On-line Monitoring of Real Time Applications for Early Error Detection

Abstract

Error confinement technologies have proven their efficiency to improve software dependability. Such mechanisms usually require efficient error detectors to swiftly signal any misbehaviour. Real-time systems, due to their timing constraints, require a richer description of correct and/or erroneous states that includes timing aspects. This paper presents real-time error detectors that can be automatically generated from formal models of the expected behaviours of software applications. The considered specifications provide the means to define quantitative temporal constraints on the execution of the application. These detectors check at run-time that the current execution matches its specification. The paper contribution is twofold. Firstly, at the theoretical level, we provide a formal definition of the expected behaviour of such detectors, ensuring a predictable behaviour of the detector system. Secondly, at a practical level, we provide a description of the complete generation process, from the models to the code of the detector.

Minotor: Monitoring Timing and Behavioral Properties for Dependable Distributed Systems

“…Most previous works on monitoring of real-time properties are based on centralized monitoring of properties expressed using Linear Temporal Logic (ltl) formulas or Timed Automata [MR10], [BLS06], [RFR08]. Unfortunately, both logic formulas and timed automata are inherently centralized and cannot be easily adapted for distributed monitoring.…”

“…Alas, on-line monitoring of complex, distributed and real-time systems is a highly complex task that, to our knowledge, has not yet been fully tackled. On-line monitoring of applications, be them distributed or not, in a centralized way is an active research domain (see, e.g., [MR10], [JRR94], [BLS06], [RFR08], [ZSLL09]), but, as we will see in Section I, these solutions do not provide a distributed implementation of monitors that can handle both distributed and real-time specifications.…”

A trustable software with a dynamic loop control mechanism

“…However, there are some applications that can cause a long delay. Thus, Robert et al (2008) proposed a method using automata theorem for analysing time of program's event in run-time monitoring; the expected outcome is the detected errors of software. This prototype reveals itself to be efficient with respect to real-time operating system deployment and decreasing event overhead.…”

“…Similar to the proposed method of this paper, Robert et al (2008), Mok and Liu (1997), and Gergeleit et al (1996) have proposed the detection techniques using time constraint for software termination. These techniques focus on the events and states of the software within a time interval.…”

A Step-Wise Approach for Integrating QoS throughout Software Development

“…In contrast, the approach of Robert et al enables generating monitoring support from non-functional requirements, represented as exceptional transitions in timed-automata [24]. In the same spirit, Duclos et al [14] have proposed to specify QoS requirements as aspects in the architectural models for providing separation of concerns, monitoring these requirements at runtime.…”