Automated Generation of Initial Configurations for Testing Component Systems

In the context of component-based systems, this paper presents the automated generation of initial states, from which an adaptive system starts to receive sequences of events that aim to provoke reconfigurations. For generating these states, also called configurations, we present a combinatorial algorithm supporting various architectural elements and relationships among them, while satisfying consistency constraints expressed by invariants. Moreover, this algorithm deals with the system-dependant instantiations of the primitive and composite components, parameters and relations, in order to produce meaningful structured configurations. While testing adaptation policies for component-based systems, this algorithm allows us to improve the capability of fulfilling coverage criteria by using different initial configurations. To illustrate the approach, the paper reports on experiments on a simulation with platoons of autonomous vehicles.