Source detection in the bitcoin network: a multi-reporting approach

Motivated by analyzing anonymity properties of Bitcoin network and identification of the origin of illegal transactions, we study the problem of detecting the source node of a transaction message in the Bitcoin network, based on the present spreading model-diffusion. We start by adopting a listening model to get the information of which part of nodes have received the message, say an observation, which is an important premise of solving source detection problem. We propose an estimator for regular trees based on independent multi-reporting observations, and theoretically give a lower bound of the correct detection probability when the observation moment tends to infinity. We show that the more independent reporting observations we have, the higher the probability of detection is, and it further approaches one. The effectiveness of our source estimator is also established in several simulations.