Silent Data Corruptions (SDCs) have become an insurmountable issue that threatens system reliability. General strategies for protecting programs from SDCs, such as dual modular redundancy, incur intolerable overheads. Another strategy is Algorithm-Based Fault Tolerance, which is highly bound to the specific algorithm and hard to generalize. In this study, we find that different implementations of the same algorithm may lead to very different SDC probabilities. We conduct a characterization study to quantify the differences and investigate the root causes. The insights we derive could help guide developers in the software engineering domain to design programs that are naturally resilient.

This paper proposes SALUS, a data-driven real-time safety monitor that detects and mitigates safety violations of an autonomous vehicle (AV). The key insight is that traffic situations that lead to AV safety violations fall into patterns and can be identified by learning from the safety violations of the AV. Our approach uses machine learning techniques to model the traffic behaviors that result in safety violations, characterize their early symptoms for training a preemptive model, and then detect real-time safety violations before actual crashes happen. To train the model, we leverage a pipeline of fuzzing techniques to tailor AV-specific safety violation symptoms and generate training data via data augmentation. Our evaluation demonstrates that the technique reduces over 97.2% of safety violations in industry-level autonomous driving systems such as Baidu Apollo, with no more than 0.018 false positive rate.