In this paper, we focus on channel balance probing. We propose a new model of the LN that accounts for parallel and unidirectional channels, which has not been done in prior work. We describe a probing algorithm that accurately updates the attacker's balance estimates without the need to directly connect to victims. We introduce an uncertainty-based metric to measure the attacker's information gain.
We implement the first probing-focused LN simulator and suggest several countermeasures against general probing (implemented considering parallel channels). We evaluate these techniques using the simulator, as well as experiments on the real network.
According to our simulations, an attacker can infer up to $80\%$~information regarding channel balances spending $\approx 20$~seconds per channel. The suggested countermeasures limit the attacker's gain at $30\%$, while also increasing the attack time by 2-4x.
In addition, we describe sophisticated attack techniques that combine fee-probing and channel jamming to get precise access to individual channel balances inside a hop, and test them against the real network.
Finally, we discuss payment flows and their concealment.
Category / Keywords: applications / privacy, Bitcoin, Lightning Network, payment channels, payment channel network Date: received 22 Mar 2021 Contact author: sergey s tikhomirov at gmail com Available format(s): PDF | BibTeX Citation Version: 20210327:071104 (All versions of this report) Short URL: ia.cr/2021/384