Concurrent real-time software systems are vulnerable to performance saturation and reliability concerns due to environmental influences. Building intelligent concurrent systems that are able to adapt to environmental changes and reconfigure themselves is the key to avoiding performance degradation of concurrent real-time software systems and ensuring the liveness property of such systems. In this paper we present a machine learning-based approach that addresses the design of agent-based intelligent concurrent software systems in order to ensure the reliability and performance properties for such systems. Although reliability and performance are conflicting requirements in most cases, we will show how to use an aspect-oriented technology by which these requirements can be designed, implemented, reused, and replaced in isolation from each other. The performance and reliability of the software system can be reasoned about by intelligent agents who can direct the system to reconfigure itself in order to adapt to the environment changes. The agents rely on the data-mining techniques to discover patterns of performance degradation or imminent signals of reliability violation and to predict policies that cope best with the environmental fluctuations.
展开▼
