Sabot: Efficient and Strongly Anonymous Bootstrapping of Communication Channels


Loading...

Date

2025

Publication Type

Conference Paper

ETH Bibliography

yes

Citations

Altmetric

Data

Abstract

Anonymous communication is vital for enabling individuals to participate in social discourse without fear of marginalization or persecution. An important but often overlooked part of anonymous communication is the bootstrapping of new communication channels. If Alice wants to communicate with Bob, she must first learn his in-system identifier. In synchronous designs, message exchange is only possible once both communication partners have agreed to communicate. Thus, Alice must notify Bob of her intent, Bob must learn her in-system identifier, and Bob must acknowledge her notification. This bootstrapping process is generally assumed to occur out-of-band, but if it discloses metadata, communication partners are revealed even if the channel itself is fully anonymized. We propose Sabot, the first anonymous bootstrapping protocol that achieves both strong cryptographic privacy guarantees and bandwidth-efficient communication. In Sabot, clients cooperatively generate a private relationship matrix, which encodes who wants to contact whom. Clients communicate with k ≥ 2 servers to obtain “their” part of the matrix and augment the received information using Private Information Retrieval (PIR) to learn about their prospective communication partners. Compared to previous solutions, Sabot achieves stronger privacy guarantees and reduces the bandwidth overhead by an order of magnitude.

Publication status

accepted

Editor

Book title

Journal / series

Volume

Pages / Article No.

Publisher

Event

32nd ACM SIGSAC Conference on Computer and Communications Security (CCS 2025)

Edition / version

Methods

Software

Geographic location

Date collected

Date created

Subject

Anonymous communication; Bootstrapping; Privacy

Organisational unit

09653 - Paterson, Kenneth / Paterson, Kenneth

Notes

Funding

Related publications and datasets