LIFT: Byzantine Resilient Hub-Sampling

TL;DR

This work investigates the Byzantine resilience of Elevator, a decentralized hub-sampling protocol, and demonstrates a critical vulnerability threshold near $p=0.02$ where coordinated Byzantine nodes can seize all hub positions. To address this, the authors introduce LIFT, a Byzantine-resilient extension that performs a two-phase defense: first letting Elevator converge, then deterministically redistributing hubs using a shared PRNG seeded by the current hub IDs, ensuring identical hub selections across all correct nodes. Empirical evaluation in a 1000-node setting shows LIFT can nearly eradicate Byzantine hubs at low participation (e.g., 5%) and markedly reduce their influence at higher participation, though effectiveness diminishes as $p$ grows. The results highlight the importance of secure randomness for hub formation in decentralized overlays and position LIFT as a practical, low-overhead defense for Byzantine-resilient distributed systems.

Abstract

Recently, a novel peer sampling protocol, Elevator, was introduced to construct network topologies tailored for emerging decentralized applications such as federated learning and blockchain. Elevator builds hub-based topologies in a fully decentralized manner, randomly selecting hubs among participating nodes. These hubs, acting as central nodes connected to the entire network, can be leveraged to accelerate message dissemination. Simulation results have shown that Elevator converges rapidly (within 3--4 cycles) and exhibits robustness against crash failures and churn. However, its resilience to Byzantine adversaries has not been investigated. In this work, we provide the first evaluation of Elevator under Byzantine adversaries and show that even a small fraction (2%) of Byzantine nodes is sufficient to subvert the network. As a result, we introduce LIFT, a new protocol that extends Elevator by employing a cryptographically secure pseudo-random number generator (PRNG) for hub selection, thereby mitigating Byzantine manipulation. In contrast, LIFT withstands adversarial infiltration and remains robust with up to 10% Byzantine nodes. These results highlight the necessity of secure randomness in decentralized hub formation and position LIFT as a more reliable building block for Byzantine-resilient decentralized systems.

Figures (9)

• Figure 1: Running of Elevator without attack. The convergence to the hub happens in less than 4 cycles.
• Figure 2: Impact of single Byzantine node with different attack strategies on hub formation in 1000-node networks.
• Figure 3: Independent Byzantine attack at 5% participation showing minimal hub infiltration due to lack of coordination among attackers.
• Figure 4: Byzantine hub infiltration at 1% participation rate showing byzantine infiltration but with honest hubs still being the majority.
• Figure 5: Byzantine hub infiltration at 2% participation rate demonstrating the critical vulnerability threshold where Byzantine nodes begin systematic hub capture.