Security Investment Over Networks with Bounded Rational Agents: Analysis and Distributed Algorithm

TL;DR

This paper models security investment decisions in networks considering human bounded rationality via prospect theory, analyzing its impact on resource allocation and proposing a distributed algorithm for strategy computation.

Contribution

It introduces a behavioral resource allocation framework incorporating prospect theory and develops a distributed algorithm for efficient security investment planning.

Findings

Human misperception significantly alters optimal resource allocation.

The distributed algorithm efficiently computes strategies in large networks.

Behavioral models improve the understanding of real-world security investments.

Abstract

This paper considers the security investment problem over a network in which the resource owners aim to allocate their constrained security resources to heterogeneous targets strategically. Investing in each target makes it less vulnerable, and thus lowering its probability of a successful attack. However, humans tend to perceive such probabilities inaccurately yielding bounded rational behaviors; a phenomenon frequently observed in their decision-making when facing uncertainties. We capture this human nature through the lens of cumulative prospect theory and establish a behavioral resource allocation framework to account for the human's misperception in security investment. We analyze how this misperception behavior affects the resource allocation plan by comparing it with the accurate perception counterpart. The network can become highly complex with a large number of participating agents. To this end, we further develop a fully distributed algorithm to compute the behavioral security investment strategy efficiently. Finally, we corroborate our results and illustrate the impacts of human's bounded rationality on the resource allocation scheme using cases studies.