Formal Modeling and Strategy Synthesis for Resource Optimization in System of Systems

Systems of Systems (SoS) face challenges related to coordinated management of the various tasks performed by constituent systems (CS), resource allocation, and SoS-level decision-making to achieve optimal performance related to costs and energy consumption. Addressing these challenges requires rigorous modeling and verification methods that accurately represent CS, capturing their interactions and synchronization. This becomes paramount as the complexity of the SoS grows with the increasing autonomy of individual CS. In this paper, we propose a methodology for efficient resource optimization in batterypowered autonomous CS within an SoS. We begin by modeling the CS and their respective controllers within a constellation of an SoS using Uppaal Stratego. Then, we synthesize a strategy for mission planning and efficient resource utilization to achieve the mission. We also apply our proposed approach to an industrial case study focused on mass removal in the construction site modeled as an SoS to validate our proposed approach. The simulation results show that the synthesized strategies significantly improved resource optimization and reduced mission completion times compared to the ones without the synthesized strategies. Our approach, based on a synergetic combination of formal model modeling and reinforcement learning, provides a viable approach to achieve efficiency in SoS contexts.