Edge Computing on Aerial and Orbital Platforms

44 concepts34 papers1 proposal

Traditional edge computing frameworks operate under a comfortable assumption: edge servers are fixed in stationary, well-connected terrestrial locations. However, as noted by Wenjing Xiao et al. in GraphEdge: Dynamic Graph Partition and Task Scheduling for GNNs Computing in Edge Network, this static paradigm fails to adapt to dynamic environments or underserved regions where terrestrial infrastructure is non-existent, congested, or destroyed.

At the same time, our ability to collect data from above has outpaced our ability to transmit it. Modern remote sensing instruments on satellites and unmanned aerial vehicles (UAVs) generate massive volumes of high-resolution data. Streaming these raw data streams back to centralized ground stations creates a severe communication bottleneck due to limited downlink bandwidth and high propagation latencies.

To solve this, a new paradigm is emerging: non-terrestrial edge computing. As detailed by Quoc-Viet Pham et al. in Aerial Computing: A New Computing Paradigm, Applications, and Challenges, this approach merges aerial access networks with edge computing to create a multi-tier, "in-the-sky" processing infrastructure. By converting aerial and orbital nodes from passive relays into active computational servers, processing power is placed directly at the data source.

Why does processing data in orbit make sense? The core intuition lies in data reduction. Downlinking raw sensor streams requires massive bandwidth and energy. As Chen Chen et al. observe in Special issue on advances in real-time image processing for remote sensing, shifting data processing from ground stations to on-board satellite systems minimizes communication bottlenecks by transmitting only processed results or high-precision subsets rather than raw data streams.

Consider the mechanism of on-board target detection. Instead of downlinking a multi-gigabyte hyperspectral image for a ground data center to find a specific object, the satellite runs local AI inference directly on its flight segment. According to Bing Zhang et al. in Progress and Challenges in Intelligent Remote Sensing Satellite Systems, converting raw imagery into high-level target information directly on-orbit reduces the data transmission burden from gigabytes to kilobytes. This allows for the real-time delivery of actionable insights, such as specific coordinates of detected objects.

Furthermore, satellites can perform what Shangguang Wang et al. term "in-orbit speculative inference" in Tiansuan Constellation: An Open Research Platform. This mechanism allows satellites to autonomously identify and discard low-utility data—such as images obscured by cloud cover—before it ever enters the downlink queue. The authors demonstrate that filtering out irrelevant data before transmission can reduce downlink bandwidth requirements by up to 90%, fundamentally improving the economic efficiency of data-heavy missions (James Murphy et al. in Machine Learning in Space: A Review of Machine Learning Algorithms and Hardware for Space Applications).

This architectural shift from a ground-centric to a flight-centric processing chain drastically reduces latency. For instance, Michael Kerr et al. in Advanced Data Chain Technologies for the Next Generation of Earth Observation Satellites Supporting On-Board Processing for Rapid Civil Alerts and Murray Kerr et al. in EO-ALERT: NEXT GENERATION SATELLITE PROCESSING CHAIN FOR RAPID CIVIL ALERTS demonstrate that generating products and alerts in-situ slashes the latency for critical civil alerts (such as extreme weather or ship detection) from hours to under five minutes.

To support these complex workloads, modern orbital platforms are adopting terrestrial software paradigms. Chao Wang et al. explore this in Satellite Computing: A Case Study of Cloud-Native Satellites, detailing how "cloud-native" satellites use containers and microservices to execute on-orbit AI inference. Similarly, deploying telecommunications infrastructure—such as the 5G User Plane Function (UPF)—directly onto satellite hardware enables local data routing and processing in space, minimizing signaling latency and reducing dependency on ground-based backhaul (Ruolin Xing et al. in From Earth to Space: A First Deployment of 5G Core Network on Satellite). As emphasized by C. Huang et al. in Suggestions on expanding application scenarios of commercial aerospace market, this shift toward space-based processing mitigates ground-to-space data bottlenecks and facilitates high-speed communication across satellite constellations.

The Aerial Edge: Dynamic, On-Demand Compute

Closer to Earth, unmanned aerial vehicles (UAVs) extend the edge computing paradigm to the atmosphere. While satellites provide global coverage, UAVs offer localized, highly flexible deployment. Instead of relying on distant cloud servers, ground devices can offload computation-heavy tasks to nearby flying base stations equipped with localized processing units or "cloudlets" (June-Woo Ryu et al. in Multi-Access Edge Computing Empowered Heterogeneous Networks: A Novel Architecture and Potential Works).

The primary advantage of an aerial mobile edge computing (MEC) server is its mobility. Unlike static base stations, a UAV can actively alter its position to establish direct line-of-sight communication paths and minimize the physical distance to ground users. This spatial flexibility is highly effective across several domains:

The Optimization Frontier: Trajectories and Energy Constraints

While shifting computation to aerial and orbital edges solves the communication bottleneck, it introduces a complex physical challenge: mobility and energy constraints. Unlike a terrestrial data center plugged into a power grid, a UAV operates on a strict battery budget, and a satellite relies on finite solar-charged batteries. Furthermore, because these edge servers are constantly moving, the channel quality between the ground user and the server changes continuously.

Therefore, the system's efficiency relies on solving a multi-dimensional optimization problem. As Zhiying Wang et al. explain in Multi-UAV Enabled MEC Networks: Optimizing Delay Through Intelligent 3-D Trajectory Planning and Resource Allocation, coordinating multiple mobile resources to minimize task processing delay requires the joint optimization of UAV trajectories, task offloading decisions, and computational resource allocation.

To maximize the operational lifetime of an aerial edge server, researchers must balance the energy spent on propulsion (flying) against the energy spent on communication and computation. Yuwen Qian et al. address this in User Association and Path Planning for UAV-Aided Mobile Edge Computing With Energy Restriction by jointly optimizing user association, flight paths, and ground transmission power. By mathematically modeling these trade-offs, the system ensures that the UAV moves to the most mathematically optimal position to service ground devices without exhausting its battery prematurely, proving that the success of non-terrestrial edge computing depends as much on mechanical trajectory planning as it does on algorithmic efficiency.

Key Foundational Work

MineFleet®: an overview of a widely adopted distributed vehicle performance data mining system (2010) — contributed 1 concept elements to this direction.
Object tracking in the presence of occlusions using multiple cameras: A sensor network approach (2013) — contributed 1 concept elements to this direction.
History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_orbital_edge_computing_1ba61889,
  title = {orbital edge computing},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/orbital-edge-computing/rdr_1ba6188986094f529bee1f615e00116d},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-19},
}

This exact version (version 1, frozen text):

@misc{lacuna_orbital_edge_computing_1ba61889_v1,
  title = {orbital edge computing},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/orbital-edge-computing/rdr_1ba6188986094f529bee1f615e00116d/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-19},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake