Predictability Versus Adaptability in Real-Time Computing Thomas

Abstract

The dichotomy of “adaptivity vs predictability” is as ill founded as the old “communism vs capitalism”. Just as capital is fundamental to any economy whether socialist or not, so predictability is essential to real-time systems whether adaptive or not. Two connotations of predictability that are relevant include: (1) the ability to anticipate the environment external to the information processing system, and (2) the ability to understand and determine the degrees of behavior of the internal real-time system state. Both of these issues are of importance to real-time systems since in these systems the internal system state is highly synchronized with the external environment. Stated another way, correctness in real-time systems is contingent not only on the ordering of internal events but on the ordering of both internal and external events. This synchronization in part defines real-time systems. Adaptivity is the ability of the system to adjust its internal state to be consistent with both the current internal system resources and external environmental condition. Adaptivity might be better opposed with the word “rigidity”. Consequently adaptivity and predictability are two aspects of a real-time system and are not mutually exclusive. A system can be both highly predictable and highly adaptable. However, in gerneral low predictability and high adaptivity is assumed.KeywordsInternal StateInformation Processing SystemExternal EventAdaptive SystemInternal SystemThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.