Traditional real-time computing concepts and techniques are focused
on static, synchronous, relatively small-scale, mostly centralized,
device-level subsystems. Many real-time systems, particularly distributed
ones, are relatively large-scale, above the device level, and at least
partially dynamic and asynchronous. We call such systems "mesosynchronous."
For example, mesosynchronous systems often are found in military surveillance
and force projection platforms, and in network-centric warfare (plus
civilian domains). Hence the lives of both friends and foes depend on
the timeliness properties of such systems being dependably acceptable
according to application- and situation-specific criteria. The real-time
research community has historically failed to perceive and appreciate
this—admittedly difficult and domain-knowledge intensive—problem, especially for end-to-end timeliness in distributed mesosynchronous
real-time systems.
» Download Paper [PDF, 202KB]
