Collaborative Research: SOTERIA: Satisfaction and Risk-aware Dynamic Resource Orchestration in Public Safety Systems

Public Safety Systems (PSS) deal with victims, first responders, and emergency control centers (ECCs) that coordinate the rescue missions during and aftermath of a disaster. Currently, there is limited understanding of how humans make decisions to utilize personal and shared resources in the wake of disasters; moreover, the existing models are mostly qualitative. Limited information availability in disaster scenarios and resource uncertainty in PSSs pose two major challenges: (1) how to achieve the victims’ satisfaction accounting for their risk-aware (autonomous) decision-making characteristics about dynamic resource orchestration? and (2) how to design a resilient disaster response system incentivizing the victims’ participation in crowdsourcing? The innovative SOTERIA project addresses these challenges by proposing a novel behavioral decision-making model that captures humans’ decision-making characteristics under risk, stemming from uncertain availability of resources. It introduces a new field in game theory, called Satisfaction Games that aims to satisfy the victims’ minimum service requirements rather than maximizing their overall satisfaction, addressing the resource management problems in PSSs. A novel bio-inspired Disaster Response Network (DRN) is also proposed that mimics the inherent robustness of biological networks of living organisms, to support the victims’ participation in reliable crowdsourcing, while the truthfulness and quality of the collected information are evaluated based on the newly invented Bayesian Prospect Theory. The proposed methods and the system will be evaluated using real-world field data. The SOTERIA project will create new distributed control, optimization, and scalable resource management techniques in PSSs. The novelty lies in the integrated approach to efficiently managing resources considering human behavior and Tragedy of the Commons phenomena about shared resources, thus enhancing the victim’s satisfaction instead of utility maximization. To deal with incomplete information about resource availability, the proposed reinforcement learning techniques will enhance the victim’s decision-making capability in real time. Research outcomes of this project have tremendous potential for supporting ECC activities and saving peoples’ lives and infrastructures during and after disasters, by designing robust situation-aware disaster response networks and advancing state of the art research in Prospect Theory, Tragedy of the Commons, and Satisfaction Games. The project will train undergraduate and graduate students and the research findings will be disseminated via a project website and high-quality publications. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.