Skip to Main Content
Reconfigurable Field Programmable Gate Arrays (FPGAs) are growing the attention of developers of mission- and safety-critical applications (e.g., aerospace ones), as they allow unprecedented levels of performance, which are making these devices particularly attractive as ASICs replacement, and as they offer the unique feature of in-the-field reconfiguration. However, the sensitivity of reconfigurable FPGAs to ionizing radiation mandates the adoption of fault tolerant mitigation techniques that may impact heavily the FPGA resource usage. In this paper we consider time redundancy, that allows avoiding the high overhead that more traditional approaches like N-modular redundancy introduce, at an affordable cost in terms of application execution-time overhead. A single processor executes two instances of the same software sequentially; the two instances are segregated in their own memory space through a soft IP core that monitors the processor/memory interface for any violations. Moreover, the IP core checks for any processor functional interruption by means of a watchdog timer. Fault injection results are reported showing the characteristics of the proposed approach.