Search

Partitioning attacks have been a known threat since the invention of cryptocurrencies. Attackers could deliberately fork the chain by re-routing network traffic into two or more separate chains and spend money on each piece, effectively spending multiples of their money. Apostolaki et. al. [{Apostolaki and {Laurent Vanbever}, 2017] were among the first to quantify the threats of such attacks on Bitcoin. They suggest a number of ways to mitigate this risk which were combined into a tool named SABRE. Jyothi explored the possibility that a solar superstorm could damage the undersea fiber-optic cables that connect the Internets of different continents, and considered the mostly likely ramifications of the damage. She concluded that such an event would likely cause major connectivity issues across the northern hemisphere and may disconnect much of North America’s internet from the eastern hemisphere for weeks. There is also concern that undersea cables could be deliberately destroyed as acts of terrorism or war or by natural disasters such as earthquakes. In this work, we construct a simulation to properly quantify the effects of a global-scale network partition on the blockchain. We hope to provide the groundwork for preventative measures to be taken to minimize the harm that such partitions might cause in the future. We do this by modifying SimBlock [{Yusuke, 2019], a blockchain simulator created to study the effect of different network topologies, to allow initiating and recovering from partitions and also add metrics to capture their effects. To quantify the severity of partitions we use a number of metrics, including the rate of agreement improvement after a new block has been minted and the average rate of block propagation across regions. We also examine the number of forks in the blockchain that result from partitions and identify the break-points at which forks begin to appear. Finally, we quantify the duration that partitions of various sizes can persist before

Blockchain plays a significant role in cryptocurrencies and growing applications like smart contracts. However, prior blockchain algorithms did not consider large-scale network partitioning a considerable concern while relying heavily on a reliable global network. Previous works have shown a possibility of a massive disruption on the Internet. The author in [Jyothi, 2021] discusses the case of Internet disorder due to solar superstorms, which can disconnect different geographical regions from each other for months. Partitioning attacks are also notable concerns that should be considered, in which their goal is to cut connections between a set of nodes and the rest of the network. In the case of network partitioning, the main chain will fork into branches, and miners in different disconnected regions will create multiple blocks in parallel. The longest chain rule in current blockchain systems accepts only one of the branches after the network is recovered, and because of that, all blocks in other branches will be pruned. Losing a considerable number of mined blocks is not tolerable and significantly impacts the reliability of the ledger and miners' benefit. In this work, we aim to improve blockchain resilience by designing a partition-tolerance blockchain system that: (1) split into branches when network partition happens. (2) merge existing branches into one when the network goes back to normal. (3) ensure the safety and integrity of the blockchain. Newly mined blocks will be collectively signed by a group of miners with a BFT protocol similar to ByzCoin[Kogias et al., 2016], where the consensus group is formed by the miners of the previous w blocks. When a network partition happens, only part of the consensus group can be reached; thus the number of signers w_b of the new block will be less than w. If a block with w_b signers is published, every node in the partition learns that they are now in a branch with around w_b/w of the total hashing power, and it can be id