New feedback method hides network topology while preserving consensus behavior

Researchers Yushan Li, Jiabao He, Julien M. Hendrickx, and Dimos V. Dimarogonas have developed a feedback-based framework to preserve the topology privacy of networked systems running consensus protocols. The core idea is to deliberately violate topology identifiability conditions, making it impossible for an adversary to uniquely or accurately recover the true network structure from observed data. This is achieved by injecting carefully designed feedback signals that mask the real edges without altering the system's intended consensus behavior. The paper derives conditions for both partial observation (characterizing topology unsolvability) and full observation (constructing a solution space that enforces inaccuracy).

The method is fully distributed, relying only on local communication at each node, and includes a novel privacy budget mechanism that limits the amount of perturbation added. The authors establish performance guarantees by quantifying the tradeoff between consensus deviation (how much the behavior strays from ideal) and topology privacy (how hard it is to guess the real edges). They also propose a low-complexity heuristic algorithm to optimally allocate privacy resources across existing edges. Comparative simulations demonstrate that their design outperforms baseline approaches in preserving privacy while keeping consensus dynamics intact.