SAFE: Secure Aggregation with Failover and Encryption

TL;DR

This paper introduces SAFE, a secure aggregation algorithm for federated learning that enhances scalability, reduces resource use, and simplifies implementation by organizing learners in a chain and encrypting traffic.

Contribution

SAFE is a novel secure aggregation method that improves scalability and resource efficiency for federated learning with fixed participant sets.

Findings

Outperforms existing solutions by 70x with 36 nodes

Scales better and is less resource demanding

Easy to implement on constrained platforms

Abstract

We propose and experimentally evaluate a novel secure aggregation algorithm targeted at cross-organizational federated learning applications with a fixed set of participating learners. Our solution organizes learners in a chain and encrypts all traffic to reduce the controller of the aggregation to a mere message broker. We show that our algorithm scales better and is less resource demanding than existing solutions, while being easy to implement on constrained platforms. With 36 nodes our method outperforms state-of-the-art secure aggregation by 70x, and 56x with and without failover, respectively.