Securing and scaling cryptocurrencies

Aviv ZoharIJCAI 2017 · DOI

TL;DR

This paper explores the fundamental challenges of securing and scaling cryptocurrencies, focusing on Bitcoin's design. It delves into the trade-offs between security and performance, highlighting how increasing transaction throughput can compromise the system's integrity and discussing novel approaches like DAG-based protocols and off-chain transaction channels to address these issues.

What Problem Does This Paper Address?

Cryptocurrencies, particularly Bitcoin, aim to provide a decentralized digital currency system. However, achieving both security and scalability presents a significant challenge. Bitcoin's core mechanism relies on a consensus protocol that synchronizes a ledger of transactions. This system is designed to be secure against attackers who might try to double-spend by creating conflicting transaction histories. The security of this system hinges on two key assumptions: that attackers are not overwhelmingly powerful (controlling less than 50% of computational resources) and that messages propagate quickly relative to block creation times.

As cryptocurrencies gain adoption, the demand for processing more transactions per second increases. This leads to a scalability problem, as Bitcoin's current design has inherent limits on transaction throughput (around 3-4 transactions per second). To increase this, either block sizes or block creation rates must increase. However, larger blocks or faster creation times exacerbate the message propagation issue, potentially weakening the security assumptions. This paper investigates how to overcome these limitations and maintain security while improving scalability.

Methods

The paper discusses several approaches to address the security and scalability challenges:

  • Nakamoto Consensus (Proof-of-Work): This is Bitcoin's foundational consensus mechanism. It uses computational puzzles (Proof-of-Work) to secure the network. Miners compete to solve these puzzles to add new blocks to the blockchain. The longest chain, with the most accumulated Proof-of-Work, is considered the valid one. This mechanism is designed to make it prohibitively expensive for attackers to create a longer, fraudulent chain.
  • GHOST Protocol (Greedy Heaviest Observed Sub-Tree): This protocol modifies the Nakamoto consensus by considering blocks that form forks in the network. Instead of solely relying on the longest chain, GHOST considers the "weight" of blocks in a subtree, allowing parallel blocks to contribute to the chain's security.
  • Inclusive Blockchains: This approach extends GHOST by not only considering off-chain blocks but also their contents. It aims to establish a consistent ledger by creating a total ordering of all known blocks, even those not on the main chain. Inconsistent transactions (e.g., double-spends) are rejected.
  • SPECTRE Protocol: This protocol utilizes a Directed Acyclic Graph (DAG) structure for blocks, moving away from a strict chain. It employs a pairwise voting mechanism between blocks, inferred from the causal structure, to achieve a consistent ledger. SPECTRE aims to work without explicit bounds on message delays, making it more scalable.
  • Off-chain Transaction Channels: These are mechanisms that allow participants to conduct multiple transactions off the main blockchain, only settling the final state on-chain. This significantly reduces the load on the main network. Examples include bi-directional channels and techniques leveraging trusted execution environments.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • The Inherent Trade-off: The paper clearly illustrates the fundamental tension between increasing transaction throughput and maintaining the security guarantees of a decentralized system. Faster block creation or larger blocks, while improving scalability, can weaken the system's resilience to attacks.
  • Beyond the Longest Chain: The exploration of GHOST and SPECTRE highlights that the simple "longest chain" rule might not be optimal for scalability. Alternative consensus mechanisms that consider more of the network's activity (like parallel blocks or DAG structures) can offer better performance.
  • The Power of DAGs: The SPECTRE protocol's use of a DAG structure is a significant departure from traditional blockchains. This structure allows for a more flexible and potentially more scalable way to represent transaction history and achieve consensus.
  • Incentives are Crucial: The paper emphasizes that the security of cryptocurrencies is deeply intertwined with their incentive mechanisms. Miners are rewarded for securing the network, but the design of these incentives can lead to unintended consequences, such as "selfish mining," where miners strategically delay block publication to gain an advantage.
  • Off-chain Solutions are Key for Scalability: The discussion on off-chain transaction channels (like the Lightning Network) points towards a practical path for achieving massive scalability by moving the bulk of transactions away from the main blockchain.

Key Results

The paper doesn't present specific quantitative results from experiments but rather discusses the theoretical properties and potential benefits of the proposed protocols:

  • GHOST and Inclusive Blockchains: These protocols aim to improve security in the face of faster block creation and offer better scalability by accounting for off-chain blocks and their contents.
  • SPECTRE Protocol: This protocol is designed to be highly scalable and to adjust to changing network conditions without explicit bounds on message delays. It offers weaker, but sufficient, properties for a consistent ledger.
  • Off-chain Transaction Channels: These channels allow for a significant increase in transaction throughput by reducing the burden on the main blockchain.

Limitations and Future Work

  • SPECTRE's Non-Transitive Relation: The paper acknowledges that the SPECTRE protocol's pairwise voting can lead to non-transitive relations (Condorcet paradox), meaning some conflicting transactions might be rejected. However, it argues this is acceptable as such conflicts often arise from double-spending attempts.
  • Self-Stabilization in Reductions: The paper notes that some reductions from PoW-based consensus to classic consensus mechanisms might not be self-stabilizing, meaning they might not recover from catastrophic events like network splits.
  • Future Work: The paper implicitly suggests continued research into optimizing incentive mechanisms, further developing DAG-based protocols, and exploring the integration of off-chain solutions with on-chain security. The ongoing evolution of cryptocurrencies and the exploration of new consensus mechanisms and scaling solutions are presented as active areas of research.

Summary updated

BibTeX
@inproceedings{zohar2017securing,
  author = {Aviv Zohar},
  title = {{Securing and scaling cryptocurrencies}},
  booktitle = {International Joint Conference on Artificial Intelligence},
  year = {2017},
  doi = {10.24963/ijcai.2017/742},
}
Figure