Resisting Selfish Mining Attacks in the Bicomp

Rui Tian, Wei Gong

Abstract

Selfish mining, which is an attack on the integrity of the Bitcoin network, was first proposed by Cornell researchers Emin Gun Sirer and Ittay Eyal in 2013. Selfish mining attack also exists in most Nakamoto consensus protocols. Generally speaking, selfish mining strategy can comprise a Nakamoto consensus system with less than 25% mining power of the whole system. We have discussed how the Bicomp can resist selfish mining in our former paper "Bicomp: A Bilayer Scalable Nakamoto Consensus Protocol". In this technical report, we give a detailed derivation on the conditions a selfish attacker should meet to earn more revenues through selfish mining. And we also get a conclusion that through adjusting macroblock difficulties together with tenure lengths, the Bicomp protocol has high resistant towards selfish mining.

Introduction

In the world of blockchain, "selfish mining" is a classic strategy where a miner (or pool) keeps a newly discovered block secret rather than broadcasting it immediately. By mining on top of this private block, they hope to create a longer chain than the public one. When they eventually reveal their private chain, the honest miners' work is "orphaned" (discarded), allowing the selfish miner to claim a disproportionate share of rewards.

While standard Nakamoto consensus (like Bitcoin) is theoretically vulnerable to attackers with as little as 25% of the network's hashing power, this report analyzes how the bilayer protocol, Bicomp, behaves. Bicomp separates the process into microblocks (for transactions) and macroblocks (which package microblocks into a trunk chain). The challenge is determining if Bicomp’s unique "tenure" and "transaction diversity" rules make it harder for selfish miners to succeed.

Modeling the Attack

The authors use a State Machine approach to model the competition between a selfish miner (Pool A) and the honest network (Pool B). States are defined based on the "lead" the private chain has over the public chain (e.g., State 1 means the selfish miner is one block ahead).

State machine of the selfish mining attack in Bicomp.

Transition probabilities ($p_0$ through $p_3$) describe the likelihood of A or B finding the next macroblock header under different conditions. These probabilities are essential for calculating the expected rewards for both parties across the various states.

Definitions of transition probabilities used in the state machine.

A crucial assumption in Bicomp is that because selfish miners only see their own transactions, they struggle to produce blocks with the same "transaction diversity" as the honest majority. In Bicomp, if two chains are the same length, the one with more diverse transactions wins, creating a natural disadvantage for the attacker.

Key Insights and Ideas

  • The Diversity Penalty: Selfish miners in Bicomp are at a natural disadvantage. Because they are a minority, they cannot generate as many diverse transactions as the rest of the network. This forces them to maintain a larger height advantage to successfully replace the public chain, raising the bar for a successful attack.
  • Configurable Resistance: The paper shows that the system's resistance isn't hard-coded. By adjusting two parameters—macroblock PoW difficulty and tenure length—network operators can effectively "dial in" the level of security against selfish mining.
  • The "Multi-Header" Fix: The authors propose a structural fix: allowing multiple leaders to generate headers in each round. If honest miners can mine on alternative headers while the attacker is hiding theirs, the attacker loses their head-start, effectively neutralizing the selfish strategy.

Key Results

The paper culminates in a mathematical condition for profitable selfish mining in Bicomp: $$\frac{2 - 2q_0 - q_1 + 2q_2}{2 - 2q_0 + 2q_2} < \alpha$$ Where $q_n$ represents the probability of being in a specific state of "lead" and $\alpha$ is the attacker's share of hashing power.

The authors conclude that through the combination of tenure mechanisms and the proposed "multi-header" modification, Bicomp can be made highly resistant to selfish mining. Unlike standard Bitcoin, where the 25% threshold is relatively static, Bicomp’s architecture allows for a much higher security margin.

Limitations and Future Work

While the authors provide the inequality needed to calculate the exact mining power threshold, they note that solving it requires Monte Carlo simulations, which are not included in this specific report. Additionally, the model assumes constant network spreading times and fixed rewards. In a real-world network, variable latency and fluctuating transaction fees could complicate the "diversity" advantage. Future work will involve detailed formal proofs for the "multi-header" fix to ensure it does not introduce other security vulnerabilities.

Summary updated

Abstract

Selfish mining, which is an attack on the integrity of the Bitcoin network, was first proposed by Cornell researchers Emin Gun Sirer and Ittay Eyal in 2013. Selfish mining attack also exists in most Nakamoto consensus protocols. Generally speaking, selfish mining strategy can comprise a Nakamoto consensus system with less than 25% mining power of the whole system. We have discussed how the Bicomp can resist selfish mining in our former paper "Bicomp: A Bilayer Scalable Nakamoto Consensus Protocol". In this technical report, we give a detailed derivation on the conditions a selfish attacker should meet to earn more revenues through selfish mining. And we also get a conclusion that through adjusting macroblock difficulties together with tenure lengths, the Bicomp protocol has high resistant towards selfish mining.
BibTeX
@misc{tian2018resisting,
  author = {Rui Tian and Wei Gong},
  title = {{Resisting Selfish Mining Attacks in the Bicomp}},
  year = {2018},
  eprint = {1809.06289},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1809.06289},
}
Figure