Software has been increasingly embedded into systems (e.g., autonomous cars, traffic control systems, power distribution systems) to increase the precision of their functionalities, deliver automation, and make them smarter. Those systems have been combined and formed Systems-of-Systems (SoS) to realize futuristic software applications, such as smart cities. However, the complexity exhibited by them has also increased, claiming for techniques to support their engineering and quality. In this direction, modeling and simulation (M&S) have been established as valuable resources to deal with such complexity, potentially supporting an accurate prediction of SoS software correctness. The main contribution of this paper is reporting the results and advances achieved in the last five years of research on the use of M and S techniques to support Software Engineering of SoS. We report our experience, present learned lessons, and point for important challenges that must still be addressed.