• DocumentCode
    270149
  • Title

    The schedulability region of two-level mixed-criticality systems based on EDF-VD

  • Author

    Müller, Dirk ; Masrur, Alejandro

  • Author_Institution
    Dept. of Comput. Sci., Tech. Univ. Chemnitz, Chemnitz, Germany
  • fYear
    2014
  • fDate
    24-28 March 2014
  • Firstpage
    1
  • Lastpage
    6
  • Abstract
    The algorithm Earliest Deadline First with Virtual Deadlines (EDF-VD) was recently proposed to schedule mixed-criticality task sets consisting of high-criticality (HI) and low-criticality (LO) tasks. EDF-VD distinguishes between HI and LO mode. In HI mode, the HI tasks may require executing for longer than in LO mode. As a result, in LO mode, EDF-VD assigns virtual deadlines to HI tasks (i.e., it uniformly downscales deadlines of HI tasks) to account for an increase of workload in HI mode. Different schedulability conditions have been proposed in the literature; however, the schedulability region to fully characterize EDF-VD has not been investigated so far. In this paper, we review EDF-VD´s schedulability criteria and determine its schedulability region to better understand and design mixed-criticality systems. Based on this result, we show that EDF-VD has a schedulability region being around 85% larger than that of the Worst-Case Reservations (WCR) approach.
  • Keywords
    processor scheduling; EDF-VD schedulability criteria; HI mode workload; HI task deadline downscaling; LO task; earliest deadline first-with-virtual deadlines; high-criticality task; low-criticality task; mixed-criticality task set scheduling; schedulability conditions; schedulability region; two-level mixed-criticality systems; virtual deadline assignment; Estimation; Processor scheduling; Resource management; Safety; Schedules; Time factors; Upper bound; EDF-VD; mixed criticality; real-time scheduling; resource efficiency;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Design, Automation and Test in Europe Conference and Exhibition (DATE), 2014
  • Conference_Location
    Dresden
  • Type

    conf

  • DOI
    10.7873/DATE.2014.269
  • Filename
    6800470