https://www.mdu.se/

In its broadest sense, collaboration can be defined as a united effort to accomplish activities that ultimately lead to the achievement of shared objectives. This requires the sharing of information, planning, risks, and rewards between collaborating entities to a certain extent. In the case of systems of systems (SoS) where diverse and autonomous constituent systems (CS) interact to achieve shared goals, collaboration presents multifaceted challenges resulting from its distinct characteristics, including interconnectedness among the CS, heterogeneity, scalability concerns, dynamic environments, emergent behavior, stakeholder alignment challenges, and intricate decision-making processes. In order to enhance the achievement of the SoS goals, we propose a policy-guided collaboration approach. In this regard, we establish a learning-based policy generation process with the goal of guiding the decision-making behavior of CS. The practicality of the proposed approach is illustrated through a focused analysis of a high-rise building fire incident response system. Based on simulation results, the proposed approach performs better than the conventional approach in terms of SoS specific task completion time, performance with changes in simulation inputs, and efficiency. We also conducted a sensitivity analysis of task completion time by varying independent decision variables such as the number of CS instances and the size of collaborative tasks.