Using Crypto-currencies to Measure Financial Activities and Uncover Potential Identities of Actors Involved
- Author(s): Huang, Yuxing;
- Advisor(s): Snoeren, Alex C;
- Levchenko, Kirill
- et al.
Bitcoin is a digital currency that has recently gathered significant interest. From ecommerce
sites to darkweb marketplaces, merchants accept Bitcoin as a form of payment.
Every day, millions of dollars are transacted across Bitcoin’s payment network. The
value of a single bitcoin has increased from $500 to $3,000 in a one-year period since
A part of the interest may stem from the decentralized design of Bitcoin. A
peer-to-peer network collectively generates new coins and maintains the distributed
transaction ledger, also known as the blockchain. The blockchain records transactions
between public keys, rather than between real-world identities. This detachment from
real-world identities makes it hard to measure financial activities and identify actors on
the network, such as four cases that we study: (i) botnets stealing computational cycles,
(ii) speculatively investing in digital currencies, (iii) delaying the processing of Bitcoin
payments, and (iv) purchasing ads with illegal contents.
Despite this challenge, the decentralized design of Bitcoin and similar digital
currencies offers public information on every transaction and the associated identities.
This dissertation demonstrates that, using the four cases as examples, we can leverage
this public information to analyze financial activities — e.g. measuring the cost and
revenue — and to potentially uncover the identities of the actors involved.
In particular, we can measure the revenue and cost for Cases (i) through (iii). For
(i), we show that botnets made a modest income of $118,000 between 2012 and 2013,
but for some botnets we estimate the cost to victims to be more than twice the botnets’
revenue. For (ii), we develop a new way to estimate the profitability of investing in digital
currency markets. By simulating multiple investment strategies, we show the drastic
variations in profitability and thus the extreme risks associated with digital currency
investment. For (iii), we show that an adversary delayed Bitcoin transaction processing
time from 0.33 to 2.67 hours, at a modest cost of $4,900 per day. Furthermore, we can
uncover the potential identities of the actors involved. For (i), we identify 10 distinct
botnet operations. For (iv), we identify ads paid for by potentially the same criminals.