Balancing system availability and lifetime with dynamic hidden Markov models

Abstract

Electronic components in space applications are subject to high levels of ionizing and particle radiation. Their lifetime is reduced by the former (especially at high levels of utilization) and transient errors might be caused by the latter. Transient errors can be detected and corrected using memory scrubbing. However, this causes an overhead that reduces both the availability and the lifetime of the system. In this work, we present a mechanism based on dynamic hidden Markov models (D HMMs) that balances availability and lifetime of a multi-resource system by estimating the occurrence of permanent faults amid transient faults, and by dynamically migrating the computation on excess resources when failure occurs. The dynamic nature of the model makes it adaptable to different mission profiles and fault rates. Results show that our model is able to lead systems to their desired lifetime, while keeping availability within the 2% of its ideal value, and it outperforms static rule-based and traditional hidden Markov models (HMMs) approaches.

Self-optimization of resilient topologies for fallible multi-robots

“…Transient faults can present themselves and then and disappear over time. In reliability engineering, probability distributions are typically used to model the initial time and the arrival times of permanent and transient faults, respectively [54].…”

Probabilistic timing analysis of time‐randomised caches with fault detection mechanisms

“…the time interval between two consecutive fault detection operations. This probability can be computed from the formula in [23] using Kolmogorov's definition as…”

“…To more reliably classify permanent faults amid transient faults, a D-HMM-based detection mechanism was proposed by Panerati et al [23]. HMMs are defined by the following components:…”

“…the time interval between two consecutive fault detection operations. This probability can be computed from the formula in [23] using Kolmogorov's definition as where is the cumulative density function of the time‐randomised variable failure describing the time at which the failure occurs.…”

“…In the literature on reliability, several probability distributions are used to model failure times [23]. A recurring one is the exponential distribution; however, the exponential representation is somewhat imprecise as it lacks the ability to capture the increasing failure probability due to the accumulated wear of a component.…”