cs.DCMar 4, 2026

Constraint-Aware Execution Planning for Hybrid Space-Ground Compute Workloads

Authors: Subhadip Mitra

Organizations: RotaStellar

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Low Earth orbit (LEO) satellites increasingly carry compute hardware capable of on-board processing, yet each satellite generates roughly two orders of magnitude more data than it can downlink per orbit. This mismatch forces operators to decide, for every workload, which computation runs on-board and which runs on the ground, how intermediate data crosses the space-ground boundary through narrow contact windows, and how to maintain delivery guarantees over noisy channels. We present Constraint-Aware Execution (CAE), a planning system that takes a satellite identifier, a workload expressed as a directed acyclic graph of processing steps, and a set of orbital and resource constraints, and produces a deterministic, physically grounded execution plan. CAE operates in four phases: (1) orbital environment construction via SGP4 propagation with eclipse detection and ground station pass prediction, (2) compute placement using a cost model that compares on-board resource consumption against transfer overhead, (3) transfer insertion with adaptive forward error correction and security overhead modeling, and (4) greedy first-fit scheduling into orbital windows under power, thermal, compute, and communication constraints. We evaluate CAE against five representative workload patterns across satellites in distinct orbital regimes and demonstrate that the system produces feasible plans in under two seconds, correctly exploits onboard data reduction to minimize transfer volume, and adapts FEC and multi-pass allocation to varying channel conditions. CAE is deployed as a production API computing plans for any cataloged satellite using live two-line element data.

Figures & tables

Figure 1: CAE four-phase planning pipeline. The orbital environment computed in Phase 1 is consumed by all subsequent phases. Each phase refines the plan incrementally: placement decides where steps run, transfer insertion adds data movement steps, and scheduling assigns everything to orbital windows.
Figure 2: Cylindrical shadow model for eclipse detection. A satellite is in eclipse when it lies on the anti-sun side of Earth ( d<0d<0 ) and its perpendicular distance from the Earth-Sun line is less than Earth’s radius ( r⊥<R⊕r_{\perp}<R_{\oplus} ). The model treats the shadow as an infinite cylinder, neglecting penumbra and atmospheric refraction.
StationLatLonProvider
Svalbard †78.2315.39KSAT
Troll−- 72.012.53KSAT
Awarua−- 46.53168.38KSAT
Fairbanks †64.86−- 147.85NASA
Wallops37.94−- 75.47NASA
McMurdo−- 77.85166.67NASA
Table 1: Ground station network. All stations support S-band and X-band; stations marked with † also support Ka-band.
ParameterValue
Carrier frequency8.2 GHz
Transmit power10 dBW
Tx antenna gain6 dBi
Rx antenna gain34 dBi
Implementation loss2 dB
Atmospheric loss0.5 dB
Table 2: Default X-band link parameters.
Channel BERFEC RateOverhead
>10−5>10^{-5}1/2100%
>10−7>10^{-7}3/433%
≤10−7\leq 10^{-7}7/814%
Table 3: Adaptive FEC rate selection based on channel BER.
Figure 3: Split learning workload DAG after placement and transfer insertion. Feature extraction achieves 40:1 data reduction on-board, minimizing downlink volume. Transfer steps (dashed) are automatically inserted at space-ground boundaries by Phase 3. Data volumes include FEC and security overhead.
WorkloadStepsDL (MB)UL (MB)Conf.
ML Inference510.500.99
Split Learning1052.87.60.99
EO with QA10560.000.99
Federated123.75.80.99
Store-Forward7157.500.99
Table 4: Plan metrics across workloads for ISS (NORAD 25544). Duration is wall-clock time from first step to last step completion.
Figure 4: Example schedule for the split learning workload on the ISS. The top row shows power state (sunlit at 80W, eclipse at 25W). The middle row shows ground station pass windows. The bottom row shows the scheduled execution: on-board steps (cyan) are placed in sunlit windows for full compute capacity, transfers (orange) are pinned to pass windows, and ground compute (yellow) begins immediately after data arrives. Abbreviations: SVB = Svalbard, FAI = Fairbanks, AWA = Awarua, SNG = Singapore, ORE = Oregon.
Figure 5: Transfer volume breakdown (MB) by workload. Raw data (cyan) is the payload after on-board processing. FEC overhead (red) depends on channel quality. Security and CCSDS framing (purple) adds encryption and packet headers.
Figure 6: Data reduction cascade for the on-board ML inference workload. Each processing step reduces data volume, with inference achieving 50:1 compression. The overall 190:1 reduction means only 10.5 MB needs to be downlinked from the original 2 GB capture. Bar heights use a compressed scale for visibility.
More in this topic

Computation Offloading

See all 15 papers

Explore similar work

CardsList
Apr 14, 2026

Optimizing Earth Observation Satellite Schedules under Unknown Operational Constraints: An Active Constraint Acquisition Approach

Earth Observation (EO) satellite scheduling (deciding which imaging tasks to perform and when) is a well-studied combinatorial optimization problem. Existing methods typically assume that the operational constraint model is fully specified in advance. In practice, however, constraints governing separation between observations, power budgets, and thermal limits are often embedded in engineering artefacts or high-fidelity simulators rather than in explicit mathematical models. We study EO scheduling under unknown constraints: the objective is known, but feasibility must be learned interactively from a binary oracle. Working with a simplified model restricted to pairwise separation and global capacity constraints, we introduce Conservative Constraint Acquisition~(CCA), a domain-specific procedure designed to identify justified constraints efficiently in practice while limiting unnecessary tightening of the learned model. Embedded in the Learn&Optimize framework, CCA supports an interactive search process that alternates optimization under a learned constraint model with targeted oracle queries. On synthetic instances with up to 50tasks and dense constraint networks, L&O improves over a no-knowledge greedy baseline and uses far fewer main oracle queries than a two-phase acquire-then-solve baseline (FAO). For n≤30n\leq 30, the average gap drops from 65--68% (Priority Greedy) to 17.7--35.8% using L&O. At n=50n{=}50, where the CP-SAT reference is the best feasible solution found in 120s, L&O improves on FAO on average (17.9% vs.\ 20.3%) while using 21.3 main queries instead of 100 and about 5×5\times less execution time.

Mohamed-Bachir Belaid

NILU, Climate and environmental research institute, Kjeller, Norway

Constrained OptimizationLearning to OptimizeRemote Sensing
AskPDF
Mar 19, 2026

iSatCR: Graph-Empowered Joint Onboard Computing and Routing for LEO Data Delivery

Sending massive Earth observation data produced by low Earth orbit (LEO) satellites back to the ground for processing consumes a large amount of on-orbit bandwidth and exacerbates the space-to-ground link bottleneck. Most prior work has concentrated on optimizing the routing of raw data within the constellation, yet cannot cope with the surge in data volume. Recently, advances in onboard computing have made it possible to process data in situ, thus significantly reducing the data volume to be transmitted. In this paper, we present iSatCR, a distributed graph-based approach that jointly optimizes onboard computing and routing to boost transmission efficiency. Within iSatCR, we design a novel graph embedding utilizing shifted feature aggregation and distributed message passing to capture satellite states, and then propose a distributed graph-based deep reinforcement learning algorithm that derives joint computing-routing strategies under constrained on-board storage to handle the complexity and dynamics of LEO networks. Extensive experiments show iSatCR outperforms baselines, particularly under high load.

Jiangtao Luo, Bingbing Xu, Shaohua Xia +1

the School of Communications and Information Engineering, Chongqing University of Posts and Telecommunications, Chongqing, 400065 China.

LEO Satellite NetworksEdge ComputingReinforcement Learning
AskPDF
Mar 19, 2026

Which Workloads Belong in Orbit? A Workload-First Framework for Orbital Data Centers Using Semantic Abstraction

Space-based compute is becoming plausible as launch costs fall and data-intensive AI workloads grow. This paper proposes a workload-centric framework for deciding which tasks belong in orbit versus terrestrial cloud, along with a phased adoption model tied to orbital data center maturity. We ground the framework with in-orbit semantic-reduction prototypes. An Earth-observation pipeline on Sentinel-2 imagery from Seattle and Bengaluru (formerly Bangalore) achieves 99.7-99.99% payload reduction by converting raw imagery to compact semantic artifacts. A multi-pass stereo reconstruction prototype reduces ~306 MB to ~1.57 MB of derived 3D representations (99.49% reduction). These results support a workload-first view in which semantic abstraction, not raw compute scale, drives early workload suitability.

Durgendra Narayan Singh

Independent Researcher, Twin Falls, ID, USA

Topic pending
AskPDF