Transient and Permanent Adaptive Fault Classifier for FPGA Applications in the Space

Abstract

Field Programmable Gate Array devices (FPGAs) are used in many applications. FPGAs are subjected to faults especially in the space where they are subjected to radiation. Faults in FPGAs may be recoverable (transient) or non-recoverable (permanent). Recoverable faults can be resolved by reconfiguring the system, on the other hand non-recoverable faults, require the relocation of the logic to a non-faulty area. This paper proposes an adaptive fault classifier that can be used to differentiate fault types. The classification guides the system to use the suitable recovery strategy. Experimental results show the operation of the classifier and adaptation layers, the proposed classifier layer is smaller in area compared to previous work. The adaptation layer works satisfactory to cope with environment changes. Both of the layers work independently of the design on any other layer.