DisAgg: Distributed Aggregators for Efficient Secure Aggregation in Federated Learning

TL;DR

DisAgg introduces a distributed aggregation protocol for federated learning that reduces communication and computation costs while maintaining privacy, enabling efficient large-scale model training.

Contribution

It proposes a novel client-aggregator architecture that eliminates local masking and homomorphic encryption, improving efficiency over existing secure aggregation methods.

Findings

DisAgg achieves a 4.6x speedup over OPA in processing 100k-dimensional updates from 100k clients.

The protocol maintains privacy against a curious server and limited colluding clients.

DisAgg reduces endpoint computation and communication costs in federated learning.

Abstract

Federated learning enables collaborative model training across distributed clients, yet vanilla FL exposes client updates to the central server. Secure-aggregation schemes protect privacy against an honest-but-curious server, but existing approaches often suffer from many communication rounds, heavy public-key operations, or difficulty handling client dropouts. Recent methods like One-Shot Private Aggregation (OPA) cut rounds to a single server interaction per FL iteration, yet they impose substantial cryptographic and computational overhead on both server and clients. We propose a new protocol called DisAgg that leverages a small committee of clients called Aggregators to perform the aggregation itself: each client secret-shares its update vector to Aggregators, which locally compute partial sums and return only aggregated shares for server-side reconstruction. This design eliminates local masking and expensive homomorphic encryption, reducing endpoint computation while preserving privacy against a curious server and a limited fraction of colluding clients. By leveraging optimal trade-offs between communication and computation costs, DisAgg processes 100k-dimensional update vectors from 100k 5G clients with a 4.6x speedup compared to OPA, the previous best protocol.