Weaving the fabric of the control loop through aspects

Self-adaptive systems are systems that are able to autonomously adapt to changing circumstances without human intervention. A number of frameworks exist that can ease the design and development of such systems by providing a generic architecture that can be reused across multiple application domains. In this paper, we study the applicability of aspect-oriented programming (AOP) to see where and how AOP can provide an interesting alternative for implementing parts of the architecture of self-adaptive frameworks. In particular, we relate two existing self-adaptive frameworks to the body of work on aspect-oriented programming techniques for self-adaptation. We present an aspect-oriented architecture for self-adaptive systems to show how AOP can be used for framework customization, event brokering and event aggregation. The potential of AOP for efficiency is shown by evaluating the architecture in a case study on decentralized traffic monitoring. Finally, we explore the potential and challenges of AOP for building scalable and decentralized self-adaptive systems.