Investigating real-time control-flow error detection in hardware: How fast can we detect errors and take action?. (November 2021)

• Record Type: Journal Article
• Authors: Hoppe, Augusto; Kastensmidt, Fernanda Lima; Becker, Jürgen
• Is Part Of: Microelectronics and reliability. Volume 126(2021)
• Publication Date: 2021-11
• Subjects: Control-flow monitor -- Real-time -- Tracing -- Control-flow errors (CFEs) -- Error detection; Electronic apparatus and appliances -- Reliability -- Periodicals; Miniature electronic equipment -- Periodicals; Appareils électroniques -- Fiabilité -- Périodiques; Équipement électronique miniaturisé -- Périodiques; Electronic apparatus and appliances -- Reliability; Miniature electronic equipment; Periodicals; 621.3815
• DOI: 10.1016/j.microrel.2021.114264
• Languages: English
• ISSNs: 0026-2714

Abstract

Abstract: Control-flow monitoring is a promising approach to detect early anomalies and program execution deviations in safety and security domains. Many techniques based on branch tracing have been proposed, but there is no model to identify such methods' timing bounds. Critical systems and detection mechanisms operate with real-time constraints, and such a model is sorely needed. We propose here a novel model to estimate latency bounds on trace-based monitoring techniques. We evaluate our trace-based method's detection capabilities and real-time model through two fault injection campaigns. We can detect 98% of all branch opcode faults and 100% of all explicit control-flow errors, providing clearly defined latency bounds.