Tackling Sequential and Coordinated Attacks in Security Domains with Real-time Information
Abstract
Many real-world security problems exhibit the challenge of sequential and coordinated attacks on targets such as critical infrastructure. In contrast, security agencies have to effectively allocate limited security resources to the targets in response to these attacks, upon receiving real-time information regarding them. In fact, these sophisticated attacks could mislead security agencies to focus on resolving attacks that have been happened, leaving other important targets vulnerable to subsequent attacks. This project proposes to study the problem of sequential and coordinated attacks with real-time information using a game-theoretic approach, taking into account strategic and sequential interactions between security agencies and attackers. There are several research challenges in this problem domain that can be broadly summarized in three points: (i) how to accurately model strategic interactions between security agencies and attackers in a sequential manner; (ii) how to design effective defense plans for security agencies against sequential and coordinated attacks, taking into account real-time information regarding these attacks; and (iii) how to properly analyze and quantify strategic behavior, the loss and benefit of sequential attack and defense actions. This project proposes to address these challenges by bringing together techniques from game theory, behavioral economics, machine learning, and optimization to build a game-theoretic framework that incorporates real-time information, behavior of sequential and coordinated attacks, and strategic interactions of players. This research will contribute to methodological advancements in the field of Artificial Intelligence (AI), particularly in multi-agent systems and game theory. More specifically, this proposal will result in: (i) new game-theoretic models which can capture a variety of real-world security scenarios in which sequential and coordinated attacks are a major concern; (ii) new practical game-theoretic algorithms for computing effective strategic defense plans developed based on a variety of solution concepts for games including classic equilibrium concepts, cognitive hierarchy theory, and robust optimization; and (iii) important theoretical results about sequential and coordinated attack behavior and its impact on defense plans of security agencies in different adversarial domain settings. Finally, this project has the potential to provide new practical real-world applications which generate advanced computational AI solutions to tackle sequential and coordinated attacks for different real-world security problems. Some domain examples include public safety and security (e.g., terrorist attacks) and cybersecurity (e.g., botnet prevention).
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Oct 22, 2020
- Source ID
- W911NF2010344
Entities
People
- Thanh H. Nguyen
Organizations
- Army Contracting Command
- United States Army
- University of Oregon