Scalable hierarchical Byzantine-robust federated aggregation

Specify hierarchical federated aggregation protocols that remain robust against poisoning coalitions distributed across geographic aggregation partitions while maintaining computational scalability as the connected-vehicle fleet grows.

Background

The Tier 3 cloud agent uses Byzantine fault-tolerant aggregation under an honest-majority assumption. Fleet growth creates computational challenges because pairwise consistency evaluation becomes more expensive, while geographic partitioning can reduce the observation pool available to detect poisoning campaigns spanning multiple regions.

The paper proposes hierarchical aggregation as a partial mitigation: RSU edge agents could validate local update distributions before forwarding aggregated gradients to the cloud. Although hierarchical federated learning exists in related vehicular IDS work, the specific protocol needed for the proposed three-tier architecture is not specified and remains an open problem.

References

Four open problems bound what the framework can currently guarantee. Sensor-manipulation adversaries who target both V2X messages and the physical observations used to validate them can evade Tier 1 cross-modal reasoning - closing this gap requires hardware-rooted sensor attestation that current automotive standards do not yet specify. Certificate propagation latency creates a revocation window that behavioral flagging only partially closes, and resolving it requires changes at the SCMS standards level rather than the detection layer. The autonomous execution of Drop and Quarantine responses operates without regulatory authorization frameworks that do not yet exist, and near-term deployment may require constraining autonomous action to Escalation-only pending future standardization. Finally, Byzantine fault tolerance at fleet scale degrades when poisoning coalitions are distributed across geographic aggregation partitions, and hierarchical aggregation protocols that address this remain unspecified.

— Autonomous Cyber Defense in Connected Vehicles: A Multi-Agent Approach to V2X Security  (2608.19135 - Medam, 19 Aug 2026) in Section VI.D, “Scalability of Federated Aggregation Under Fleet Growth”; Section VII conclusion