This paper, published in the Journal of Surveillance, Security and Safety, introduces a sophisticated new threat to blockchain networks: the Eclipse Fork After Withholding (EFAW) attack. By blending network-layer isolation with protocol-layer manipulation, the authors demonstrate how attackers can extract significantly higher rewards from mining pools than previously thought possible.
TL;DR
The EFAW attack combines the "Eclipse" network attack with the "Fork After Withholding" (FAW) mining strategy to create a more potent threat to Proof-of-Work blockchains. By isolating miners within a victim pool, an attacker can discard their work to inflate their own share of rewards, consistently outperforming traditional mining attacks.
What Problem Does This Paper Address?
In Proof-of-Work (PoW) blockchains like Bitcoin, miners often join "pools" to combine their computational power and receive steady payouts. However, this collective structure creates vulnerabilities.
Existing attacks like Block Withholding (BWH) (where an attacker joins a pool but never submits valid blocks) and FAW (where an attacker holds back a valid block to intentionally cause a "fork" when someone else finds one) are well-documented. However, these usually focus on the protocol layer. The authors identify a gap: how can an attacker use network-level control to amplify these protocol attacks? They specifically look at the "Eclipse" attack—where an attacker controls all of a node's peer connections—to see if it can make mining attacks more profitable and harder to defend against.
Methods
The authors propose the EFAW attack model, which operates in three main phases:
- Infiltration: The attacker sends "infiltrator" miners to join a victim mining pool.
- Eclipsing: Simultaneously, the attacker launches an eclipse attack against a subset of honest miners within that same victim pool.
- Strategic Manipulation: Because the attacker controls the network view of the eclipsed miners, they can intercept their work.
- The attacker discards the Partial Proof-of-Work (PPoW) submitted by eclipsed miners, which reduces the pool's total "shares" and increases the attacker's relative payout.
- The attacker retains Full Proof-of-Work (FPoW) (actual valid blocks) found by eclipsed miners and releases them strategically to cause forks, just like in a standard FAW attack.
The researchers developed mathematical models to calculate the "Relative Extra Reward" (RER) for attackers in both single-pool and two-pool scenarios. They then validated these models using Monte Carlo simulations with $10^9$ rounds to ensure statistical accuracy.
Datasets
Not discussed in the paper. As this is a theoretical and simulation-based study of blockchain mechanics, the authors did not use external datasets. Instead, they used normalized computational power (where the total network power equals 1) and synthetic simulations to model miner behavior and attack outcomes.
Key Insights and Ideas
- The Power of Discarding: The most novel part of EFAW is the attacker's ability to discard the eclipsed miners' partial work. In a normal pool, every share counts toward the payout. By "silencing" eclipsed miners, the attacker effectively steals their portion of the pool's earnings without the pool manager realizing the miners are even working.
- FAW as a Floor: The study proves that the FAW attack is the "lower bound" of EFAW. If the attacker fails to eclipse any miners (an eclipse ratio of 0), the attack simply degrades into a standard FAW attack.
- Direct Proportionality: The attacker’s profit is directly proportional to the number of miners they can successfully eclipse ($q$). The more "dark" they can keep the victim pool's members, the higher their ROI.
- Solving the "Miner's Dilemma": Traditional BWH attacks often result in a "prisoner's dilemma" where both the attacker and victim lose money. FAW solved this by making the attack profitable; EFAW takes it a step further by making it highly profitable.
Key Results
- Increased Profitability: In a single-pool attack where the attacker eclipses 50% of the victim's miners ($q=0.5$), the attacker can achieve a 32.54% Relative Extra Reward.
- Multi-Pool Scaling: When attacking two pools simultaneously, the rewards are even higher. Under optimal conditions ($q=0.5$ and a high probability of winning forks), the extra reward reached 43.16%.
- Simulation Accuracy: The Monte Carlo simulations almost perfectly matched the theoretical mathematical models (e.g., a predicted 2.13% reward matched a 2.13% simulated result), proving the robustness of their formulas.
- Victim Loss: The study confirmed that victim pools always suffer a loss of rewards, which increases as the attacker eclipses more of their members.
Limitations and Future Work
- The Cost of Eclipsing: The authors acknowledge that launching an eclipse attack requires significant network resources and IP addresses. While they argue these costs are "limited" compared to mining rewards, they did not include a formal cost-benefit analysis of the network overhead in their equations.
- Detection and Defense: The paper focuses on the attack to help maintainers understand the threat. It does not propose new defensive measures specifically tailored to EFAW, though it mentions existing eclipse attack detection methods.
- Hybrid Attacks: The authors suggest that future research could combine EFAW with other strategies like "Stubborn Mining" or "Self-holding" to see if the profit ceiling can be pushed even higher.
Summary updated