Title:Micro Analysis of Natural Forking in Blockchain Based on Large Deviation Theory

Abstract:Natural forking in blockchain refers to a phenomenon that there are a set of blocks at one block height at the same time, implying that various nodes have different perspectives of the main chain. Natural forking might give rise to multiple adverse impacts on blockchain, jeopardizing the performance and security of the system consequently. However, the ongoing literature in analyzing natural forking is mainly from the macro point of view, which is not sufficient to incisively understand this phenomenon. In this paper, we fill this gap through leveraging the large deviation theory to conduct a microscopic study of natural forking, which resorts to investigating the instantaneous difference between block generation and dissemination in blockchain. Our work is derived comprehensively and complementarily via a three-step process, where both the natural forking probability and its decay rate are presented. Through solid theoretical derivation and extensive numerical simulations, we find 1) the probability of the mismatch between block generation and dissemination exceeding a given threshold dwindles exponentially with the increase of natural forking robustness related parameter or the difference between the block dissemination rate and block creation rate; 2) the natural forking robustness related parameter may emphasize a more dominant effect on accelerating the abortion of natural forking in some cases; 3) when the self-correlated block generation rate is depicted as the stationary autoregressive process with a scaling parameter, it is found that setting a lower scaling parameter may speed up the failure of natural forking. These findings are valuable since they offer a fresh theoretical basis to engineer optimal countermeasures for thwarting natural forking and thereby enlivening the blockchain network.