Information-Theoretically Secure Aggregation for Lightweight Federated Learning: Resilient to Dropouts and Adversaries

2026-07-22

AI Summary

This work addresses the vulnerability of gradient-sign-based federated learning on resource-constrained devices to inference attacks, as well as the high overhead and poor compatibility of existing secure aggregation protocols. The authors propose a lightweight, information-theoretically secure single-round aggregation framework that, under an honest-majority assumption, reveals only the final aggregated sign to the server. Key innovations include a majority-voting polynomial, an inverse-form exponent reduction technique to lower polynomial degree, and a combination of single-round secure multiplication with MDS code decoding, achieving linear offline computational and storage complexity. Experiments demonstrate a 99.5% reduction in online communication and an 85.7% decrease in latency compared to conventional approaches, along with accuracy improvements of 20.65% under client dropouts and 10.74% in the presence of malicious behavior.

Abstract

On-device federated learning (FL) enables privacy-preserving and personalized model training on resource-constrained devices such as smartphones and IoT nodes. To reduce communication cost, sign-based methods (e.g., signSGD) transmit one-bit gradients. However, exposing gradient signs makes them vulnerable to inference attacks, while existing secure aggregation schemes are often incompatible with such methods or incur significant computational and communication overhead. We propose a lightweight and information-theoretically secure aggregation framework tailored for sign-based FL. The framework securely computes the majority vote (MV) polynomial through single-round secure multiplication, ensuring end-to-end information-theoretic security under the honest-majority assumption while revealing only the final aggregated sign to the server. To enhance efficiency and scalability, we introduce two key techniques. First, inverse-form exponent reduction halves the effective MV polynomial degree, reducing both communication and computation costs. Second, we propose single-round secure multiplication, achieving linear offline complexity and storage with only a single online communication. Together, these techniques reduce online communication by up to 99.5% and latency by up to 85.7% compared to conventional approaches. Also, by leveraging inherent MDS-code-based decoding, the framework achieves robustness against both dropouts and adversarial behaviors, yielding accuracy gains of up to 20.65% and 10.74%, respectively. Overall, the proposed framework establishes a practical foundation for large-scale, low-latency, and information-theoretically secure aggregation in sign-based FL.

Problem

• federated learning
• secure aggregation
• sign-based methods
• information-theoretic security
• dropouts and adversaries

Innovation

• information-theoretic security
• sign-based federated learning
• secure aggregation
• majority vote polynomial
• MDS-code robustness