Deception for Characterizing Adversarial Strategies in Complex Networked Systems

• Main Authors: Al Amin, Md Ali Reza, Foytik, Peter, Shetty, Sachin, Dorismond, Jessica, Gamarra, Marco
• Format: Conference Proceeding
• Language: English
• Subjects: adversarial strategy; complex networked; Complexity theory; Costs; Cyber deception; Force; Measurement; Mission critical systems; Monte Carlo methods; POMCP; Systematics
• ISSN: 2331-9860
• DOI: 10.1109/CCNC49033.2022.9700549

There is a need for systematic characterization of complex networked systems involving friendly forces and opponent forces to understand the adversary opportunities and capabilities to cause harm and develop counter-strategies that would minimize the adversarial impact. Specifically, we need the ability to quantify the incurred cost (cost of protecting friendly forces) and induced cost (opponent cost to cause damage to friendly systems). In this paper, we address the problem of characterizing adversarial strategies and develop a suite of metrics that quantify the opportunity and capability of the adversary. We also propose a deception strategy that would force the opponent to redirect away from mission-critical systems by optimally placing decoy nodes and thereby expending opponent resources with little gain. Our adversary strategy characterization involves an integrated graph theoretic and Bayesian network that tracks lateral propagation of adversaries in a complex networked sys-tem. Our deception strategy is devised by a Partially Observable Monte Carlo Planning (POMCP) Framework that provides a sequence of actions that force an attacker to be redirected to a decoy network with high probability.