Modelling and analyzing adaptive self-assembly strategies with Maude

Building adaptive systems with predictable emergent behavior is a difficult task and it is becoming a critical need. The research community has accepted the challenge by introducing approaches of various nature: from software architectures to programming paradigms and analysis techniques. Our white-box conceptual approach to adaptive systems based on the notion of control data promotes a clear distinction between the application and the adaptation logic. In this paper we propose a concrete instance of our approach based on (i) a neat identification of control data; (ii) a hierarchical architecture that provides the basic structure to separate the adaptation and application logics; (iii) computational reflection as the main mechanism to realize the adaptation logic; (iv) probabilistic rule-based specifications and quantitative verification techniques to specify and analyze the adaptation logic. We show that our solution can be naturally realized in Maude, a Rewriting Logic based framework, and illustrate our approach by specifying, validating and analyzing a prominent example of adaptive systems: robot swarms equipped with self-assembly strategies.