基于元胞自动机的SRAM_FPGA单粒子故障传播机理及容错方法研究

SEU Mitigation is a research challenge in space application. With device feature size shrinking, single event effect (SEE) is becoming more complicated and SEU error rate is ever increasing. Since SEE error is propagating in the logic circuit, affecting the interrelated sub-ciucuits and even resuting the failure of circuit function. The accuracy and speed of traditional SEE modeling technique and the performance and cost of traditional SEE hardening technique show more and more insufficiency. This proposal is mainly focusing on the characteristic of SEE soft error propagation and SEU mitigation for SRAM-based FPGA. We plan to introduce Cellular Automata (CA) towards SEE soft error dynamic propagation progress. The main content to be researched is as below: (1)In order to state the science meaning of SEE soft error propagation (SEP) characteristic, we plan to study the effect of interrelated module and output of specific circuit induced by SEE soft error. (2) Aiming at the SEE SEP simulation model based on CA, dynamic propagation progress will be studied. Also the critical propagation path should be determimated. (3)Aiming at researching the content of SEE soft error in SRAM-based FPGA, the principle and algorithm of SEU fault injection should be studied. Through extensive experimental justification and system integration, a highly effective and robust SEU fault injection system is expected to be implemented. (4) Based on SEE error's charaeteristic, from the aspect of FPGA's place and routing strategy , error detection method will be studied, which aims to decrease the losses in resource and performanee. We plan to analyze the condition by which SEU-MBE happens and gives some according solutions to SEU- MBE. For the proposed approaches, it should show a remarkable inerease of SEE immunity for FPGA's routing resource. The output of this proposal is benefit to enlarging the comprehending of SEE soft error propagation and enriching the analytical approach of soft error in SRAM-based FPGA. It can provide powerful theories and methods support for the device in radiation environment. Also has positive significance to promote the popularization and application of the cellular automata.

工艺缩减导致单粒子效应引起的二次失效持续增加，给传统的单粒子效应建模方法及加固方法提出了新挑战。以SRAM_FPGA为研究对象，明确单粒子故障传播特性的科学内涵，确定其研究内涵、描述方法和研究内容；探索故障传播关键路径的确定方法，建立元胞自动机单粒子软故障传播模型；研究模型的表征参数及其测试方法，实现单粒子故障注入并验证模型有效性；研究关键路径三摸冗余算法和布局布线优化算法，提升布线资源抗单粒子翻转能力；探讨单粒子多位翻转与用户电路布局的空间几何关系，揭示单粒子多位翻转机理，实现抗单粒子多位翻转容错方法。 研究成果可深化对单粒子故障传播特性的认知，丰富SRAM_FPGA软错误分析理论和和抗单粒子效应加固设计方法，为无辐射加固工艺的非宇航级VLSI在空间应用提供理论依据和方法指导。同时，研究成果对促进元胞自动机理论的推广应用具有积极意义。

工艺缩减导致单粒子效应引起的二次失效持续增加，给传统的单粒子效应建模方法及加固方法提出了新挑战。.项目在研究SRAM型FPGA微结构及单粒子效应表征基础上，重点分析了重离子、质子和大气中子作用下的单粒子效应机理，研究了单粒子效应机理作用下软错误产生的物理过程及FPGA软错误敏感部位，基于TCAD模拟了大气中子单粒子效应。建立了不同结构SRAM单元仿真电路，进行单粒子效应理论分析，得出SRAM型FPGA产生单粒子翻转的主要影响因素。.项目根据软错误形成机理，面向FPGA不同微结构的元素对软错误行为模式进行建模，在此基础上，针对开短路软错误，提出基于有向图的互连资源开短路软错误传播行为分析方法，采用XDL语言及Pajek软件进行有向图网络特征分析。针对SRAM单元翻转，提出基于元胞自动机的单粒子软错误传播行为模型分析方法，有效拓展了早期设计阶段对单粒子翻转在SRAM型FPGA逻辑层电路的演化过程的分析能力，为容软错误的可靠性设计和评估提供合理的指导和帮助。.本项目在对三模冗余传统方法可靠性分析基础上，考虑到单粒子带来的共因失效问题，提出一种基于多资源的冗余设计结构及功能安全性评估方法，通过将相同的逻辑功能采用不同的FPGA资源布局，提高了错误检测率并有效避免共因失效问题，并采用PDS方法对冗余设计进行功能安全性分析。.针对可用性和性能权衡需求，利用FPGA可重构特性，对基于动态重构的自适应SRAM型FPGA容错结构进行初步探索，以兼顾系统执行任务的高效性和可用性的目的，为系统功能灵活性和可靠性之间的权衡提供了很好的解决方案。针对典型Markov过程中的假设条件不符合动态重构结构的可用性分析的问题，提出多阶段Markov模型对该结构的可用性和性能权衡进行分析。.研究成果可深化对单粒子故障传播特性的认知，丰富SRAM_FPGA软错误分析理论和和抗单粒子效应加固设计方法，为无辐射加固工艺的非宇航级VLSI在空间应用提供理论依据和方法指导。同时，研究成果对促进元胞自动机理论的推广应用具有积极意义。