Topological analysis and simulations on centrality in the Lightning Network

Bitcoin is the first decentralized cryptocurrency ever created, and due to the transaction limit of the blockchain it is not scalable. To overcome the scalability problem, payment channel networks have been developed. Payment channel networks reduce the number of transactions loaded on the blockchain, allowing unbounded off-chain payments in almost instant time. Lightning Network is the most prominent payment channel network built for Bitcoin. With its time-based smart contract called HTLC, it allows off-chain payments across channels without trusting other participants, with small fees to route payments. Payments are routed with a few hops from peers across the network. Topological studies show that since channel creation takes funds from the blockchain nodes tend to create few channels by connecting to central nodes in the network rather than creating ad hoc ones. A very important research field is understanding if there are nodes that may have an important or central position in the network or if there are hubs. These nodes can compromise the decentralization of the network and if they are not cooperative, they can increase the rate of payments failure. This work concerns the research of these nodes on the Lightning Network. There are different centrality measures in graph theory that define different concepts of centrality of a node based on several criteria. Main centrality measures will be presented and Lightning Network’s nodes that are central to these measures will be identified. For each centrality measure, central nodes will be removed, and the topology of the network will be studied. A random node removal will also be performed to test the network’s resistance and behavior to random attacks or offline nodes. For each centrality measure, simulations were conducted on the Lightning Network removing the central nodes. In addition of the remotion of the total list of central nodes three intermediate simulations will also be carried out for each quarter of the list. Subsequently, simulations using CLoTH will be performed. CLoTH is a Lightning Network simulator. It simulates payments on the Lighthning Network and produces performance measures, such as probability of payment success and average payment time. The topological results showed that removing a few central nodes is sufficient to detach many nodes in the network. Most of these nodes are isolated, confirming that nodes have few connections to central nodes. Payments need on average more hops to reach the destination. CLoTH simulations confirmed the average number of hops, adding details regarding increment in the average time for payments and more payment failures. The results showed that Lightning Network has some nodes that are important to the network. The removal of these central nodes creates inconveniences in the success of the payments not comparable to a random removal of nodes. However, despite their removal, the network is still fairly well connected.