Context: Systems of Systems (SoS) constitute a type of complex software systems resulting from integrating heterogeneous constituent systems that are independently operable on their own but are networked together for a common goal. Each constituent system has its own purpose and could operate and collaborate voluntarily with other constituent systems to achieve a common goal that cannot be treated by any of them in isolation. Objective: A constituent system may be deployed or undeployed at run-time within an SoS. Emergent behaviors may be undesirable and affect the behaviors of each constituent system and lead to unexpected operations and a lack of permanent status in the SoS. Thus, we need to continuously extract and represent the actual behaviors within the SoS at run-time. Method: In this paper, we implement the first step our “Architecture Mining” approach. Thus, we monitor an SoS and develop Discovery algorithm to extract the actual behaviors. The actual behaviors are presented by a “Discovered Model” dynamically and automatically built from the execution traces. Results: To implement our approach, we applied it to a case study entitled Smart City, which is an SoS including six types of constituent systems. We extracted the actual behaviors executed at run time from the SoS execution traces, which have never been modeled in any constituent system nor expected by the designer.