Skip to content

The 1st International Workshop on Federated Learning at the Orbital Edge: Intelligence across Space, Air and Ground (OrbitFL'26)

Co-located with IoT 2026 · November 17–20, 2026 · Newcastle upon Tyne, United Kingdom

Call for Papers

Edge intelligence is expanding beyond terrestrial networks. Increasingly affordable and capable satellites, high-altitude platforms, and unmanned aerial vehicles are transforming the Internet of Things into a three-dimensional space–air–ground infrastructure in which sensing, training, and inference happen progressively closer to the data source. Low Earth Orbit constellations alone are projected to grow from roughly 10,000 active satellites in 2024 to an estimated 100,000 by 2030, and an expanding share of these nodes will process data on board rather than relaying it to terrestrial stations.

Federated and distributed learning offer a natural paradigm for this setting: exchanging model updates instead of raw observations preserves data locality and reduces traffic over scarce, low-bandwidth links. Moving this paradigm into orbit, however, introduces constraints that terrestrial deployments rarely face at the same intensity. The network topology changes continuously as satellites move, ground contacts are short and intermittent, on-board power varies with solar illumination and eclipse cycles, hardware is heterogeneous and must tolerate radiation effects, and the data collected across regions and sensors is inherently non-IID and subject to temporal drift.

OrbitFL'26 brings together researchers from the IoT, edge AI, satellite systems, and remote-sensing communities to address these challenges in a dedicated venue within IoT 2026. We welcome contributions spanning theory, systems design, and real-world applications across the full space–air–ground continuum.

Topics of Interest

Topics of interest include, but are not limited to:

  • Federated, decentralised, and hierarchical learning for LEO satellite constellations, high-altitude platforms, and UAV swarms
  • Orbital edge computing: on-board training, aggregation, and inference close to the data source
  • Energy-efficient and sustainable on-device learning under limited, time-varying power budgets (solar/eclipse cycles, battery-aware scheduling)
  • Communication-aware learning over intermittent, heterogeneous links (RF, laser inter-satellite links, NTN, ground-to-satellite, UAV–satellite)
  • Predictive contact planning, client selection, and resource-aware scheduling under orbital dynamics
  • Learning with non-IID data, concept drift, and temporal drift: clustered and personalised federated learning across sensors and regions
  • Label-efficient learning: semi-supervised, self-supervised, and active learning for remote sensing with scarce, delayed, or noisy labels
  • Reconfigurable and accelerator-based on-board ML (FPGA, GPU), quantization, pruning, sparsification, and lightweight model compression
  • Robust and secure aggregation under radiation faults, stragglers, and adversarial conditions
  • Applications: Earth observation, environmental monitoring, wildfire and flood detection, and disaster response via space–air–ground IoT

Important Dates

All deadlines are Anywhere on Earth (AoE).

  • Paper submission deadline: August 20, 2026
  • Notification of acceptance: September 20, 2026
  • Camera-ready paper due: October 2, 2026
  • Workshop date: November 17, 2026

Submission Guidelines

Authors are invited to submit original research papers of up to 6 pages (including figures and references), formatted according to the ACM Primary Article Template (sigconf format, double-column). Papers must be submitted in PDF format.

All submissions will undergo a single-blind review process. Each paper will receive at least three reviews from members of the Programme Committee. Submitted work must be original and must not be under review at any other venue.

Authors should submit their papers via the EasyChair submission page.

Publication

Accepted papers will be published in the companion proceedings of IoT 2026 and made available through the ACM Digital Library. At least one author of each accepted paper must register for and attend the workshop to present the work. A Best Paper Award, selected by the Workshop Co-Chairs on the basis of Programme Committee review scores, will be conferred on one of the presented papers.

Programme

OrbitFL'26 is planned as a half-day workshop (approximately 3.5 hours). The tentative programme includes:

  • An invited keynote at the intersection of federated learning and non-terrestrial systems
  • Two paper sessions (20-minute presentations plus 5 minutes for questions)
  • A moderated panel and roadmap discussion collecting open problems and charting directions for future editions

The detailed programme will be published after the notification of acceptance.

Organization

Workshop Chairs

Mark Adrian Gambito Mark Adrian Gambito
University of Messina, Italy
Kanaka Sai Jagarlamudi Kanaka Sai Jagarlamudi
Western Sydney University, Australia
Nan Yang Nan
Yang
Western Sydney University, Australia

Program Committee

To be announced.

Contact

For any enquiry regarding OrbitFL'26, please contact the Workshop Co-Chairs: