Source-linked AI summary
SkyShare: Constellation-wide Sky Sharing for LEO-Radio Astronomy Coexistence
Farzad Mehri, Dara Ron, Nishanth Sastry, Satyaki Roy, Vijay K. Shah
TL;DR
LEO constellations threaten radio astronomy because existing reactive avoidance does not safely account for aggregate sidelobe interference and can create large coverage losses. SkyShare uses predictive, constellation-wide beam scheduling with EPFD modeling and observatory data; across 25 RAS sites, it substantially reduces service disruption while protecting observations.
Problem
LEO constellations expose radio astronomy sites worldwide to interference, while local avoidance can miss aggregate sidelobe emissions and sacrifice coverage.
Method
SkyShare combines orbital prediction, ITU-compliant EPFD modeling, observatory data, and an EPFD-budgeted Region of Interest to schedule spot beams through a flow-based algorithm.
Results
Across 25 radio astronomy sites, SkyShare substantially reduces service disruption compared to beam avoidance while maintaining protection for scientific observations.
Takeaways & Limitations
Predictive, network-wide coordination improves coverage while protecting RAS operations, unlike reactive avoidance that can cause substantial coverage loss without proportional interference reduction.
Takeaways & Limitations
SkyShare relies on predictable satellite motion and radio astronomy observations scheduled in advance for proactive assignment planning.
Abstract
from arXiv · showhide
Rapidly growing low-Earth-orbit (LEO) constellations increasingly operate in the spectrum shared with radio astronomy services (RAS), creating escalating interference risks for sensitive scientific observations. Existing mitigation mechanisms rely on reactive beam steering, or avoidance near observatories but fail to account for aggregate sidelobe emissions-leading to residual interference and substantial, unnecessary capacity loss. We present SkyShare, a constellation-wide sky-sharing system that enables predictive, interference-aware spot beam scheduling to protect radio astronomy while preserving network coverage. SkyShare integrates high-fidelity orbital prediction with International Telecommunication Union (ITU)-compliant Equivalent Power Flux Density (EPFD) modeling, and real-time observatory data via Operational Data Sharing (ODS) to jointly optimize beam-cell assignments over observation windows. To make constellation-scale coordination tractable, we introduce a concept of EPFD-budgeted Region-of-Interest(RoI) that bounds residual sidelobe interference while confining optimization to a minimal, provably sufficient set of cells. Building on RoI, we formulate LEO-RAS coexistence as a scalable scheduling problem and design SkySched, a flow-based algorithm that is optimal in special cases and yields scalable near-optimal solutions in the general NP-hard setting. SkyShare operates entirely in the control plane and requires no satellite hardware changes. Using real Starlink constellation geometries, we evaluate SkyShare across 25 single-dish, Ku-band RAS sites worldwide. Compared to Starlink boresight avoidance, SkyShare reduces unserved cells by up to 90.68% while remaining within EPFD limits.
1 Introduction
LEO constellations expand connectivity but expose radio astronomy services to interference that reactive, local avoidance cannot reliably control. SkyShare reframes coexistence as predictive, constellation-wide scheduling that coordinates satellite–cell assignments to protect observations while preserving coverage.
- Motivation: LEO constellations make every location vulnerable to radio-frequency interference, threatening sensitive radio astronomy observations.RAS instruments detect extremely weak cosmic signals and require stringent interference protection.
- Current practice: Existing Starlink-style avoidance uses operational observatory data but reacts through beam disabling or steering near RAS sites.The TBA approach illustrates this threshold-based exclusion strategy.
- Current practice: Aggregate sidelobes from many satellites can violate EPFD limits even when distant cells individually contribute little interference.This limits the protection achievable through local geographic restrictions.
- Current practice: Reactive avoidance creates persistent coverage gaps, disproportionately affecting rural and remote users while residual sidelobe interference can continue limiting RAS sensitivity.Disabling more surrounding cells yields diminishing EPFD reduction because aggregate sidelobes dominate.
- SkyShare insight: SkyShare uses predictable satellite motion, scheduled observations, and satellite diversity to reassign spot beams proactively across the constellation.SkySched jointly accounts for future positions, observation schedules, and aggregate interference in its assignments.
- Contributions: The paper introduces SkySched, a flow-based scheduler that is optimal in special cases and near-optimal in general settings.SkyShare formulates coexistence scheduling under EPFD constraints and uses SkySched for scalable operation.
2 Background and Preliminaries
This section describes LEO spot-beam operation, RAS observations, EPFD-based protection, and Telescope Boresight Avoidance (TBA). It also identifies aggregate sidelobe interference as a limitation of reactive avoidance.
- LEO Satellites and Spot Beam Operation: LEO constellations use electronically steerable phased-array antennas to allocate multiple narrow spot beams across ground locations.Each satellite can form up to N_sb independent beams simultaneously.
- LEO Satellites and Spot Beam Operation: Each quasi-Earth-fixed cell is served by at most one spot beam at a time.Spot beams continuously steer fixed ground footprints during brief satellite overpasses.
- RAS Observations and ODS: RAS sites vary widely in antenna size, geographic distribution, and satellite visibility, while observations are scheduled in advance and metadata can be shared through ODS.ODS publishes telescope location, pointing direction, schedules, and frequency bands for near-real-time coexistence adaptation.
- RAS Protection Criteria: Time-averaged EPFD aggregates instantaneous interference over an observation and must remain below the harmful threshold throughout its duration.The observation is discretized into time slots to capture interference from moving satellites.
- Telescope Boresight Avoidance (TBA): TBA disables or redirects beams near telescope boresights but does not explicitly validate the EPFD constraint.Its inner and outer boresight rules use local angular or distance thresholds.
- Telescope Boresight Avoidance (TBA): Aggregate sidelobe emissions from satellites outside TBA regions can remain harmful, limiting protection while reactive beam actions create coverage gaps.The problem worsens as satellite and user populations scale.
3 SkyShare System
SkyShare uses observatory metadata, orbital prediction, and EPFD modeling to schedule constellation-wide beam–cell assignments predictively. Its SkySched scheduler enforces time-averaged EPFD constraints while accounting for traffic demand to reduce coverage loss.
- Design Goals: SkyShare targets strict EPFD compliance, minimal coverage loss, and predictive operation for LEO–RAS coexistence.The system is designed to protect RAS sites while preserving nearby network coverage.
- System Overview: SkyShare operates in the operator cloud control infrastructure over scheduling windows such as Starlink Gen2-mini’s 15-second intervals.Satellite–cell assignments are computed for each window.
- System Overview: ODS supplies telescope location, pointing direction, operating frequency, and observation schedule to identify when and where protection is required.These metadata are consumed by the constellation-wide scheduler.
- System Overview: The Constellation State Predictor propagates internal orbital parameters to compute precise satellite positions throughout each scheduling interval.These predicted states support visibility and interference calculations.
- System Overview: SkyShare computes satellite visibility, relative telescope geometry, window-level EPFD contributions, and an interference-relevant Region of Interest around each observatory.The RoI confines scheduling decisions to cells that materially affect interference.
- SkySched: SkySched jointly assigns satellite beams to cells using visibility, observation metadata, EPFD thresholds, and cell-level demand.It enforces time-averaged EPFD constraints while seeking minimal coverage loss.
4 Interference Modeling and Protection Constraints
The section models constellation-wide LEO interference at radio astronomy sites using ITU-compliant EPFD and enforces time-averaged protection constraints. An EPFD-budgeted Region of Interest confines optimization while bounding residual sidelobe interference from non-RoI cells.
- Interference Sources: Four interference regimes combine satellite and RAS main-lobe or sidelobe interactions, with aggregate satellite sidelobe emissions potentially producing non-negligible interference.Satellite sidelobe-to-RAS-main-lobe contributions can be significant because of the telescope’s high main-lobe gain.
- Interference Sources: Aggregate sidelobe interactions become increasingly important with constellation density, making RAS protection a constellation-wide rather than local geometric problem.Existing threshold-based boresight avoidance fails to account for this aggregate interference source.
- EPFD Modeling: EPFD represents the equivalent single-source power flux density producing the same RAS main-beam received power as aggregate interference from multiple satellites.Instantaneous EPFD is formed by an incoherent linear sum of active spot-beam contributions, with time-sharing represented by normalized beam capacity.
- Protection Constraint: RAS protection requires time-averaged EPFD over the observation duration to remain below the observation-specific detrimental threshold TH_RAS.The threshold depends on observation type, bandwidth, required detection performance, and source visibility; exceeding it is interpreted as RAS data loss.
- EPFD-Budgeted RoI: The EPFD-budgeted RoI assigns residual budget ΔT_H_RAS to non-RoI cells and optimizes only cells whose joint assignments could exceed that budget.Smaller residual budgets strengthen guarantees but enlarge the optimization scope, while the selected RoI can be the minimum set whose complement fits within the budget.
5 LEO–RAS Coexistence Problem Formulation
The paper formulates LEO–RAS coexistence as maximizing served ground cells during an observation while respecting satellite, cell-service, handover, and time-averaged EPFD constraints. The resulting mixed-integer optimization is NP-hard.
- Objective: The objective maximizes the number of ground cells whose traffic demand is fully served throughout a radio astronomy observation.The optimization considers cells within the EPFD-budgeted Region of Interest.
- Demand Model: Each cell’s normalized demand represents the fraction of full spot-beam capacity required on average over the observation duration.A demand of 1 requires continuous service from one full spot beam, while zero-demand cells are excluded.
- Assignment Constraints: Satellite capacity, single-satellite service, and spot beam–cell consistency constrain assignments across time intervals and scheduling windows.Each satellite has at most N_sb simultaneous spot beams, and each cell is served by at most one satellite under the stated constraints.
- Network Interpretation: The flow-network representation uses satellite–cell assignment arcs, with arc costs capturing EPFD contributions and capacities enforcing spot-beam limits.The network layers correspond to scheduling intervals or windows.
- Complexity: Problem P0 is NP-hard.This motivates the flow-based special-case solution and heuristic treatment of the general case.
6 SkySched Algorithm
SkySched converts interference-aware satellite–cell assignment into min-cost flow over a time-expanded network. It is exact under restricted unit-window, uniform-demand conditions and uses iterative feasibility repair for the general NP-hard case.
- Algorithm Overview: SkySched models satellite–cell assignments as min-cost flow, with EPFD interference represented by edge costs and spot-beam availability by capacities.The algorithm is centralized and polynomial-time for its flow-based procedure.
- Special Case: Under N_w = 1 and unit demand, P0 reduces to interference-minimization subproblems solvable by min-cost flow.The special case permits associations to change every interval and serves cells with full spot beams or not at all.
- Special Case: The largest integer q satisfying the EPFD threshold gives the optimal objective value of P0 under the special-case assumptions.Any assignment achieving q★ is optimal for the original special-case problem.
- Special Case: For fixed integer q, the minimum-cost q-unit flow yields exactly EPFD★(q), establishing an exact solution for the special-case subproblem.Feasible problem assignments correspond one-to-one with integral q-unit flows.
- General Case: For heterogeneous demand and multi-interval windows, SkySched uses scaled min-cost flow followed by feasibility repair and repeated capacity updates.Overloaded satellite-window assignments are pruned, while underused capacity can trigger re-solving until convergence or an iteration limit.
- Execution: SkySched computes orbital geometry and EPFD contributions from TLE and ODS data, then searches over flow demand to return the best feasible assignment found.The procedure accounts for antenna patterns and satellite–cell–RAS off-axis angles within the RoI.
7 Evaluation
SkyShare is evaluated against TBA and DTBA using Starlink Gen2-mini simulations across 25 Ku-band RAS sites. It reduces coverage loss while maintaining EPFD protection, improves sensitivity, and provides scalable near-optimal scheduling with measurable robustness limits.
- Coverage and EPFD Interference Tradeoff: SkyShare reduces total unserved cells by over 90.68% on average across 25 Ku-band RAS sites compared with TBA and DTBA.The evaluation reports substantially lower service disruption while maintaining strict protection for observation sites.
- Sensitivity: At full coverage (U = 0), SkyShare achieves a 0.264 Jy sensitivity floor versus 0.530 Jy for TBA, a 3 dB improvement.At U = 5, SkyShare provides up to 15 dB better sensitivity.
- Geographic and Physical Factors: At 60° latitude, SkyShare improves coverage by 94.85% relative to TBA and 92.72% relative to DTBA.For telescope diameters above 30 m, SkyShare achieves full coverage under the 0.4 Jy EPFD threshold; for 10 m telescopes, it reduces unserved cells by 99.33%.
- Scalability and Optimality: SkyShare scales polynomially as Gurobi runtime grows exponentially with RoI size.The flow-based scheduler achieves over 99% of the Gurobi optimum across tested instances, including heterogeneous-demand cases.
- Robustness to System Errors: Pointing errors up to 12° produce 0% mean EPFD violations, while 30 km satellite position error produces mean violations exceeding 80%.SkyShare outperforms TBA and DTBA across pointing-error levels but is sensitive to satellite position error.
8 Discussion and Limitations
SkyShare’s gains arise from predictive, constellation-wide coordination, but its current design assumes centralized control, single-beam service, and limited network and measurement scope. Several extensions remain open, including multi-beam, multi-operator, multi-observatory, and broader emission modeling.
- Discussion: Predictable satellite motion and scheduled observations enable proactive coordination that improves coverage while protecting RAS operations.The discussion contrasts this with reactive methods that treat satellites independently and ignore aggregate sidelobe interference.
- Operational assumptions: SkyShare assumes centralized control-plane scheduling, ODS metadata, and assignments fixed within each scheduling window.Distributed operation is left as a future extension, and evaluated 15-second windows bound reassignment and handover opportunities.
- Single-Beam vs. Multi-Beam Service: The current model does not exploit the larger feasible assignment set of multi-beam architectures and would require splittable-flow formulation or demand discretization.Schedules remain feasible under multi-beam service, but splitting demand increases scheduling complexity.
- Limited Networking Scope: The evaluation excludes latency, queueing, power-control dynamics, inter-cell interference, and multi-operator scenarios.These network-level objectives and multi-operator settings remain future work.
- System Model Fidelity: Poorer measured sidelobe suppression can enlarge the RoI or reduce the available EPFD budget, potentially increasing runtime or unserved cells.Validation with operational traffic traces and real-world network and antenna measurements remains future work.
- Interference Scope: The current interference scope excludes leakage into adjacent protected bands, harmonic emissions, and spurious emissions outside the satellite operating band.Frequency-dependent emission masks and separate EPFD constraints remain to be incorporated.
- Multiple simultaneous RAS observations: The evaluation protects one RAS observation at a time, while simultaneous observatory observations would introduce overlapping EPFD constraints.Joint optimization across many simultaneous observations is identified as an important direction.
9 Related Work
Prior work addresses LEO–RAS coexistence through operator–observatory efforts and beam-avoidance mechanisms, while dynamic spectrum-sharing research largely targets terrestrial opportunistic access. SkyShare differs by coordinating satellite assignments predictively across the constellation.
- LEO Satellite and RAS Coexistence: Prior LEO–RAS work includes detected intended and unintended emissions and mechanisms whose availability impacts are stronger at lower frequencies, smaller antennas, or wider spot-beam footprints.Wider footprints create larger angular exclusions and affect satellites for longer periods.
- Dynamic Spectrum Sharing: Dynamic spectrum-sharing frameworks commonly study terrestrial cognitive radio and database-driven access rather than constellation-wide LEO–RAS coordination.The passage introduces this literature as a separate research context for spectrum sharing.
10 Conclusion
SkyShare combines predictive orbital and observatory information with EPFD-aware scheduling to protect RAS while preserving LEO coverage. Across 25 sites, it reduces service disruption relative to beam avoidance, supporting constellation-wide coordination as a practical coexistence approach.
- Conclusion: SkyShare integrates real-time ODS data, orbital prediction, and ITU-compliant EPFD modeling into control-plane scheduling.Its EPFD-budgeted RoI supports MILP formulation and a flow-based algorithm with optimal special cases and scalable near-optimal general solutions.
- Conclusion: Across 25 radio astronomy sites, SkyShare substantially reduces service disruption compared with beam avoidance while maintaining RAS protection.The conclusion presents constellation-wide coordination as practical and necessary for sustainable satellite–RAS spectrum sharing.
A.1 Starlink Downlink Antenna Model
The Starlink downlink model divides 2000 MHz into eight channels and represents each satellite antenna as a phased array serving multiple spot beams. Beam gain is modeled from the element gain and array factor.
- 2000 MHz of aggregate spectrum is divided into 8 channels of 250 MHz each.
- Each gen2-Mini satellite has 5 downlink phased-array antennas, each illuminating 8 spot beams with 2 polarizations, for 80 beams total.
- Each downlink antenna is modeled as a 25 × 40 half-wavelength uniform rectangular array approximating a 34 dB boresight gain.
- Spot-beam gain is specified as the product of single-element gain and array factor for a two-dimensional rectangular array.
A.2 Satellite beam Capacity Model
The capacity and scheduling formulation combines received-power and achievable-capacity modeling with EPFD-aware assignment constraints. The supplied material then characterizes computational complexity and flow-based solution methods, including a special-case optimality result.
- Capacity model: Satellites dynamically control EIRP so ground-level Ku-band PFD remains near the regulatory maximum across elevation angles.
- Capacity model: The capacity model uses spot-beam capacity, received power, effective receiver aperture, noise temperature, and satellite elevation to evaluate service feasibility.
- Deployment: SkyShare recomputes beam–cell assignments periodically from predicted satellite states and ODS metadata using rolling-horizon control-plane scheduling.
- Deployment: SkyShare operates in the control plane without satellite hardware or onboard beamforming changes and is presented as compatible with existing LEO architectures.
- Deployment: Scheduling windows of several seconds to tens of seconds can track predictable constellation dynamics while limiting configuration churn and enabling proactive deployment.
- Complexity and algorithms: The general scheduling problem is NP-hard via a reduction from Bin Packing, even when the EPFD constraint is non-binding.
- Complexity and algorithms: For fixed served demand, a flow graph maps feasible assignments to integral flows, allowing min-cost flow to minimize EPFD in the special case.
- Complexity and algorithms: The special-case procedure precomputes EPFD values, searches over served demand, and uses feasibility tests to obtain the maximum feasible served count.