The evolving topology of the Lightning Network: Centralization, efficiency, robustness, synchronization, and anonymity

Abstract

The Lightning Network (LN) was released on Bitcoin's mainnet in January 2018 as a solution to favor scalability. This work analyses the evolution of the LN during its first year of existence in order to assess its impact over some of the core fundamentals of Bitcoin, such as: node centralization, resilience against attacks and disruptions, anonymity of users, autonomous coordination of its members. Using a network theory approach, we find that the LN represents a centralized configuration with few highly active nodes playing as hubs in that system. We show that the removal of these central nodes is likely to generate a remarkable drop in the LN's efficiency, while the network appears robust to random disruptions. In addition, we observe that improvements in efficiency during the sample period are primarily due to the increase in the capacity installed on the channels, while nodes' synchronization does not emerge as a distinctive feature of the LN. Finally, the analysis of the structure of the network suggests a good preservation of nodes' identity against attackers with prior knowledge about topological characteristics of their targets, but also that LN is probably weak against attackers that are within the system.

A Tale of Two Layers: The Mutual Relationship between Bitcoin and Lightning Network

“…In both snapshots it is possible to notice the presence of a few large nodes surrounded by smaller ones indistinguishable from each other. The presence of a few massively endowed nodes highly connected with the rest of the network, composed by a vast majority of relatively poorly endowed nodes, suggests an overall hub and spoke structure of the system, a feature already highlighted by Martinazzi and Flori (2020). Moreover, in Table 1 we show some topological measures collected for the LN at the beginning and at the end of the sample period.…”

“…Finally, following the approach proposed in Martinazzi and Flori (2020), we decide to use as main representation of the LN's configuration its topological efficiency. It depends on key elements of the structure of the network, such as its density and the distribution of the capacity stored in its channels, hence it is a measure capable to aggregate a great deal of relevant information about the functioning of the network.…”

“…For this reason, we employ the topological efficiency proposed by Marchiori (2001, 2003) to assess its likelihood to disseminate information through its nodes, which is a critical aspect for successfully routing transactions in such a multi-hop framework. In line with Martinazzi and Flori (2020), we provide therefore a network theory analysis of the LN, but in this case we propose an evaluation of the efficiency on a daily basis to better assess the impact of market dynamics. We consider the period from 12 February 2018, when the LN started, to 12 August 2020.…”

Counting Down Thunder

“…Furthermore, recent entries have discussed the possibility of discovering the private channel balances by probing [40,41] and analyzed how much privacy could be retained, if noisy channel balances were to be made public [42]. A number of papers analyzed the graph-theoretic properties of the Lightning Network graph and discussed possible consequences regarding decentralization and routing [39,43,44], as well as graph-based privacy properties [45]. Similarly, Béres et al [46] and Tikhomirov et al [47] empirically analyzed the privacy properties of PCNs with the assistance of model-based traffic simulation.…”

Link and Node Removal in Real Social Networks: A Review

“…In the last few decades, a number of studies investigated the response of real networks to link/node removal (LNR) in what is called "network attack analysis" because it simulates the consequences of an attack on the network [1][2][3][4][5][6][7][8]. These studies found application in very different fields of science such as biology [9][10][11], ecology [12][13][14][15], transport and infrastructure science [16][17][18][19][20][21], informatics [22,23], neurology [24], economics [25,26], and social networks [27][28][29][30]. These studies aimed to (i) assess network robustness, a measure that indicates the capacity of the system to maintain its functions after LNR [6,31], and (ii) identify the LNRs that trigger the greatest amount of damage in the systems, thus revealing the links/nodes that act as key players in network functioning [5,31].…”