Emergent and Adaptive Systems of Systems

The paper provides a first attempt to formalize Systems of Systems based on game theory and Multi-Agent Systems. We propose to use control agents to enforce joint strategies, which solve problems arising from the unconstrained interaction of autonomous agents and where Nash equilibria are suboptimal. In essence, in the resulting systems new solutions emerge, evolving in Systems of Systems that can form hierarchies. In addition, we present a learning algorithm that allows the resulting Systems of Systems to adapt to varying conditions and uncertainty -in the strategies the constituents follow and/or in their payoffs. The paper presents mechanisms to formalize Systems of Systems and their two main characteristics, namely, emergence and adaptability.