Design Patterns for Self Adaptive Systems Engineering

TL;DR

This paper introduces design patterns for engineering self-adaptive software systems, focusing on the interactions within and between multiple feedback control loops based on the MAPE-K architecture.

Contribution

It proposes a systematic set of design patterns for modeling multi-loop feedback control in self-adaptive systems, addressing a gap in existing approaches.

Findings

Patterns facilitate systematic design of control interactions

Case study demonstrates pattern applicability

Addresses multi-loop control in self-adaptive systems

Abstract

Self adaptation has been proposed to overcome the complexity of today's software systems which results from the uncertainty issue. Aspects of uncertainty include changing systems goals, changing resource availability and dynamic operating conditions. Feedback control loops have been recognized as vital elements for engineering self-adaptive systems. However, despite their importance, there is still a lack of systematic way of the design of the interactions between the different components comprising one particular feedback control loop as well as the interactions between components from different control loops . Most existing approaches are either domain specific or too abstract to be useful. In addition, the issue of multiple control loops is often neglected and consequently self adaptive systems are often designed around a single loop. In this paper we propose a set of design patterns for modeling and designing self adaptive software systems based on MAPE-K Control loop of IBM architecture blueprint which takes into account the multiple control loops issue. A case study is presented to illustrate the applicability of the proposed design patterns.