Strategic Information Gathering and Revelation in Networked Multi-agent Systems
Abstract
In a multi-agent system, information available to different agents is crucially relevant to the strategies that they can follow, and hence the performance they can achieve. The problem of designing and analyzing optimal strategies for the agents with specified information patterns is a classical problem. However, the problem of how agents can either gather information from other self-interested agents to expand their own information set, or reveal information strategically to impact information sets of other agents in either a friendly or a malicious manner is not as yet well- understood. This project will study the design of strategic information revelation and gathering in multi-agent systems by using tools from game theory, mechanism design, optimization, and control theory. Three specific problems that this project focuses on are: (i) how to collect control-relevant information of a specified quality with minimum cost, if the information has to be collected from sensors that are self-interested and strategic. (ii) what is the optimal amount of information an agent should collect if there is a cost incurred in collecting such information and other agents may learn this information by observing the actions of the agents, (iii) how can an agent to use actions that are sub-optimal in the short run but mislead the other agents about the private information held by the agent?
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Sep 04, 2019
- Source ID
- W911NF1910483
Entities
People
- Vijay Gupta
Organizations
- Army Contracting Command
- United States Army
- University of Notre Dame