Markdown Version | Transcript | Session Recording | Session Materials

SPACE

Summary

The proposed SPACE Research Group (SPACE RG) met at IETF 126 to discuss long-term research challenges and architectures for space-based and non-terrestrial networks (NTNs). The session was chaired by Yorgos, with support from Nishanth Sastry. The agenda included four invited technical presentations and two group work updates.

Key topics addressed included the exponential growth of planned low Earth orbit (LEO) constellations, real-world uplink constraints and neighbor contention on LEO networks, the specific communication and computing needs of industrial NTN use cases, and routing protocol scalability in highly dynamic space topologies. The group also reviewed draft updates detailing a standard code notation for describing satellite constellations and a comprehensive taxonomy of space research tools and infrastructures.

Key Discussion Points

1. Administration and Introduction

• Slide Deck: Chairs - Introduction
• The Chairs opened the session, reviewed the IRTF Note Well guidelines, privacy/recording policies, and the code of conduct.
• Yorgos reminded participants that the group focuses on long-term research issues rather than standard development, aiming to foster collaboration on space-specific protocols and architectures.

2. Beyond Starlink: Understanding the Coming Waves of LEO Satellite Deployments

• Presenter: Dan York
• Slide Deck: Beyond Starlink: Understanding the Coming Waves of LEO Satellite Deployments
• Presentation Summary:
  • Currently, Starlink dominates the LEO space with ~12,612 satellites launched and plans for nearly 35,000. It also operates Star Shield (~1,600 satellites) for military applications.
  • A massive wave of new LEO constellations is being filed and developed globally:
    • Active/Imminent: Amazon LEO (Project Kuiper - ~8,000 planned); Eutelsat OneWeb (660 launched, 528 more filed); AST SpaceMobile and Link Global (direct-to-cell, utilizing very large satellites); Globalstar (1,946 filed, partnering with Apple and being integrated into Amazon LEO); Iridium (acquired by Rocket Lab to build vertical launcher/constellation integration).
    • China: Chinese state-backed Guangwang (14,000 planned, ~200 launched) and Shanghai-backed Thousand Sails (16,000 planned).
    • Russia: Rassvet (300 to 900 planned, ~22 launched).
    • Upcoming / Filed: Blue Origin's PteraWave (optical downlinks); EU's Iris 2; Telesat LightSpeed (Canada, ~1,600); Hanwha 3 (China, 10,000); Meridian Space (~1,200 using Spinlaunch kinetic launcher); India's Reliance Jio; Logos Space (4,000); and an additional Chinese filing for ~200,000 satellites.
    • Orbital Data Centers (ODCs): Filings have been submitted to host AI and compute centers in space. These include SpaceX (nearly 1.2 million satellites), Lumen Orbit/Star Cloud (88,000), Blue Origin's Project Sunrise, Orbital Compute (100,000), Cowboy Space (20,000), and Google's Project Suncatcher.
    • Aggregate scale: In total, filings track over 2.3 million planned satellites.
• Q&A and Discussion:
  • Launch Capacity & Sustainability: Yorgos and Dan York discussed how launch capability remains a major bottleneck. Given an orbital lifespan of 3–5 years at lower altitudes (<600 km), maintaining these constellations requires continuous, high-frequency launches. Atmospheric impacts of burning up dozens of satellites daily remain unknown.
  • Feasibility: Nitinder Mohan asked about the practical feasibility of these numbers over the next ten years. Dan noted that massive capital is flowing into the sector, particularly around AI applications (ODCs).
  • Governance and Regulation: Carlos asked about international coordination. Dan explained that filings go through national regulators (such as the US FCC or Germany's Bundesnetzagentur) which are then submitted to the ITU-R for spectrum and orbit coordination. Ali Rezaki (Sustain-RG co-chair) raised the question of whether the global population should have a larger say in space governance, given the scale of orbital deployment.
  • Interoperability: Kurtis Heimerl questioned how external researchers or the IETF can influence highly proprietary commercial systems. Rick Taylor responded that because these networks must ultimately connect to the broader Internet, the IETF's role is critical in ensuring heterogeneous compatibility and interoperability.

3. The Limitations of LEO Uplink

• Presenter: Liz Izhikevich
• Slide Deck: The Limitations of LEO Uplink
• Presentation Summary:
  • Real-world measurements were conducted using five Starlink Business Priority dishes on the roof of a Netflix building in Los Gatos, CA, uplinking UDP traffic.
  • While downlink capacities are high, ISPs allocate significantly less uplink (9:1 downlink-to-uplink ratio, typically 20–40 Mbps).
  • Key Findings:
    • Handover Loss: Single-dish uplinks experience severe multi-second bursts of packet loss (up to 80% loss for ~15 seconds) matching Starlink's 15-second topology reconfiguration/satellite handover intervals.
    • Contention: Localized neighbor contention causes significant packet loss when multiple nearby dishes contend for the same satellite's uplink capacity.
    • Bonding: Distributing traffic across multiple dishes (bonding) reduces average loss but does not eliminate shared loss events when all local dishes point to the same satellite.
    • Traffic Engineering Trick: Actively downloading at high rates (e.g., 200 Mbps) triggers the system's traffic engineering algorithms to allocate almost double the uplink capacity to the user.
• Q&A and Discussion:
  • Measurement Context: Warren Kumari noted that the Los Gatos area likely has high ambient dish density, and observed that the TE allocation model incentivizes users to generate artificial downlink traffic to secure uplink bandwidth.
  • Congestion Source: Tony Li asked if the losses were due to gateway congestion or satellite association. Liz confirmed that they mapped association data; the loss peaks occurred when all five dishes were forced onto the same satellite, and they confirmed it was not gateway-side congestion.
  • Global Capacity Limitations: Nalini Elkins asked about physical limits in remote versus urban areas. Liz noted that remote areas experience much better performance, pointing to existing theoretical work proving that LEO satellite capacity cannot scale to support dense urban populations due to orbital surface area constraints.
  • Handover Times: Daniel Gulch asked if the physical handovers could be shortened. Liz clarified that while the actual radio switch is under 10 ms, the resulting routing and configuration updates cause longer-lasting packet loss bursts.

4. Industrial Communication Services via Non-Terrestrial Networks: Requirements and Challenges

• Presenter: Florian Zeiger
• Slide Deck: Industrial Communication Services via Non-Terrestrial Networks: Requirements and Challenges
• Presentation Summary:
  • Siemens is exploring NTNs for critical industrial infrastructures, including railways (fleet management, train control), maritime, mining, smart grids, and remote plant operations.
  • Industrial traffic has unique requirements: massive telemetry (delay-tolerant, high packet volume), remote monitoring, supervisory control (sub-second delays, packet loss-sensitive), and closed-loop control (tens of milliseconds, strict packet ordering, low jitter).
  • Presented an experiment running a closed-loop robot control system (balancing a ball on a plate using Profinet) over an emulated LEO network where virtualized Programmable Logic Controllers (VPLCs) migrated between satellites.
  • Emphasized the need for "compute awareness" and joint optimization of communication, computing, and control. When network degradation is predicted, the application must gracefully adapt (e.g., slowing down the robot's movement).
• Q&A and Discussion:
  • Lars Eggert asked about potential APIs to expose network quality and predictions to applications. Florian agreed this is a critical research area, especially regarding how applications express their "service intent" to dynamic networks.

5. Towards Standardized Routing for LEO Satellite Networks: A Comparative Study

• Presenter: Lin Han
• Slide Deck: Towards Standardized Routing for LEO Satellite Networks: A Comparative Study
• Presentation Summary:
  • Analyzed routing options for LEO networks, particularly in 3GPP NTN architectures. Proactive routing is considered the most viable option for carrier/transport network functions due to hardware pre-programmed forwarding and performance demands.
  • Standard proactive terrestrial protocols (OSPF, IS-IS) cannot be applied directly to LEO networks because:
    • They are designed for relatively stable, hierarchical topologies of ~1,000 nodes, whereas LEO networks are flat, highly dynamic, and easily scale to 10,000+ nodes.
    • They struggle with interleaved orbital planes, requiring major enhancements in partitioning, convergence dependency, and flooding reduction.
  • Key development areas include Segment Routing (to minimize hop-by-hop convergence dependency), static satellite/ISL addressing, fast network state detection, and highly scalable simulation environments.
• Q&A and Discussion:
  • Interoperability: Rick Taylor questioned the necessity of standardizing routing protocols when commercial operators (like Starlink) deploy proprietary, single-vendor systems. Lin Han and Maxime Piraux argued that as direct-to-cell (3GPP) integration matures, multi-vendor interoperability at the backhaul and access boundary will become essential, especially for medium-sized constellations that cannot build entirely vertical, proprietary stacks.

6. Draft update: A Code to Describe Satellite Constellations

• Presenter: Maxime Piraux
• Slide Deck: Draft update: A Code to Describe Satellite Constellations
• Presentation Summary:
  • Presented updates to draft-piraux-space-constellation-code, which establishes a standard taxonomy and notation to model satellite constellations (access, core, and ground networks).
  • In version -02, the draft replaced YAML definitions with formal CDDL schemas to represent link patterns (in-orbit and cross-plane links) and ground station configurations (elevation, antennas).
  • Demonstrated two tools utilizing the taxonomy:
    • Aerospace Lab's internal tool which instantiates virtualized Mininet/Linux-based network emulators directly from the constellation code.
    • Juan A. Fraire's web-based, open-source Contact Plan Designer, which computes complex link contacts for both Earth and interplanetary network topologies.
• Q&A and Discussion:
  • Data Modeling: Eric Vyncke suggested separating the abstract data model (e.g., YANG or CDDL) from the specific string-based "code" serialization, allowing more flexible integration with network management tools.
  • Routing in Space: Ali Rezaki asked if routing in the satellite "core" is more energy-efficient than traditional "bent-pipe" satellite links. Maxime noted that while bent-pipe is more energy-efficient, a routed core enables richer services and ground-independent routing, which are required for modern use cases.

7. Draft update: A Typology of Space Research Infrastructures

• Presenter: Nishanth Sastry
• Slide Deck: Draft update: A typology of Space Research Infrastructures
• Presentation Summary:
  • Presented an update on a draft defining a taxonomy of space research infrastructures, initially conceived at the TAGS workshop.
  • The taxonomy categorizes tools into 10 classes: Simulators, Emulators, Testbeds, In-orbit platforms, Datasets, Measurement tools, Reference implementations, Libraries, Visualizers, and Research platforms.
  • The registry has expanded from 60 to 210 tools.
  • Announced a community-driven, machine-readable registry where researchers can add and update tools via Git pull requests. Each entry includes a "last verified" field to track active maintenance.

Decisions and Action Items

• No formal standardization or consensus decisions were made, in line with the group's status as a proposed Research Group (RG).

Next Steps

1. Constellation Notation Draft: The authors of draft-piraux-space-constellation-code invite community feedback on the CDDL schemas and general taxonomy on GitHub or the SPACE mailing list.
2. Space Typology Draft: Nishanth Sastry and Juan A. Fraire will publish the initial draft of the space research infrastructure typology on the mailing list and invite researchers to review and submit pull requests to populate the tool registry.
3. Interim Meeting: The Chairs plan to organize a virtual interim meeting in the fall (separate from the scheduled IETF meetings) to continue technical discussions. Details will be announced on the mailing list.

Related Documents

draft-piraux-space-constellation-code, draft-piraux-space-constellation-code-02-00, draft-update-a-typology-of-space-research-infrastructures-00