Worst-Case Deadline Failure Probability in Real-Time Applications Distributed over CAN (Controller Area Network) - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Journal Articles Journal of Systems Architecture Year : 2000

Worst-Case Deadline Failure Probability in Real-Time Applications Distributed over CAN (Controller Area Network)

Nicolas Navet
  • Function : Author
  • PersonId : 830599
Ye-Qiong Song
François Simonot
  • Function : Author

Abstract

Real-time applications distributed over the CAN network are generally characterised by stringent temporal and dependability constraints. Our goal is to take account of transmission errors in the design of such applications because the consequences of such disturbances are potentially disastrous. In this study, the concept of worst-case deadline failure probability (WCDFP) is introduced. The motivation of the probabilistic approach is that, in practice, the number of errors occurring during a given time period can with difficulty be bounded. To evaluate the WCDFP, we propose, on the one hand, a method of computing for each message the tolerable threshold of transmission errors under which timing constraints are guaranteed to be met. On the other hand, we also suggest an error model enabling us to consider both error frequency and error gravity. Our error model follows a generalized Poisson process and its stochastic parameters have been derived. We then propose a numerically efficient algorithm to compute the probabilities and apply the analysis to an industrial case-study of the automotive field.
No file

Dates and versions

inria-00099080 , version 1 (26-09-2006)

Identifiers

  • HAL Id : inria-00099080 , version 1

Cite

Nicolas Navet, Ye-Qiong Song, François Simonot. Worst-Case Deadline Failure Probability in Real-Time Applications Distributed over CAN (Controller Area Network). Journal of Systems Architecture, 2000, 46 (7), pp.607-617. ⟨inria-00099080⟩
128 View
0 Download

Share

Gmail Facebook Twitter LinkedIn More