Coevolutionary Network Model for Efficient Collaboration in Systems-of-Systems

Collaborative systems-of-systems (CSoSs) are defined by the Systems Engineering Body of Knowledge as groups of constituent systems that voluntarily work with each other toward a common goal. As the complexity and sociotechnical interactions of real systems increases, so too does our need to understand how to design and manage collaboration. The metric of collaboration efficiency (CE) is introduced as a measure of collaboration in any defined CSoS structure. To test the effect of constituent system parameters on CE, an agent-based model is developed that combines network evolution mechanisms (to model a system-of-systems) with evolutionary game theory (to model collaboration). Simulation results suggest that a reactive collaboration strategy or a reinforcement algorithm-based strategy converges to the highest CE in a CSoS, with 29.6% and 34.1%, respectively. The proposed approach and results allow conclusions about the best collaboration choices constituent systems should make when participating in CSoSs. Applicable to systems in energy, logistics, social organizations, infrastructure, and more, this research can improve the design and operation of systems in any CSoS.