Wiretapping a hidden network

We consider the problem of maximizing the probability of hitting a strategically chosen hidden virtual network by placing a wiretap on a single link of a communication network. This can be seen as a two-player win-lose (zero-sum) game that we call the wiretap game. The value of this game is the greatest probability that the wiretapper can secure for hitting the virtual network. The value is shown to equal the reciprocal of the strength of the underlying graph. We efficiently compute a unique partition of the edges of the graph, called the prime-partition, and find the set of pure strategies of the hider that are best responses against every maxmin strategy of the wiretapper. Using these special pure strategies of the hider, which we call omni-connected-spanning-subgraphs, we define a partial order on the elements of the prime-partition. From the partial order, we obtain a linear number of simple two-variable inequalities that define the maxmin-polytope, and a characterization of its extreme points. Our definition of the partial order allows us to find all equilibrium strategies of the wiretapper that minimize the number of pure best responses of the hider. Among these strategies, we efficiently compute the unique strategy that maximizes the least punishment that the hider incurs for playing a pure strategy that is not a best response. Finally, we show that this unique strategy is the nucleolus of the recently studied simple cooperative spanning connectivity game.

💡 Research Summary

The paper introduces a two‑player zero‑sum game, called the wiretap game, that models the problem of intercepting a hidden virtual network by placing a single wiretap on a physical link of a communication network. One player, the wiretapper, chooses one edge of the underlying graph; the other player, the hider, selects a spanning subgraph that represents the hidden virtual network. The wiretapper succeeds if his chosen edge belongs to the hider’s subgraph, otherwise he fails. The objective of the wiretapper is to maximise the probability of success, while the hider aims to minimise it.

Main theoretical contribution – The authors prove that the value of this game (the maximal success probability that the wiretapper can guarantee) is exactly the reciprocal of the strength of the underlying graph. Graph strength, originally defined through a sequence of minimum cuts, measures how “robust’’ a graph is; the result establishes a direct bridge between a classic graph‑theoretic invariant and a game‑theoretic performance metric.

Prime‑partition – To analyse optimal strategies, the paper defines a unique partition of the edge set, called the prime‑partition. This partition groups edges that share the same contribution to the graph’s strength. The authors give a polynomial‑time algorithm (based on repeated minimum‑cut computations) that constructs the prime‑partition for any graph.

Omni‑connected‑spanning‑subgraphs – Using the prime‑partition, the authors identify a special family of spanning subgraphs that are the hider’s pure best responses against any max‑min strategy of the wiretapper. These subgraphs, termed omni‑connected‑spanning‑subgraphs, contain edges from each block of the prime‑partition in a way that avoids all minimum cuts, thereby guaranteeing the smallest possible exposure to the wiretapper.

Partial order and max‑min polytope – The blocks of the prime‑partition are equipped with a partial order that reflects inclusion relationships derived from the strength contributions. From this order the authors derive a set of linear inequalities, each involving only two variables, that completely describe the max‑min polytope – the convex set of all optimal mixed strategies for the wiretapper. Remarkably, the number of inequalities is linear in the number of edges, a substantial improvement over naïve exponential descriptions.

Extreme points and equilibrium strategies – The extreme points of the max‑min polytope correspond to probability distributions that assign mass to whole blocks of the prime‑partition. These extreme points constitute the wiretapper’s equilibrium strategies that minimise the number of pure best‑response subgraphs available to the hider.

Maximising the minimal punishment – Among the equilibrium strategies, the authors focus on the one that maximises the minimal punishment the hider would incur for deviating to a non‑best‑response pure strategy. This “most punitive’’ equilibrium is shown to be unique and can be computed efficiently by solving a small linear program derived from the partial order.

Connection to the nucleolus – The final and perhaps most surprising result is that the uniquely identified equilibrium strategy coincides with the nucleolus of the simple cooperative spanning‑connectivity game recently studied in cooperative game theory. The nucleolus is the most equitable and stable payoff allocation, and its equivalence to the wiretapper’s most punitive max‑min strategy provides a deep link between non‑cooperative security games and cooperative cost‑sharing models.

Algorithmic aspects – All major components – constructing the prime‑partition, building the partial order, generating the linear description of the max‑min polytope, and computing the nucleolus – are shown to run in polynomial time. In particular, the prime‑partition can be obtained in (O(m \log n)) time using standard min‑cut algorithms, and the linear program for the nucleolus has size linear in the number of edges.

Implications and future work – By tying the wiretap problem to graph strength, the paper offers a new, theoretically grounded metric for assessing the vulnerability of communication networks to single‑link interceptions. The prime‑partition and the associated linear description provide a tractable framework for designing optimal monitoring placements. Moreover, the identified connection to the nucleolus suggests that similar techniques could be applied to other security games where cooperative and non‑cooperative perspectives intersect. Future extensions might consider multiple simultaneous wiretaps, dynamic networks, or stochastic edge weights, and explore whether analogous partitions and nucleolus‑type solutions exist in those richer settings.