ADSEng: a model-based methodology for autonomous digital service engineering

In digital service ecosystems (DSEs), business stakeholders provide the most important driving factors and managing them is a challenge. It requires systems and services to handle uncertainty. Uncertainty in DSEs can be attributed to several factors; for example, dynamic nature and the unknown deployment environment, and change and evolution of requirements. Therefore, there is a need for novel software engineering methods and tools to handle these uncertainties in DSEs. In this regard, valuable lessons can be learnt from the autonomic computing (AC) paradigm and systems that are characterized by self-* properties. This paper proposes a novel, systematic service engineering methodology called ADSEng for ecosystem-based engineering of autonomous digital services. In the current research, the means of handling uncertainty from requirements to architecture and running systems are investigated. To do this, two interrelated research problems are studied: reflexivity that is realized using AC techniques, and evolvability of the ecosystem, supported by automated transformations. Our main contributions are: (i) a modeling methodology from uncertainty specification to runtime models and (ii) quality-driven adaptation patterns embodied by digital services. The paper also presents key lessons learnt from the research experience thus far.