Challenge Problem Session

1
Challenge Problem Session
WPDRTS 2004 Tuesday April 27, 2004
2
Challenge Problem Context

• Real-Time and embedded systems are being applied
  to an ever wider range of application areas
• Embedded control systems for industrial
  automation, vehicles, military systems, etc
• This expanding range of application areas brings
  with it a widening range of system constraints
• Real-time and application specific scheduling of
  CPU and other resources
• Power
• Memory Footprint
• Configurability of many kinds of system behavior

3
Challenge Problem

• These trends challenge system designers because
  it presents two difficult scenarios
• Support a large number of hardware and software
  systems that are each appropriate for a narrow
  range of applications
• Support a small number of hardware and software
  systems that are each appropriate for a wide
  range of applications
• What methods for hardware and software design can
  be used to address the needs of this wide range
  of application areas by contributing to solution
  of one of these problems

4
Discussion Points

• Commodity CPU designs will dominate because of
  economies of scale
• Only one or two systems on a chip will exist
• Customized hardware and software will become
  common because configuration from standard
  components will be easy and cheap
• Only a few system HW/SW architectures will
  dominate because they will be good for everything
• Power will remain a significant constraint on
  system realization for the foreseeable future

5
Discussion Points

• Tension between the wish for COTS and
  customizability will not be resolved until
  functionality focused system composition
  techniques are refined to include QoS dimensions
  (Kopetz)
• Power, scheduling, application specific
  computation
• Reconfigurable Hardware will provide the best
  economic path for realizing customized SoC
  platforms

6
Discussion Points

• The configuration effort for really complex
  systems will require significant self-organizing
  support to be practical
• Emergent properties and behavior are antithetical
  to security and certification
• How much will this depend on flavors of emergent
• Certification of systems will become increasingly
  important and require HW and SW designed to
  permit it as a first principle