Non-intrusive fault tolerance in soft processors through circuit duplication

The flexibility introduced by Commercial-Off-The-Shelf (COTS) SRAM based FPGAs in on-board system designs make them an attractive option for military and aerospace applications. However, the advances towards the nanometer technology come together with a higher vulnerability of integrated circuits to radiation perturbations. In mission critical applications it is important to improve the reliability of applications by using fault-tolerance techniques. In this work, a non-intrusive fault tolerance technique has been developed. The proposed technique targets soft processors (e.g. LEON3), and its detection mechanism uses a Bus Monitor to compare output data of a main soft-processor with its redundant module. In case of a mismatch, an error signal is activated, triggering the proposed fault tolerance strategy. This approach shows to be more efficient than the state-of-the-art Triple Modular Redundancy (TMR) and Software Implemented Hardware Fault Tolerance (SIHFT) approaches in order to detect and to correct faults on the fly with low area overhead and with no major performance penalties. The chosen case study is an under development On-Board Computer (OBC) system, conceived to be employed in future missions of the Brazilian Institute of Space Research (INPE).