Operational Implications of Dynamic Power Throttling on High-Performance LEO Edge Computing

The integration of high-performance COTS (Customer Of The Shelf) processors, such as the Nvidia Jetson Orin NX, into SmallSat architectures enables advanced on-orbit edge computing but introduces significant power management challenges. Standard manufacturer power profiles typically require system reboots and lack the granularity needed to respond to dynamic LEO (Low Earth Orbit) energy constraints, such as eclipse phases or high-priority payload activation.

This paper presents a software-defined power controller developed for the EDGX Sterna platform that enforces strict, variable power ceilings in real-time without interrupting active services. The core research focuses on the operational implications of this dynamic throttling, specifically regarding data buffering and throughput recovery during energy-critical states. When power is capped below the requirement for a real-time workload (e.g., Radio Frequency Signal processing), constant data ingress outpaces processing capacity, necessitating buffering.

We detail a series of characterization experiments designed to attempt to model the non-linear relationship between enforced power limits and processing rates for static, stateless workloads. By sweeping through granular power limits, we attempt to derive a transfer function that maps available wattage to precise throughput values. We then demonstrate how this function can be used to algorithmically predict the “recovery window”, the exact duration required at a nominal power state to process both the live data stream and the accumulated backlog.

These findings offer a methodology for autonomous mission planning, allowing satellite operators to maximize data throughput and manage buffer sizes while strictly adhering to highly variable energy budgets.