Resource-Aware Task Allocator Design: Insights and Recommendations for Distributed Satellite Constellations

A new research paper presents the design and analysis of a Resource-Aware Task Allocator (RATA), a system for intelligently distributing real-time computing tasks across constellations of low Earth orbit (LEO) to medium Earth orbit (MEO) satellites. The study, led by researcher Bharadwaj Veeravalli, uses a Single-Level Tree Network (SLTN) architecture to evaluate key performance metrics like blocking probability, response time, and energy consumption under varying traffic loads. It simulates processing for tens of thousands of tasks, monitoring critical parameters including on-board compute, storage, bandwidth, and the impact of satellite eclipses on communications.

The findings reveal a pronounced non-linear scaling effect: while overall system capacity increases with the number of satellites, blocking rates and delays grow rapidly beyond a certain point. Counterintuitively, the analysis demonstrates that energy consumption remains relatively resilient under solar-aware scheduling. The primary cause of system failure shifts from energy constraints to pure CPU availability as the constellation scales. This research provides the first quantitative guidance for satellite constellation designers, identifying the precise thresholds at which a distributed space-based computing system transitions from graceful performance degradation to a complete operational collapse.