Modelling fault dependencies when execution time budgets are exceeded

Abstract

Given that real-time systems are specified to a degree of confidence, budget overruns should be expected to occur in a system at some point. When a budget overrun occurs, it is necessary to understand how long such a state persists, in order to determine if the fault tolerance of the system is adequate to handle the problem. However, given the rarity of budget overruns in testing, it cannot be assumed that sufficient data will be available to build an accurate model. Hence this paper presents a new application of Markov Chain based modelling techniques combined with forecasting techniques to determine an appropriate fault model, using Lossy Compression to fit the model to the available data. In addition, a new algorithm, DepET, for generating job execution times with dependencies is given for use in task simulators.

Forecast-based interference

“…In the ield of real-time systems, Zheng [31] applied linear regression techniques to relate the amount of resource accesses (obtained via the PMCs) to the inlation in the execution times caused by the accesses; however, in our experience, a linear relationship does not hold for most tasks and platforms. Griin et al [9] employed forecasting to determine information about the behaviour of tasks when their execution time budgets were exceeded, by constructing a model based on the observed behaviour of the tasks' execution times. While Griin et al 's work focused on the technique of extrapolation, this paper employs Deep Learning Neural Networks (DLNNs) [19], a machine learning approach capable of learning sophisticated patterns in data and making predictions based on these learned patterns.…”