University of Illinois at Urbana-Champaign
Some theoretical problems in blockchain security and efficiency
Abstract
dc:descriptionA salient factor behind the success of Bitcoin and other cryptocurrencies is the security of its underlying consensus engine: the longest-chain protocol. Recent literature has given us an excellent theoretical understanding of this protocol's behavior in the synchronous (bounded communication delay) model. One principle that emerges from this work is that the protocol's security degrades as the bound on the communication delays increases. It is therefore natural to investigate further the protocol's robustness to network delays and how one might enhance it further. This thesis studies the security of the longest-chain protocol in network models that relax the synchronous assumption. In particular, we answer the following two questions in the affirmative: 1. In a network with random, possibly unbounded delays, is the longest-chain protocol secure? This work shows that the protocol is robust to sporadic, large delays, as is wont to happen in real-world networks. Moreover, it sheds light on the dual role of delays on security: they have both a local effect and a global effect. 2. In a network with more adverse conditions, i.e., occasional periods with arbitrary delay, How can the longest-chain protocol be made secure? This work designs a finality gadget with provable security properties that ensures the security of the longest-chain protocol even under arbitrary delay. It also reveals some nuances about the CAP theorem, a fundamental impossibility result in blockchains and distributed systems. In addition, Bitcoin suffers from poor transaction throughput. This is a fundamental limitation of the longest-chain protocol. Layer-two solutions provide a means to offload most of the transactions from the blockchain, using the consensus mechanism only when disputes arise. Payment channel networks are a class of layer-two solutions that have already been deployed in practice. In this dissertation, we model these networks mathematically, explore the benefits of dynamic routing and flow control, and present a network protocol that jointly performs these actions to steer the network to an optimal operating point. The protocol is derived as a method of solving a network optimization problem. It makes use of channel prices to coordinate the actions of the nodes in a decentralized fashion.
Degree
thesis:*- Name thesis:degree_name
- Ph.D.
- Level thesis:degree_level
- Dissertation
- Discipline thesis:degree_discipline
- Electrical & Computer Engr
- Grantor
- University of Illinois at Urbana-Champaign
- Year dc:date
- 2022
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Sankagiri, Suryanarayana
- Contributors dc:contributor
-
- Hajek, Bruce
- Viswanath, Pramod
- Miller, Andrew
- Ren, Ling
- Tse, David
Subjects
dc:subject × 6Rights
dc:rights- Statement dc:rights
-
- Copyright 2022 Suryanarayana Sankagiri
- Language dc:language
- en, eng
Identifiers
dc:identifier.*- Handle dc:identifier
- https://hdl.handle.net/2142/116230