嵌入式SoC单粒子翻转软件容错与快速启动设计

Software Fault Tolerance and Fast Boot Design for SEU-resilient Embedded SoCs

摘要

在空间辐射环境中,高能粒子轰击Flash存储器易诱发单粒子翻转错误,导致数据静默破坏,严重威胁系统的稳定性.传统错误纠正码技术虽然具备良好的软件容错能力,但其解码开销较高,常成为性能瓶颈.为兼顾关键数据的正确性与启动效率,本文提出一种面向系统启动的高效容错架构.该架构基于辐射诱发错误的空间局部性特征,设计错误聚集映射机制,实现高风险错误位的集中管理,并通过统计分析动态更新错误分布模型;同时引入校验-解码分层机制,结合轻量级循环冗余校验与高强度解码策略,减少冗余解码操作.实验结果表明,本文提出的基于错误局部性的高效启动容错架构,在确保启动数据正确性的前提下,显著提升系统在辐射环境下的启动速率.

Abstract

In space radiation environments,high-energy particle strikes frequently induce single-event upsets,leading to silent data corruption that threatens system stability and reliability.Although conventional error correction codes provide strong software-level fault tolerance,their high decoding overhead often becomes a performance bottleneck during system startup.To ensure the correctness of critical data while improving boot efficiency,this work proposes an efficient fault-tolerant mechanism tailored for the startup phase.Leveraging the spatial locality of radiation-induced errors,an error aggregation mapping mechanism is designed to centrally manage high-risk bit errors.A layered verification-decoding strategy is further introduced,combining lightweight cyclic redundancy checks with high-strength decoding to reduce redundant decoding operations.Additionally,an error-bit mapping reconstruction mechanism dynamically updates the error distribution model through statistical analysis.Experimental results demonstrate that the proposed locality-aware mechanism significantly accelerates boot time under radiation while ensuring startup data integrity.

关键词

软件容错 / 高效启动 / 嵌入式SoC / 错误聚集映射 / 分层校验

Key words

software fault tolerance / efficient boot / embedded SoC / error aggregation mapping / layered verification

[1] CHEN X N.Research on ingle event upset tolerant design methods for SRAM-based FPGAs[D].Mianyang:Southwest University of Science and Technology,2024.
[2] Herbert E,Sega R,McGraw J,et al.Risk management for commercial-off-the-shelfparts based space hardware[J].Systems Engineering,2025,28(3):374-392.
[3] WANG F.Design of an ICAP-based single event upset mitigation system for SRAM-based FPGAs[D].Beijing:University of Chinese Academy of Sciences,2022.
[4] Ibe E H,Yoshimoto S,Yoshimoto M,et al.Radiation-induced soft errors[M].Tokyo:Springer Japan,2018.
[5] XU W L,LI K X,XU Z,et al.Single event effect hardening techniques for RISC-V-based systems[J].Electronics & Packaging,2025,doi:10.16257/j.cnki.1681-1070.2026.0023.
[6] Bastos R P,Picas M G,Velazco R.Medium-earth-orbit spaceflight radiation effects in triple modular system on programmable device[J].IEEE Transactions on Nuclear Science,2023,70(8):1783-1790.
[7] Paparsenos E,Tsiatouhas Y.Radiation-hardened latch design with triple-node-upset recoverability[J].AEU-International Journal of Electronics and Communications,2024,187:155539,doi:10.1016/j.aeue.2024.155539.
[8] Vasilikou K,Gogolou V,Noulis T.Radiation-hardened cells for analog design:comprehensive review and design flow[C]//Proceedings of the 13th International Conference on Modern Circuits and Systems Technologies(MOCAST),2024:1-4.
[9] Almazmomi N K.Enhanced reed-solomon error correction framework for reliable data transmission in noisy communication systems[J].Internet Technology Letters,2025,8(6):e634,doi:10.1002/itl2.634.
[10] Kakizaki T,Nakamura M,Kisaka Y.Channel-polarized multilevel coding for low-complexity forward error correction[J].Journal of Lightwave Technology,2025,43(10):4555-4563.
[11] Xu C,Wu L,Liu W,et al.Block code design based on golay code for future communication systems[C]//Proceedings of the IEEE 99th Vehicular Technology Conference(VTC2024-Spring),2024:1-6.
[12] Maity R K,Samanta J,Bhaumik J.An improved single and double-adjacent error correcting codec with lower decoding overheads[J].Journal of Signal Processing Systems,2023,95(6):721-733.
[13] Shanthi G,Rao K S,Kumar A S,et al.Multi-bit error detection and correction codes implementation using generator-parity check matrices technique[J].Microsystem Technologies,2025,31(2):509-520.
[14] Cui L,Wu F,Zhang M,et al.Enhanced LDPC coding for 3D TLC NAND flash memory:leveraging RBER difference from intra-layer variation[J].IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems,2025,doi:10.1109/TCAD.2025.3598931.
[15] Rao N P,Desai M P.Neutron induced strike:on the likelihood of multiple bit-flips in logic circuits[J].arXiv preprint arXiv:1612.08239,2016.
[16] Pintong K P.Towards lightweight real-time identification,classification,and detection of algorithm behavior in embedded systems via the power side channel[D].Binghamton,NY:State University of New York at Binghamton,2025.
[17] Chen Z,Song S,Wei Z,et al.Approximating median absolute deviation with bounded error[J].Proceedings of the VLDB Endowment,2021,14(11):2114-2126.
[18] Mramba L K,Liu X,Lynch K F,et al.Detecting potential outliers in longitudinal data with time-dependent covariates[J].European Journal of Clinical Nutrition,2024,78(4):344-350.
附中文参考文献:
[1] 陈夏楠.面向SRAM型FPGA的抗单粒子翻转容错设计方法研究[D].绵阳:西南科技大学,2024.
[3] 王 番.基于ICAP的SRAM型FPGA抗单粒子翻转系统设计[D].北京:中国科学院大学,2022.
[5] 徐文龙,李凯旋,许 峥,等.基于RISC-V的抗单粒子加固研究[J].电子与封装,2025,doi:10.16257/j.cnki.1681-1070.2026.0023.

基金资助

空间可信计算与电子信息技术实验室开放基金项目(OBCandETL-2024-05)资助;辽宁省联合自然科学基金项目(2023BSBA-123)资助;中央高校基本科研业务费专项资金项目(N25ZJL005)资助;国家自然科学基金青年基金项目(62302083)资助.