PISTIS: An Event-Triggered Real-Time Byzantine-Resilient Protocol Suite

TL;DR

This paper introduces PISTIS, a suite of real-time Byzantine-resilient protocols designed for scalable, fault-tolerant distributed systems, demonstrating robustness and low latency in challenging network conditions.

Contribution

The paper presents a novel suite of probabilistic synchronous protocols, including PISTIS, that ensure real-time Byzantine fault tolerance with practical performance and scalability.

Findings

PISTIS withstands message loss up to 50% in 49-node systems.

Provides bounded delivery latencies of a few milliseconds.

Proven correctness of protocols from reliable broadcast to consensus.

Abstract

The accelerated digitalisation of society along with technological evolution have extended the geographical span of cyber-physical systems. Two main threats have made the reliable and real-time control of these systems challenging: (i) uncertainty in the communication infrastructure induced by scale, and heterogeneity of the environment and devices; and (ii) targeted attacks maliciously worsening the impact of the above-mentioned communication uncertainties, disrupting the correctness of real-time applications. This paper addresses those challenges by showing how to build distributed protocols that provide both real-time with practical performance, and scalability in the presence of network faults and attacks, in probabilistic synchronous environments. We provide a suite of real-time Byzantine protocols, which we prove correct, starting from a reliable broadcast protocol, called PISTIS, up to atomic broadcast and consensus. This suite simplifies the construction of powerful distributed and decentralized monitoring and control applications, including state-machine replication. Extensive empirical simulations showcase PISTIS's robustness, latency, and scalability. For example, PISTIS can withstand message loss (and delay) rates up to 50% in systems with 49 nodes and provides bounded delivery latencies in the order of a few milliseconds.