FPGA-Based Neural Network Accelerators for Space Applications: A Survey

2025-04-22

AI Summary

FPGA-based neural network accelerators for space missions must simultaneously satisfy stringent real-time performance, ultra-low power consumption, and radiation hardness requirements. This paper presents a systematic survey of 32 onboard FPGA+NN co-design works, identifying— for the first time—three critical technology gaps in space applications: architecture mapping under resource and reliability constraints, inadequate EDA toolchain support for radiation-aware design, and lack of on-orbit learning capability. To bridge these gaps, we propose a holistic co-design methodology integrating fixed-point/sparse NN mapping, SEU-tolerant circuit design, dynamic partial reconfiguration optimization, and space-grade verification. Key technical advances include a novel radiation-induced soft-error mitigation mechanism and an efficient runtime scheduling strategy for reconfigurable resources. The resulting framework delivers the first engineering-ready technology roadmap for deep-space intelligent payloads, significantly enhancing on-board autonomous decision-making, real-time sensor analytics, and lossless/lossy data compression capabilities.

Technology Category

• Machine Learning: Hardware-aware ML
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Space missions are becoming increasingly ambitious, necessitating high-performance onboard spacecraft computing systems. In response, field-programmable gate arrays (FPGAs) have garnered significant interest due to their flexibility, cost-effectiveness, and radiation tolerance potential. Concurrently, neural networks (NNs) are being recognized for their capability to execute space mission tasks such as autonomous operations, sensor data analysis, and data compression. This survey serves as a valuable resource for researchers aiming to implement FPGA-based NN accelerators in space applications. By analyzing existing literature, identifying trends and gaps, and proposing future research directions, this work highlights the potential of these accelerators to enhance onboard computing systems.

Problem

• Surveying FPGA-based NN accelerators for space missions
• Addressing high-performance onboard computing needs
• Exploring radiation-tolerant neural network implementations

Innovation

• FPGAs enable flexible radiation-tolerant space computing
• Neural networks enhance autonomous space mission tasks
• Survey guides FPGA-NN accelerator research gaps