arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2602.04803eess.SP
收藏

Safe-NEureka: 一种用于卫星 onboard AI 处理的混合模块冗余 DNN 加速器

Safe-NEureka: a Hybrid Modular Redundant DNN Accelerator for On-board Satellite AI Processing

University of Bologna(博洛尼亚大学);Fondazione Chips-IT(Chips-IT基金会);ETH Zürich(苏黎世联邦理工学院)

Riccardo Tedeschi, Luigi Ghionda, Alessandro Nadalini, Yvan Tortorella, Arpan Suravi Prasad, Luca Benini, Davide Rossi, Francesco Conti

AI总结:

Safe-NEureka 是一种用于卫星 onboard AI 处理的混合模块冗余 DNN 加速器,结合冗余保护与性能优化,适用于安全关键和性能关键任务。

摘要:

低地球轨道(LEO)星座正在革新航天领域, onboard 人工智能(AI)成为下一代卫星的关键。AI 加速对于安全关键功能如自主导航、导航和控制(GNC)至关重要,其中错误不可容忍,而对于高带宽传感器数据的性能关键处理,偶尔错误是可以容忍的。因此,卫星的 AI 加速器必须结合对辐射诱导故障的鲁棒保护与高吞吐量。本文提出了 Safe-NEureka,一种用于异构 RISC-V 系统的混合模块冗余深度神经网络(DNN)加速器。它运行在两种模式下:一种是利用双模块冗余(DMR)的冗余模式,通过硬件恢复;另一种是性能模式,利用冗余数据路径最大化并行吞吐量。此外,其内存接口通过纠错码(ECC)保护,控制器通过三模块冗余(TMR)保护。在 GlobalFoundries 12nm 技术上实现,冗余模式下故障执行减少了 96%,且有可接受的 15 面积开销。在性能模式下,架构在 3x3 密集卷积上实现接近基线速度,吞吐量为 5,效率减少 11%,相比冗余模式的 48 和 53。这种灵活性确保高开销仅限于关键任务,使 Safe-NEureka 成为适用于航天应用的多功能解决方案。

英文摘要:

Low Earth Orbit (LEO) constellations are revolutionizing the space sector, with on-board Artificial Intelligence (AI) becoming pivotal for next-generation satellites. AI acceleration is essential for safety-critical functions such as autonomous Guidance, Navigation, and Control (GNC), where errors cannot be tolerated, and performance-critical processing of high-bandwidth sensor data, where occasional errors are tolerable. Consequently, AI accelerators for satellites must combine robust protection against radiation-induced faults with high throughput. This paper presents Safe-NEureka, a Hybrid Modular Redundant Deep Neural Network (DNN) accelerator for heterogeneous RISC-V systems. It operates in two modes: a redundancy mode utilizing Dual Modular Redundancy (DMR) with hardware-based recovery, and a performance mode repurposing redundant datapaths to maximize parallel throughput. Furthermore, its memory interface is protected by Error Correction Codes (ECCs), and the controller by Triple Modular Redundancy (TMR). Implementation in GlobalFoundries 12nm technology shows a 96 reduction in faulty executions in redundancy mode, with a manageable 15 area overhead. In performance mode, the architecture achieves near-baseline speeds on 3x3 dense convolutions with a 5 throughput and 11 efficiency reduction, compared to 48 and 53 in redundancy mode. This flexibility ensures high overheads are limited to critical tasks, establishing Safe-NEureka as a versatile solution for space applications.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑