Modeling Self-adaptation - A Possible Endeavour?

Abstract

Self-adaptive systems have the capability to autonomously modify their behavior at runtime in response to changes in their internal structure or execution environment. Therefore, often self-adaptation emerges as a means to solve problems related to performance or security, to increase efficiency, or to react to various hazards. Basically, self-adaptation may emerge to solve a whole spectrum of various problems or hazards occurring in the execution environment, which implies that behavior modeling for self-adaptation requires intrinsic knowledge of the system context.

A new approach to modeling self-adaptation compliant to system goals is presented in this paper. In this approach, KnowLang, a knowledge representation language for self-adaptive systems, is used to model self-adaptive behavior. Special KnowLang policies are at the core of this approach. Ideally, KnowLang policies are specified to handle specific situations by pursuing a specific goal. A policy exhibits a behavior via actions generated in the environment or in the system itself. Specific probabilistic beliefs and generic conditions determine what specific actions shall be executed. Context properties are intrinsically embedded in the self-adaptive behavior, which makes that behavior context-reactive. To demonstrate the novelty of this approach, the paper elaborates on a self-adaptive behavior of an autonomous vehicle modeled with KnowLang.