Milojević, Jovana (2024). Security and Fairness of Blockchain Consensus Protocols. (Thesis). Universität Bern, Bern

Preview

Text 24milojevic_j.pdf - Thesis Available under License Creative Commons: Attribution (CC-BY 4.0). Download (2MB) | Preview

Abstract

The increasing popularity of blockchain technology has created a need to study and understand consensus protocols, their properties, and security. As users seek alternatives to traditional intermediaries, such as banks, the challenge lies in establishing trust within a robust and secure system. This dissertation explores the landscape beyond cryptocurrencies, including consensus protocols and decentralized finance (DeFi). Cryptocurrencies, like Bitcoin and Ethereum, symbolize the global recognition of blockchain technology. At the core of every cryptocurrency lies a consensus protocol. Utilizing a proof-of-work consensus mechanism, Bitcoin ensures network security through energy-intensive mining. Ethereum, a representative of the proof-of-stake mechanism, enhances scalability and energy efficiency. Ripple, with its native XRP, utilizes a consensus algorithm based on voting for efficient cross-border transactions. The first part of the dissertation dives into Ripple's consensus protocol, analyzing its security. The Ripple network operates on a Byzantine fault-tolerant agreement protocol. Unlike traditional Byzantine protocols, Ripple lacks global knowledge of all participating nodes, relying on each node's trust for voting. This dissertation offers a detailed abstract description of the Ripple consensus protocol derived from the source code. Additionally, it highlights potential safety and liveness violations in the protocol during simple executions and relatively benign network assumptions. The second part of this thesis focuses on decentralized finance, a rapidly growing sector of the blockchain industry. DeFi applications aim to provide financial services without intermediaries, such as banks. However, the lack of regulation leaves space for different kinds of attacks. This dissertation focuses on the so-called front-running attacks. Front-running is a transaction-ordering attack where a malicious party exploits the knowledge of pending transactions to gain an advantage. To mitigate this problem, recent efforts introduced order fairness for transactions as a safety property for consensus, enhancing traditional agreement and liveness properties. Our work addresses limitations in existing formalizations and proposes a new differential order fairness property. The novel quick order-fair atomic broadcast (QOF) protocol ensures transaction delivery in a differentially fair order, proving more efficient than current protocols. It works optimally in asynchronous and eventually synchronous networks, tolerating up to one-third parties corruption, an improvement from previous solutions tolerating fewer faults. This work is further extended by presenting a modular implementation of the QOF protocol. Empirical evaluations compare QOF's performance to a fairness-lacking consensus protocol, revealing a marginal 5\% throughput decrease and approximately 50ms latency increase. The study contributes to understanding the practical aspects of QOF protocol, establishing connections with similar fairness-imposing protocols from the literature. The last part of this dissertation provides an overview of existing protocols designed to prevent transaction reordering within DeFi. These defense methods are systematically classified into four categories. The first category employs distributed cryptography to prevent side information leaks to malicious insiders, ensuring a causal order on the consensus-generated transaction sequence. The second category, receive-order fairness, analyzes how individual parties participating in the consensus protocol receive transactions, imposing corresponding constraints on the resulting order. The third category, known as randomized order, aims to neutralize the influence of consensus-running parties on transaction order. The fourth category, architectural separation, proposes separating the task of ordering transactions and assigning them to a distinct service.

Item Type:	Thesis
Dissertation Type:	Single
Date of Defense:	7 March 2024
Subjects:	000 Computer science, knowledge & systems 500 Science > 510 Mathematics
Institute / Center:	08 Faculty of Science > Institute of Computer Science (INF)
Depositing User:	Hammer Igor
Date Deposited:	18 Mar 2024 16:52
Last Modified:	18 Mar 2024 17:12
URI:	https://boristheses.unibe.ch/id/eprint/4966

Actions (login required)

View Item