ML Security PapersOne-Shot Secure Aggregation: A Hybrid Cryptographic Protocol for Private Federated Learning in IoT
Imraul Emmaka , Tran Viet Xuan Phuong
Published on arXiv
2511.23252
Model Inversion Attack
OWASP ML Top 10 — ML03
Key Finding
Achieves sub-second execution on Raspberry Pi 4 with constant ~12× communication expansion and collusion resistance against the server plus up to N-2 clients, outperforming multi-round MK-CKKS baselines in communication efficiency.
Hyb-Agg
Novel technique introduced
Federated Learning (FL) offers a promising approach to collaboratively train machine learning models without centralizing raw data, yet its scalability is often throttled by excessive communication overhead. This challenge is magnified in Internet of Things (IoT) environments, where devices face stringent bandwidth, latency, and energy constraints. Conventional secure aggregation protocols, while essential for protecting model updates, frequently require multiple interaction rounds, large payload sizes, and per-client costs rendering them impractical for many edge deployments. In this work, we present Hyb-Agg, a lightweight and communication-efficient secure aggregation protocol that integrates Multi-Key CKKS (MK-CKKS) homomorphic encryption with Elliptic Curve Diffie-Hellman (ECDH)-based additive masking. Hyb-Agg reduces the secure aggregation process to a single, non-interactive client-to-server transmission per round, ensuring that per-client communication remains constant regardless of the number of participants. This design eliminates partial decryption exchanges, preserves strong privacy under the RLWE, CDH, and random oracle assumptions, and maintains robustness against collusion by the server and up to $N-2$ clients. We implement and evaluate Hyb-Agg on both high-performance and resource-constrained devices, including a Raspberry Pi 4, demonstrating that it delivers sub-second execution times while achieving a constant communication expansion factor of approximately 12x over plaintext size. By directly addressing the communication bottleneck, Hyb-Agg enables scalable, privacy-preserving federated learning that is practical for real-world IoT deployments.
Key Contributions
- Hyb-Agg: a hybrid secure aggregation protocol combining MK-CKKS homomorphic encryption with ECDH-based additive masking to achieve one-shot (single non-interactive upload per round) FL aggregation with constant per-client communication overhead
- Formal security proof under RLWE, CDH, and random oracle assumptions, guaranteeing gradient privacy against a semi-honest server colluding with up to N-2 clients while eliminating partial decryption rounds
- Empirical evaluation on resource-constrained hardware (Raspberry Pi 4) demonstrating sub-second execution and a constant ~12× communication expansion factor independent of participant count
🛡️ Threat Analysis
Hyb-Agg is a secure aggregation protocol that protects individual client model updates (gradients) from a semi-honest aggregation server and up to N-2 colluding clients, directly defending against gradient leakage attacks in federated learning. The guidelines explicitly include secure aggregation protocols that defend against gradient leakage even when the primary contribution is systems/communication efficiency.