Author: Elias Rohrer 2018-09-14 13:44:34
Published on: 2018-09-14T13:44:34+00:00
In an email exchange between Elias Rohrer and Kulpreet Singh, they discussed several aspects related to the Lightning Network (LN). Firstly, Elias explained that message complexity could be a significant challenge in the distributed case of LN. Goldberg et al. established that it would take approximately O(n^3) in the asynchronous model. However, he suggested ideas like introducing rounds or batching payments to address this issue. Secondly, Elias mentioned that replacing the current source routing scheme's route selection algorithm with a flow-based approach should be possible, which is interesting to see. He identified that combining such a flow-based approach with the atomic multi-path payment (AMP) scheme proposed in the beginning of this year could be particularly promising. Elias did not look into possible library candidates for a prototype implementation in the Go client yet but mentioned that it seems exciting. Furthermore, Elias suggested having a look at average maximum flows between all or random nodes as a first idea, which could help to understand how much payment capacity can be unlocked by implementing a multi-path approach such as AMP. Kulpreet had some questions about Elias's paper. The first question was about the practicality of the distributed execution as proposed in the Goldberg paper. He wondered how it would work when a waterfall occurs after nodes talk to their neighbors. Elias responded that periodic route calculation phases, where all routing nodes collaborate every x seconds to help find routes for all the payments in that phase, could be an option. The second question was whether Elias has an implementation of the push-relabel algorithm in Go that could be tried out in LN or if he had seen one as part of another library. Elias replied that he had not looked into that yet. In addition, Kulpreet shared his work on measuring node centrality in the LN graph and the dependence on some nodes to route payments and prevent network partitions. He measured the centrality of nodes and the central point dominance as defined in the seminal paper by Lindon C. Freeman, and he measured the number of articulation points in the network as per Robert E. Tarjan. He elaborated on the choice of the two metrics and what they mean for LN. Elias found this work interesting and suggested that treating LN as a flow network could be beneficial. Rene Pickhardt also participated in the email thread, sharing his views about autopilot and some resources related to it.
Updated on: 2023-05-25T02:19:04.659453+00:00