites with power and backhaul, which leaves maritime regions, mountainous terrain, disaster zones, and sparsely populated areas without nearby computation. An airborne node can be repositioned within minutes, establishes a line-of-sight channel to ground terminals, and can be withdrawn when demand falls.

Platform Tiers

Low-altitude platforms are rotary-wing and fixed-wing unmanned aircraft operating from tens to a few hundred meters, offering short round-trip latency and fine-grained positioning but limited endurance and payload. High-altitude platform stations, typically solar-powered aircraft or stratospheric balloons at roughly 20 kilometers, trade latency for persistence and cover footprints tens of kilometers across. Satellite constellations in low Earth orbit extend the hierarchy to global scale with propagation delays of a few milliseconds one way and with onboard processors constrained by radiation tolerance and thermal budget. A survey of aerial mobile edge computing compares the three tiers on deployment flexibility, coverage area, and cost per served user.

Computation Offloading and Trajectory Design

The central optimization problem is deciding what to compute where. A ground device may execute a task locally, offload it to an aerial node, or relay it further to a terrestrial cloud, and the choice depends on task deadline, data volume, channel quality, and residual battery on both ends. Because the aerial node moves, the channel gain is a function of the flight path, which couples the offloading decision to trajectory planning and makes the joint problem non-convex and often mixed-integer. Published formulations apply successive convex approximation, Lyapunov optimization for queue stability, game-theoretic bargaining, and deep reinforcement learning, and are catalogued in an ACM Computing Surveys review of resource management in unmanned aerial vehicle edge computing. Energy is the binding constraint throughout, since propulsion power for a multirotor typically exceeds its computing power by an order of magnitude.

Integration with Space-Air-Ground Networks

Aerial computing is usually described as one layer of a space-air-ground integrated network, in which orbital, atmospheric, and terrestrial segments share a common control plane. Research in this direction addresses handover between fast-moving nodes, service migration so that a running container follows a user as coverage changes, federated learning across nodes that cannot share raw data, and security under intermittent connectivity. A survey of air computing in 6G wireless networks sets out the architectural interfaces and the open standardization questions, including how orchestration frameworks designed for static data centers should be adapted to platforms with finite flight time.

Applications

Aerial computing has applications in a range of fields, including:

• Disaster response, where ground infrastructure is damaged or absent
• Precision agriculture and livestock monitoring over large fields
• Maritime and offshore connectivity for vessels and platforms
• Real-time video analytics for search, rescue, and infrastructure inspection
• Temporary capacity augmentation at large public events
• Internet of Things data collection in remote sensor deployments