基于功能重分化的LUT型胚胎电子阵列进化自修复
An Evolutional Self-repair Method Based on Re-differentiation for LUT-based Embryonics Array
-
摘要: 胚胎电子阵列的进化自修复是在胚胎电子阵列的行/列移除基础上提出的一种新型自修复方法。利用演化硬件理论进行进化自修复计算时,求解空间大,且需要对所有个体进行电路功能评估,计算量巨大.提出了一种基于功能重分化的进化自修复方法,以电子细胞组成的电路网表为基础,在具有故障电子细胞的胚胎电子阵列上通过物理重映射,搜索对当前阵列具有更大适应能力的目标电路形式,完成进化自修复.详细分析了故障电子细胞对阵列结构的影响,对物理重映射进行了数学建模,并建立了进化自修复流程.该方法无需进行电路功能评估,降低了运算量,为进化自修复的快速实现打下了基础.最后,通过一个实验电路的仿真,对所提方法进行了验证.Abstract: Evolutional self-repair is a new self-repair method for embryonics array besides row/column elimination and cell elimination. The evolution hardware theory can be used for the evolutional self-repair. But in evolve process, the solution space is large, and the function of every individual circuit should be evaluated, the computed amount is huge. A novel evolutional self-repair method based on function re-differentiation is proposed in this paper. Based on the circuit netlist composed of electronic cells, the target circuit is physical re-mapped on the embryonics array with faulty cells, and the circuit form with larger self-repair capacity is searched. The faulty cells' impact on embryonics array is analyzed in detail, and a mathematical model is built for the physical re-mapping, and the function re-differentiation based evolutional self-repair process is established. The proposed method does not require the evaluation of circuits' function, and the amount of computation is reduced, which make a foundation to achieve fast evolutional self-repair. Finally, the proposed method was validated through an experimental circuit simulation.