On the Performance of Wireless PBFT-Based Blockchain Network With IEEE 802.11

Ziyi Zhou, Oluwakayode Onireti, Xinyi Lin, Lei Zhang, Muhammad Ali ImranIEEE Syst. J 2024 · DOI

Abstract

Practical Byzantine Fault Tolerance (PBFT) is a promising consensus mechanism for energy-constrained IoT devices, but its performance in wireless environments is poorly understood. This paper presents a mathematical framework to evaluate PBFT over IEEE 802.11 networks, revealing that wireless contention triggers costly leader re-elections (view changes) that can lead to exponential delays. The authors provide an optimization strategy to balance throughput and reliability in these decentralized ecosystems.

Introduction

As decentralized 6G and IoT ecosystems evolve, blockchain is increasingly proposed for secure data management. PBFT is often preferred over Bitcoin’s Proof-of-Work due to its energy efficiency. However, PBFT was designed for stable wired connections. In wireless settings, nodes must compete for radio spectrum. Message collisions can be misinterpreted as node failures, triggering "view changes" that severely degrade performance.

The PBFT Consensus Mechanism

PBFT relies on a series of voting rounds to reach consensus. As shown in Figure 1, the process involves multiple phases—Request, Pre-Prepare, Prepare, Commit, and Reply—requiring intense communication between the primary node and backups.

The multi-phase PBFT consensus process involves significant message exchange between nodes.

In a wireless environment, the high volume of messages increases the likelihood of collisions. If a message in a critical phase is lost, the system may fail to reach consensus within a timeout period, forcing a view change to elect a new leader.

Methodology

The researchers developed a discrete-time Markov chain to model the unsaturated IEEE 802.11 broadcast scheme. This model allows for the calculation of transmission success probabilities under contention. They derived closed-form equations for:

  1. Transmission Success Probability: The likelihood of a message reaching its destination without collision.
  2. Transaction Confirmation Delay: The total time from request to finality.
  3. Throughput: The rate of successful transaction processing.

A key contribution is the derivation of "view change delay," which quantifies the time lost when the network cycles through faulty or collided primary nodes. To optimize performance, the authors used the De Moivre-Laplace theorem to find the ideal contention window size ($W$) and packet arrival rate ($\lambda$).

Key Insights: Contention and Interference

The study highlights a "double whammy" effect: channel contention directly lowers success rates and indirectly triggers slow view changes. While normal confirmation delay grows linearly with network size, view change delay grows exponentially, making collisions in large networks extremely costly.

Furthermore, the presence of non-PBFT nodes—regular Wi-Fi users sharing the same spectrum—significantly impacts performance.

A wireless network model showing the primary node, PBFT backup nodes, and interfering non-PBFT nodes.

As illustrated in Figure 3, these external nodes compete for the same airwaves, suggesting that wireless blockchains may require dedicated spectrum or prioritized "lanes" to maintain reliability.

Results and Optimization

Numerical simulations validated the analytical models, showing a tight match. For a 28-node network, the authors identified an optimal transmission probability of 0.0059. They found that larger contention windows (e.g., 128 or 256) are generally superior for maintaining stability in wireless clusters.

The results also revealed a "zig-zag" effect in success probability. Because PBFT's fault tolerance ($f$) is a step function based on the number of nodes ($n$), adding nodes only improves resilience at specific thresholds, leading to non-linear performance gains.

Conclusion

This research provides a critical tool for planning wireless blockchain deployments. By quantifying the impact of view changes and external interference, it offers a path toward more robust decentralized networks in the 6G era. Future work will extend this framework to 5G/6G cellular networks and explore adaptive protocols that respond dynamically to channel conditions.

Summary updated

BibTeX
@article{zhou2024performance,
  author = {Ziyi Zhou and Oluwakayode Onireti and Xinyi Lin and Lei Zhang and Muhammad Ali Imran},
  title = {{On the Performance of Wireless PBFT-Based Blockchain Network With IEEE 802.11}},
  journal = {IEEE Syst. J.},
  year = {2024},
  doi = {10.1109/jsyst.2024.3443541},
}
Figure