Restoring Reliability in Fault Tolerant Reconfigurable Systems

Abstract

The new generations of SRAM-based FPGA devices, built on nanometer technology, are the preferred choice for the implementation of reconfigurable computing platforms. However, smaller technological scales increase their vulnerability to manufacturing imperfections and hence to the occurrence of electromigration. Moreover, the large internal RAM (for configuration purposes or as embedded memory blocks) makes them more prone to soft errors. The incorporation of self-reconfiguration capabilities in recent FPGAs, allied to the use of soft and hard microprocessor cores, facilitates the offset of these vulnerabilities by enabling the development of self-restoring fault tolerant reconfigurable systems. In the methodology presented in this paper, the embedded microprocessor is also responsible for the implementation of online self-test-and-repair strategies, based on modular redundancy and on self-reconfiguration. The detection of faults, caused by soft or hard errors, may be followed by repairing actions, depending on the fault type. This approach leads to smoother system degradation, extending its lifetime and improving its reliability.