Xilinx FPGA自主配置管理容错设计研究

Fault-Tolerance Design of Xilinx FPGA with Self-Hosting Configuration Management

针对SRAM型FPGA在空间辐射环境下容易受到单粒子效应影响的问题,在分析可重配置的XilinxFPGA的结构和故障模式的基础上,提出一种基于自主配置管理的Xilinx FPGA容错设计方案。综合运用诸如逻辑电路三模冗余、块存储器EDAC校验、动态回读、动态局部重配置及周期全局重配置等方法实现故障的屏蔽、检测和修复。该方案覆盖了FPGA的各种单粒子效应故障模式,并且在芯片内部实现了自主配置管理,具有体积小、成本低、可靠性高的特点。

For mitigating the single event effects when employing SRAM-based FPGA devices in space applications, a fault tolerance design with self-hosting configuration management is proposed on the basis of analysis for the architecture and failure modes of reconfigurahle Xilinx FPGA. The fault tolerance methods such as triple module redundancy of logic circuit, the error detection and correction of block RAM, configuration memory readback, dynamic partial reconfiguration and periodic global reconfiguration are adopted to mask, detect and recover the faults. This fault tolerance strategy covers all kinds of FPGA failure modes caused by the single event effects and accomplishes a self management of the configuration within the chip, which costs low, occupies small space, and obtains high reliability.