Collusion Attack from Hubs in The Blockchain Offline Channel Network

Abstract

Offline channels can improve the scalability of blockchains by reducing the number of transactions in the blockchain. Offline channels provide Path-Based fund Transfer (PBT) service which allows a pair of peers without a mutual channel to transfer fund between them using paths in the channel network. In PBTs, peers allow a 3rd party to use their channel for fund transfer in exchange for a transfer fee. There are channels in the Bitcoin Lightning network which are designed to collect such PBT transfer fees. An analysis of Bitcoin’s Lightning network revealed the existence of hubs or nodes with very high degree in the channel network. There are only 10 nodes who own more than 50% funds in the Lightning network. These nodes are designed to facilitate PBTs among peers with a low degree (number of channels) in exchange for transfer fees. The emergence of hubs in channel network created the possibility of collusion attack on the channel network where a group of hubs deliberately make few channels non-operational to prevent PBTs involving a selected set of hubs (victims of the collusion attack). In this paper, we model such collusion attack using cooperative game theory and using Banzhaf index we classify the vulnerability of the hubs from the collusion attacks. We propose a design principle of the channel network that can decrease the possibility of collusion attacks.