Adaptable and Adaptive Systems: The Intelligent Control Paradigm for Software Architecture

This paper presents a model-based architectural approach to evolving, or growing, adaptive software systems. The architecture is based on the Viable System Model (VMS) developed by Stafford Beer. The VSM, as a meta-model, tries to capture the essential invariants required by successful, or viable, systems. These types of systems are called Complex Adaptive Systems in Complexity Theory and Intelligent Control Systems in control engineering. I call the VSM-based architecture the Viable Software Architecture. The architecture incorporates the structure and behaviour of viable systems in software. Additionally, the approach permits “piecemeal” growth of the system to include high-levels of behaviour. The goal is to provide a method whereby software can be adapted (designtime, by humans) toward becoming an adaptive system (at runtime, self-controlling software). The Viable Software Architecture is developed in a sequence of stages using UML. Then a component framework specification is given based on the architecture. The overall goal of the approach is discussed.