Detecting Sybil Identities in Distributed Multi-Robot Exploration via Geometric Ranking

Sybil attacks pose a significant threat to cooperative multi-robot exploration by allowing an adversary to introduce forged identities that manipulate shared coverage information and disrupt distributed coordination. In this work, we investigate the impact of strategically placed Sybil identities on distributed exploration using adversarial placement strategies proposed in prior work. Building on this threat model, we present IsoRank, a geometry-driven Sybil detection framework that exploits spatial consistency constraints inherent to physically realizable robot teams. IsoRank integrates density-based clustering, global centroid-isolation analysis, and local neighbor-consistency verification to identify identities exhibiting anomalous geometric behavior. The proposed method operates using only broadcast position information and requires neither centralized infrastructure nor specialized hardware. We evaluate IsoRank under distributed multi-robot exploration across varying robot densities and environment configurations. Experimental results, including ablation and parameter-sensitivity analyses, show that the combined multi-stage design consistently improves detection performance over the individual geometric stages across varying robot densities and environment configurations. These results demonstrate that spatial consistency provides an effective lightweight signal for detecting Sybil identities in distributed multi-robot exploration.