空天应用中芯片自愈型可重构硬件结构与容错机制研究

The high reliability and strong fault-tolerant capability are very important for those electronic systems in space aircraft. How to make these systems more reliable is an urgent thesis should be conquered currently. Based on the theory of Embryonics Hardware, a new type electronic hardware named the Chip Self-healing Reconfigurable System, which with the abilities of self-diagnosing and self-repairing, was proposed. If this new electronic hardware was used to construct the space computers, their reliability will be significantly improved. Nowadays, because of the low hardware efficiency, application systems designed with the Chip Self-healing Hardware has the disadvantage of the serious circuit redundency. So, few could be used in space aircraft by now. In this requisition, three methods will be carried out to reduce the consumption of the reconfigurable hardware, they are the architecture amelioration, the cell simplification and the layout optimization of cell array. Firstly, a new reconfigurable architecture with four logic levels is improved, the molecules in this new model can repair by themselves. In this work, the design strategy of Tissue Self-repairing is substituted by Molecule Self-repairing, therefore, the hardware ultilization of cells is heightened. Secondly, a new architecture is proposed to simplify the circuit structure of the Configuration Memory in cells, where only three cells' configuration data is stored. In this method, the transient failure and the stuck-at fault in the Configuration Memory can both be tolerated. Thirdly, follow the reliability model to the cell array, several criteria are putted forward. The number of spare cells can be cut down according to the model analyzation results. All research achievements could be the radical metdology theory in designing the high reliable and self-healing systems.

机载电子系统的高可靠、强容错设计是空天应用研究需要解决的关键技术。基于仿生胚胎硬件理论的芯片自愈型可重构硬件，能实现芯片级在线自主故障诊断和自主修复，用于空天计算机系统和电子设备设计，可显著提高可靠性，但目前设计因设计冗余严重而无法实用。本项目面向空天应用从硬件系统体系结构、细胞电路模块组成和细胞阵列布局三个方面研究芯片自愈型可重构硬件的简化设计方法：⑴采用具有分子自修复能力的新四层结构模型，改善传统结构因修复层级高导致硬件利用率低的问题；⑵采用只保存三个细胞配置信息的配置存储器设计，大幅减小了细胞电路面积，且可同时容错瞬时故障和永久故障；⑶通过可靠性分析给出细胞阵列的布局优化准则，可减少电路的冗余空闲细胞数。研究成果可作为高可靠、自修复系统设计的理论基础。

空天应用环境中，机载电子系统的高可靠、强容错设计是迫切需要解决的关键技术。基于仿生胚胎硬件理论的芯片自愈型可重构硬件，具有芯片级在线自主故障诊断和自主修复特点，可显著提高空天环境数字电子系统性能，但因自修复覆盖不全面和资源冗余严重等问题难以实用。本项目从硬件结构体系、细胞电路设计方法和应用设计可靠性提高方法三个方面开展研究：(1)提出新的三层结构模型，采用增强低层级修复能力的方法提高修复灵活性和可靠性，通过去除高层级自修复方法提高硬件利用率，并给出相应的分层自修复策略；(2)提出新的细胞内部电路模块自测试和自修复结构，实现了细胞内所有模块电路的自测试和自修复能力的全覆盖，给出了电路简化设计方法；(3)研究应用设计过程的可靠性提高方法，通过分析细胞阵列在不同自修复策略、不同细胞单元粒度大小和不同细胞阵列布局结构下的可靠性变化规律，给出设计者选择能够获得最大可靠性的最佳设计策略和设计参数的方法。