Active Learning from Blackbox to Timed Connectors

Coordination models and languages play a key role in formally specifying the communication and interaction among different components in large-scale concurrent systems. In this paper, we use active learning to extract timed connector models from black-box system implementations. Firstly, parameterized Mealy machine (PMM) is introduced as an operational semantic model for channel-based coordination language Reo. With product and link operators defined, we can construct complex connectors by joining basic ones in form of PMM. Moreover, with a concretize mapping function, PMMs can be easily transformed into Mealy machines, and the latter can be extracted by an optimized L* algorithm.