Performance of Bitcoin Protocol Variants

井上万実(1451011)


Bitcoin is the first widespread distributed virtual currency and the outstanding point of its system is that it allows to make possible to create secure transactions with untrustworthy clients. However there are concerns about the security and performance of Bitcoin system. This work makes three contributions: First, we analyze real-world Bitcoin traffic data spanning over a year and create a queueingtheoretic model from them. We modeled these transactions as two classes: highpriority and low-priority, and we reproduced known scalability problems of the Bitcoin protocol. Second, we perform a Monte Carlo simulation of the Bitcoin protocol and one of its proposed variants called “Heavy-Chain protocol” [SZ15], to seek possible solution for the scalability problems. Our analysis focuses on the two perspectives: on its performance from the viewpoint of the number of transactions, and on the robustness of the protocol against the “51%attack”, which is currently considered as the most critical attack for the Bitcoin. We show that the Heavy-Chain protocol incurs no significant performance penalty and is more robust against the 51%attack, except 25% increase of the block chain size. Finally, we performed a preliminary evaluation for the “delayed checking of blocks”, as a possible mitigation for reducing the block chain size. In conclusion, the thesis shows the Heavy-chain can shorten the block time interval without increasing the risk of the 51%attack. Also, by giving up complete validation in advance of sending a block, we mitigate the side effect caused by Heavy-chain and shortening block time interval.