Designing Efficient and Secure Algorithms for Payment Channel Networks

Abstract

Major blockchains such as Bitcoin and Ethereum suffer from scalability issues. For instance, Bitcoin can handle up to about 7 transactions per second, whereas custodian payment networks such as Visa can handle tens of thousands of transactions per second. One of the promising solutions to scale blockchains such as Bitcoin is Payment Channel Networks (PCNs). In this thesis, we propose new solutions to make the existing PCNs, such as the Bitcoin's Lightning network, more efficient and secure. Chapter 3 studies watchtowers, which are entities that watch the blockchain on behalf of their offline clients to protect the clients’ payment channels. In practice, there is no trust between clients and watchtowers, and it is challenging to incentivize watchtowers to well-behave (e.g., to refuse bribery). To tackle this problem, Chapter 3 proposes a novel reputation system, and uses the system to incentivize watchtowers to not only deliver their promised service but also reduce their service fees in competition with each other. Chapter 4 proposes Spear, a new multi-path payment method that can support redundant payments. We show that this redundant payments can significantly improve the success rate of payment transfers in the Lightning network. The challenge in supporting redundant payments is that the recipient may overdraw funds from the redundant paths. In Chapter 4, we explain how this can be done without requiring high computation or major changes to the Lightning network. Finally, Chapter 5 presents a powerful probing technique called Torrent to discover balances of channels in the Lightning network. Unlike the existing techniques, Torrent uses multi-path payments instead of single-path payments. This—together with a novel max flow algorithm designed for the Lightning Network—allows a single probing node to push a large flow of payments through a target channel, making it possible to discover balances of even remote channels. Another major advantage of Torrent is that it can speed up balance discovery through parallel payment transfers and pre-computation of payment paths.