Scalable Multi-domain Trust Infrastructures for Segmented Networks

TL;DR

This paper proposes a leaderless, scalable trust infrastructure for segmented networks using Edwards curve digital signatures, addressing issues of trust delegation and scalability in distributed environments.

Contribution

It introduces a leaderless trust domain model utilizing Edwards curve signatures to improve scalability and robustness in distributed trust infrastructures.

Findings

Distributed signing overhead remains constant regardless of trust domain size.

The method enhances trust delegation without relying on a central trusted dealer.

Scalability is improved by replacing elliptic curve signatures with Edwards curve signatures.

Abstract

Within a trust infrastructure, a private key is often used to digitally sign a transaction, which can be verified with an associated public key. Using PKI (Public Key Infrastructure), a trusted entity can produce a digital signature, verifying the authenticity of the public key. However, what happens when external entities are not trusted to verify the public key or in cases where there is no Internet connection within an isolated or autonomously acting collection of devices? For this, a trusted entity can be elected to generate a key pair and then split the private key amongst trusted devices. Each node can then sign part of the transaction using their split of the shared secret. The aggregated signature can then define agreement on a consensus within the infrastructure. Unfortunately, this process has two significant problems. The first is when no trusted node can act as a dealer of the shares. The second is the difficulty of scaling the digital signature scheme. This paper outlines a method of creating a leaderless approach to defining trust domains to overcome weaknesses in the scaling of the elliptic curve digital signature algorithm. Instead, it proposes the usage of the Edwards curve digital signature algorithm for the definition of multiple trust zones. The paper shows that the computational overhead of the distributed key generation phase increases with the number of nodes in the trust domain but that the distributed signing has a relatively constant computational overhead.