Disaster Management and Information Transmission Decision-Making in Public Safety Systems

This paper introduces a Multi-Agency DisAster Management (MADAM) framework for Unmanned Aerial Vehicle (UAV)-assisted public safety systems, based on the principles of game theory and reinforcement learning. Initially, the information quality and criticality (IQC) provided by each agency to an UAV-assisted public safety network is introduced and quantified, and the concept of Value of Information (VoI) that measures each agency's positive contribution to the overall disaster management process is defined. Based on these, a holistic cost function is adopted by each agency, reflecting its relative abstention from the information provisioning process. Each agency aims at minimizing its personal cost function in order to better contribute to the disaster management. This optimization problem is formulated as a non- cooperative game among the agencies and it is proven to be an exact potential game, thus guaranteeing the existence of at least one Pure Nash Equilibrium (PNE). We propose a binary log- linear reinforcement learning algorithm that converges to the optimal PNE. The performance of the proposed approach is evaluated through modeling and simulation under several scenarios, and its superiority compared to other approaches is demonstrated.