A Case for Application-Aware Space Radiation Tolerance in Orbital Computing

TL;DR

This paper proposes a cost-effective, application-aware approach to enhance space radiation tolerance in in-orbit AI computing, specifically for DNN tasks on COTS hardware, by exploiting layer sensitivity and spatial correlation to mitigate radiation errors.

Contribution

It introduces RedNet, a novel method that leverages domain knowledge of DNN layer sensitivity and spatial correlation to improve radiation tolerance without heavy resource overhead.

Findings

RedNet reduces radiation error influence to approximately zero.

It accelerates in-orbit DNN inference by 8.4% to 33%.

Validation on satellite payloads confirms effectiveness and efficiency.

Abstract

We are witnessing a surge in the use of commercial off-the-shelf (COTS) hardware for cost-effective in-orbit computing, such as deep neural network (DNN) based on-satellite sensor data processing, Earth object detection, and task decision.However, once exposed to harsh space environments, COTS hardware is vulnerable to cosmic radiation and suffers from exhaustive single-event upsets (SEUs) and multi-unit upsets (MCUs), both threatening the functionality and correctness of in-orbit computing.Existing hardware and system software protections against radiation are expensive for resource-constrained COTS nanosatellites and overwhelming for upper-layer applications due to their requirement for heavy resource redundancy and frequent reboots. Instead, we make a case for cost-effective space radiation tolerance using application domain knowledge. Our solution for the on-satellite DNN tasks, \name, exploits the uneven SEU/MCU sensitivity across DNN layers and MCUs' spatial correlation for lightweight radiation-tolerant in-orbit AI computing. Our extensive experiments using Chaohu-1 SAR satellite payloads and a hardware-in-the-loop, real data-driven space radiation emulator validate that RedNet can suppress the influence of radiation errors to $\approx$ 0 and accelerate the on-satellite DNN inference speed by 8.4%-33.0% at negligible extra costs.