A method is presented for automated improvement of embedded application reliability. The compilation process is guided using Genetic Algorithms and a Multi-Objective Optimization Approach (MOOGA). Even though modern compilers are not designed to generate reliable builds, they can be tuned to obtain compilations that improve their reliability, through simultaneous optimization of their fault coverage, execution time, and memory size. Experiments show that relevant reliability improvements can be obtained from efficient exploration of the compilation solutions space. Fault-injection simulation campaigns are performed to assess our proposal against different benchmarks and the results are assessed against a real ARM-based System on Chip under proton irradiation.

No abstract

No abstract

A high-level C++ hardening library is designed for the protection of critical software against the harmful effects of radiation environments that can damage systems. A mathematical and empirical model to predict system behavior in the presence of radiation induced faults is also presented. This model generates a quick evaluation and adjustment of several reliability vs. performance trade-offs, to optimize radiation hardening based on the proposed C++ hardening library. Several simulations and irradiation campaigns with protons and neutrons are used to build the model and to tune it. Finally, the effects of our hardening approach are compared with other hardened and non-hardened approaches.

This work presents a new Dual-Core LockStep approach to enhance fault tolerance in microprocessors. The proposed technique is based on the combination of software-based data checking and trace-based control-flow checking through an external hardware module. The hardware module is connected to the trace interface and is able to observe the execution of all the processors in the architecture. The proposed approach has been implemented for a dual core commercial processor. Experimental results demonstrate that the proposed technique has a high error detection capability with up to 99.63% error coverage.