基于仿生学的多层自适应容错重构阵列研究

Bio-inspired method to develop the multi-layer and self-adaptive reconfigurable array

目前传统的可重构阵列面临着故障监测可靠性低、故障修复机制不灵活、时间与资源消耗随阵列规模急剧增大等困境。为此,根据生物体内遗传物质的精确自复制、免疫监督、组织自修复等容错修复机制,提出一种多层分布式自适应可重构电子阵列。阵列逻辑上分为3层,其中可配置逻辑层根据不同的用户配置信息完成系统功能;免疫层对可配置逻辑层的细胞进行故障监控;重构层在收到故障信号后寻找替代细胞、重新布线、转移配置信息,完成修复。以4位并行乘法器为例,对可重构阵列的功能和容错能力进行了验证。实验证明此可重构阵列实现了在线分布式自测试、自重构,多重联合故障检测增加了检测的可靠性,分块并行修复机制有效控制了故障修复的时间,增加了重构灵活度。资源与时间消耗基本不随阵列规模增大而增大。

Traditional reconfigurable arrays are facing issues of low reliability for monitoring the faults,low flexibility for doing reconfiguration,huge time and resources consuming when the scale of the circuits expand. Here,a distributed multi-layer reconfigurable array is proposed according to some novel fault-tolerant and repairing mechanisms of the organism like precise duplication of the DNA,immune surveillance and self-repairing of the tissues. The array has three layers logically. The programmable logic layer does the specific logical or sequential work according to the customs＇ configuration bits. The immune layer monitors cells in configurable layer. The reconfigurable layer finds the replacement cell and routes the related signals after accepting a fault signal. The fault-tolerant capability and performance of the array are verified by a 4bits parallel multiplier circuit. It is also demonstrated that the array has the ability of distributed online self-testing and self-repairing. The reliability of the monitoring is improved,and the repairing time is controlled by the distributed and parallel repairing mechanism. Reconfiguration is more flexible,and resources and time consuming will not increase as the scale of the circuits expand.