Introduction

• PDEVS formalism:
  • mathematical description of complex discrete systems using components
  • method and book (Zeigler et al.) widely popular
  • but: not used in standard simulation environments
• Basic obstacle for practical PDEVS use:
  • uses Moore architecture (output depends on state, not on input)
  • Mealy only via transitory states
    • → changes the expected order of concurrent events
    • → makes description complicated
    • → behaviour of such components depends on context
  • serious drawback for a library of reusable blocks
• Example compswitch:
  • model
    • 
  • idea
    • negative/positive numbers leave through upper/lower output
  • behaviour in PDEVS
    • comparator output only after internal switch to transitory state
    • input of outputswitch already on its way
    • → some numbers use wrong output
• RPDEVS:
  • "revised" PDEVS by Preyser et al.
  • basic goal: make PDEVS fit for real-life libraries
  • allows for Mealy-like behaviour directly
    • one generic state transition function δ(s,e,x)
    • output function λ(s, e, x)
    • λ is always called before δ
  • simulator changes the way chains of concurrent events are handled
• Example singleserver:
  • model
    • 
    • push strategy = entities proceed as far as possible
  • queue has internal state isBlocked
    • changed by input bl
  • server has output working
    • connected to queue.bl
  • server gets ready and queue holds several entities
    • → complex cascade of concurrent events
  • RPDEVS: model aborts ("too many λ steps")