How to Choose Communication Links in an Adversarial Environment?

Game-theoretic Robustness of Many-to-one Networks

“…Our work is inspired by [4] and [5], which use game-theoretic analysis of robustness of many-to-many networks. In [4], the strategic interactions between a network manager, whose goal is to keep the network connected by choosing a spanning tree, and an attacker, whose goal is to disconnect the network by attacking a link, were modeled as a zero-sum game.…”

“…Furthermore, an efficient algorithm was provided to compute an optimal attack. In [5], the game model was generalized to include link attack costs, which can vary based on the targeted links, resulting in a non-zero-sum game.…”

“…While the definition of our game resembles that of [4] and [5], it is actually fundamentally different:…”

“…-First, our game models many-to-one networks, while [4] and [5] modeled manyto-many networks. We believe that studying adversarial games in many-to-one networks is more important as these networks are usually more vulnerable to attacks.…”

“…Since we believe that a general theory of adversarial network games and graph robustness metrics is possible, we have kept our notions as similar to that of [4] and [5] as possible, even though our model and methodology is different.…”