Reducing CO₂ emissions in a peer-to-peer distributed payment network: Does geography matter in the lightning network?

Danila Valko and Daniel Kudenko
Computer Networks
Bitcoin, Ethereum and other blockchain-related globally distributed peer-to-peer networks have become an important payment infrastructure. Because they are naturally decentralized and distributed, they can consume resources unevenly, raising some sustainability concerns. The Lightning Network (LN) is a growing but understudied second layer network built on blockchain infrastructure and designed to make fast and anonymous global multi-pass payments. At a conservative estimate, this infrastructure produces about 1.4 million tons of carbon dioxide (CO₂) per year. Thus, in this paper, we aim to justify a general approach to consider topology and the geographical distribution of this type of network in the design and investigate ways to prevent excessive CO₂ emissions. While the LN itself shows great promise as a scalable and widely adopted solution, there has been limited research exploring its structure, distribution, and performance from a sustainability perspective. This study contributes to analyzing the LN’s topology, geospatial distribution, and pathfinding algorithms. By examining real-world data snapshots of the LN, we investigate the relationship between payment routes that are produced by native pathfinding algorithms, geographical distribution of the network and the carbon intensity of electricity in the countries involved in the final payment paths. Our analysis highlights the important structural and geospatial characteristics of the LN and reveals a significant correlation between the length of payment paths, geographical distance, carbon intensity of electricity and other features. To tackle sustainability concerns, we propose an original pathfinding heuristic that effectively prevents excessive carbon dioxide emissions in LN infrastructure. Our computational experiments have shown that, under optimal parameters, such a heuristic can prevent the associated CO₂ emissions both directly – by limiting path lengths, number of intercountry and intercontinental hops – and indirectly – by giving more weight to channels covering places with lower carbon intensity of electricity. Technically, the highest result it achieves is as follows: the average path length reduced by 28.7%, the average number of intercontinental hops by 28.7%, the average number of intercountry hops by 21.3%, and the average carbon intensity by 9.4%. This solution also maintains a compromise between the average locktimes and fee ratios. In conclusion, we discuss that geographic distribution is a rather important characteristic of decentralized peer-to-peer payment networks, which is usually underestimated at the network design stage.
April / 2024