Analysis and design of blockchain systems

Since the advent of Bitcoin in 2008, cryptocurrencies and the blockchain systems they are built upon have seen parabolic growth in almost every aspect of human culture. As blockchain technology continues to evolve, the need for rigorous analysis and thoughtful design of blockchain protocols has become increasingly critical. This work focuses on the fundamental principles of blockchain systems, with a particular emphasis on analyzing Proof-of-Work (PoW) blockchains and designing new algorithms and incentive mechanisms to enhance their performance and security. Central to the design of these protocols is the need to characterize and incentivize rational behavior, as cryptocurrencies depend on it to function. Understanding how participants, such as miners and users, behave strategically within a decentralized environment is crucial to ensuring the security, efficiency, and scalability of blockchain networks. In this dissertation, three major contributions are presented which explore the incentives, decision-making processes, and potential vulnerabilities inherent to these systems in order to provide insights that aim to improve both the design and operation of future blockchain protocols. The first contribution introduces BlockReduce, a novel Layer 1 PoW cryptocurrency designed to address the long-standing scalability problem. By analyzing the core architectural bottlenecks that limit throughput in existing PoW blockchains, BlockReduce proposes targeted protocol-level enhancements that significantly increase transaction throughput while maintaining the same degree of security as a traditional PoW system. The second contribution explores the behavior of rational miners under adversarial conditions, specifically in the context of double-spend attacks. It proposes a new reward function that incorporates the incentive for an attacker to attempt a double-spend, thereby internalizing the security risks posed by high-value transactions. The analysis frames mining as a dynamic mean-field game, identifying equilibrium strategies which guarantee that an attacker cannot profit from a double-spend attack. The third contribution presents a formal computational framework for understanding miner behavior in PoW systems, with a specific emphasis on the strategies miners follow in selecting which block to extend. By modeling blockchain growth as a Partially Observable Stochastic Game (POSG) and reducing it to a class of Partially Observable Markov Decision Processes (POMDPs) through application of the the mean field assumption, this work offers a rigorous method to analyze the stationary behavior of PoW systems under various system conditions and miner strategies. In so doing, this work provides the first proof of optimality of Bitcoin's Longest Chain Rule (LCR).

Name thesis:degree_name

Doctor of Philosophy

Discipline thesis:degree_discipline

Electrical and Computer Engineering

Grantor

The University of Texas at Austin

Year dc:date.issued

2025

Author dc:creator

• Georghiades, Yanni

Advisors dc:contributor.advisor

• Vishwanath, Sriram
• Garg, Vijay K. (Vijay Kumar), 1963-

Committee members dc:contributor.committeemember

• Lizy K. John
• Cesare Fracassi
• Tej Anand

• Blockchain
• Algorithmic game theory
• Incentive design
• Security
• Cryptocurrency
• Distributed systems
• Proof-of-work
• Mean field games
• Markov decision process

Identifier URI

https://doi.org/10.26153/tsw/61412

OAI identifier oai:identifier

oai:repositories.lib.utexas.edu:2152/134085